Name a constant better.

Halton has less correlation artifacts than Sobol for DoF, but
converges slower.

Cargo.lock

@@ -1,5 +1,7 @@
 # This file is automatically @generated by Cargo.
 # It is not intended for manual editing.
+version = 3
+
 [[package]]
 name = "ansi_term"
 version = "0.11.0"
@@ -9,15 +11,6 @@ dependencies = [
  "winapi",
 ]
 
-[[package]]
-name = "approx"
-version = "0.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3f2a05fd1bd10b2527e20a2cd32d8873d115b8b39fe219ee25f42a8aca6ba278"
-dependencies = [
- "num-traits",
-]
-
 [[package]]
 name = "arrayvec"
 version = "0.5.2"
@@ -133,6 +126,15 @@ dependencies = [
 [[package]]
 name = "color"
 version = "0.1.0"
+dependencies = [
+ "colorbox",
+]
+
+[[package]]
+name = "colorbox"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3d27d55561009760957654f467735e73806f8bc2d081cc4a22e93403ecd156fc"
 
 [[package]]
 name = "compact"
@@ -155,6 +157,10 @@ version = "0.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "24ce9782d4d5c53674646a6a4c1863a21a8fc0cb649b3c94dfc16e45071dea19"
 
+[[package]]
+name = "data_tree"
+version = "0.1.0"
+
 [[package]]
 name = "fastapprox"
 version = "0.3.0"
@@ -184,12 +190,6 @@ dependencies = [
  "wasi",
 ]
 
-[[package]]
-name = "glam"
-version = "0.15.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "411e0584defa447c328f25c756ba3d0685727ecc126b46c3c1176001141cd4b6"
-
 [[package]]
 name = "half"
 version = "1.7.1"
@@ -240,14 +240,6 @@ version = "0.2.94"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "18794a8ad5b29321f790b55d93dfba91e125cb1a9edbd4f8e3150acc771c1a5e"
 
-[[package]]
-name = "math3d"
-version = "0.1.0"
-dependencies = [
- "approx",
- "glam",
-]
-
 [[package]]
 name = "memchr"
 version = "2.4.0"
@@ -350,20 +342,22 @@ dependencies = [
  "bvh_order",
  "clap",
  "color",
+ "colorbox",
  "compact",
  "copy_in_place",
  "crossbeam",
+ "data_tree",
  "fastapprox",
- "glam",
  "half",
  "halton",
  "kioku",
  "lazy_static",
- "math3d",
  "nom",
  "num_cpus",
  "openexr",
  "png_encode_mini",
+ "rmath",
+ "rrand",
  "rustc-serialize",
  "scoped_threadpool",
  "sobol_burley",
@@ -569,6 +563,22 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "rmath"
+version = "0.1.0"
+dependencies = [
+ "bencher",
+ "rand 0.6.5",
+]
+
+[[package]]
+name = "rrand"
+version = "0.1.0"
+dependencies = [
+ "bencher",
+ "rand 0.6.5",
+]
+
 [[package]]
 name = "rustc-serialize"
 version = "0.3.24"
@@ -607,15 +617,15 @@ checksum = "1d51f5df5af43ab3f1360b429fa5e0152ac5ce8c0bd6485cae490332e96846a8"
 
 [[package]]
 name = "sobol_burley"
-version = "0.3.0"
+version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "26e3528b09b1f1b1e152342a4462d1e80d568dc5623a0772252a6e584a53d550"
+checksum = "3441b32bbb896e372f1b8e7eb51a983713aef99599c32c0eb69183aa490cb6a0"
 
 [[package]]
 name = "spectral_upsampling"
 version = "0.1.0"
 dependencies = [
- "glam",
+ "rmath",
 ]
 
 [[package]]

Cargo.toml

@@ -3,8 +3,10 @@ members = [
     "sub_crates/bvh_order",
     "sub_crates/color",
     "sub_crates/compact",
+    "sub_crates/data_tree",
     "sub_crates/halton",
-    "sub_crates/math3d",
+    "sub_crates/rmath",
+    "sub_crates/rrand",
     "sub_crates/spectral_upsampling",
 ]
 
@@ -30,13 +32,14 @@ nom = "5"
 num_cpus = "1.8"
 openexr = "0.7"
 kioku = "0.3"
-sobol_burley = "0.3"
+sobol_burley = "0.4"
 png_encode_mini = "0.1.2"
 rustc-serialize = "0.3"
 scoped_threadpool = "0.1"
 time = "0.1"
-glam = "0.15"
 fastapprox = "0.3"
+colorbox = "0.3"
+
 
 # Local crate dependencies
 [dependencies.bvh_order]
@@ -47,12 +50,18 @@ path = "sub_crates/color"
 
 [dependencies.compact]
 path = "sub_crates/compact"
-[dependencies.halton]
 
+[dependencies.data_tree]
+path = "sub_crates/data_tree"
+
+[dependencies.halton]
 path = "sub_crates/halton"
 
-[dependencies.math3d]
-path = "sub_crates/math3d"
+[dependencies.rmath]
+path = "sub_crates/rmath"
+
+[dependencies.rrand]
+path = "sub_crates/rrand"
 
 [dependencies.spectral_upsampling]
 path = "sub_crates/spectral_upsampling"

README.md

@@ -13,9 +13,7 @@ efficiently handle very large data sets, complex shading, motion blur, color
 management, etc. presents a much richer and more challenging problem space to
 explore than just writing a basic path tracer.
 
-
 ## Building
-
 Psychopath is written in [Rust](https://www.rust-lang.org), and is pretty
 straightforward to build except for its OpenEXR dependency.
 
@@ -36,7 +34,6 @@ documented in the [OpenEXR-rs readme](https://github.com/cessen/openexr-rs/blob/
 Once those environment variables are set, then you should be able to build using
 the same simple cargo command above.
 
-
 # PsychoBlend
 
 Included in the repository is an add-on for [Blender](http://www.blender.org)
@@ -53,6 +50,15 @@ doesn't support them yet.
 - Exports dupligroups with full hierarchical instancing
 - Limited auto-detection of instanced meshes
 
+# Contributing
+
+I'm not looking for contributions right now, and I'm likely to reject pull
+requests.  This is currently a solo project and I like it that way.
+
+However, if you're looking for projects _related_ to Psychopath to contribute to,
+[OpenEXR-rs](https://github.com/cessen/openexr-rs) is definitely a
+collaborative project that I would love more help with!  And I fully expect more
+such projects to come out of Psychopath in the future.
 
 # License
 
@@ -63,13 +69,3 @@ See LICENSE.md for details.  But the gist is:
 * Most crates under the `sub_crates` directory are dual-licensed under MIT and Apache 2.0 (but with some exceptions--see each crate for its respective licenses).
 
 The intent of this scheme is to keep Psychopath itself copyleft, while allowing smaller reusable components to be licensed more liberally.
-
-
-# Contributing
-
-This is a personal, experimental, for-fun project, and I am specifically
-not looking for contributions of any kind.  All PRs will be rejected
-without review.
-
-However, feel free to fork this into an entirely new project, or examine
-the code for ideas for a project of your own.

build_native.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+RUSTFLAGS="-C target-cpu=native" cargo build --release

example_scenes/cube.psy

@@ -1,60 +1,73 @@
-Scene $Scene_fr1 {
-    Output {
-        Path ["test_renders/cube.png"]
-    }
-    RenderSettings {
-        Resolution [960 540]
-        SamplesPerPixel [16]
-        Seed [1]
-    }
-    Camera {
-        Fov [49.134342]
-        FocalDistance [9.559999]
-        ApertureRadius [0.250000]
-        Transform [0.685881 0.727634 -0.010817 0.000000 -0.317370 0.312469 0.895343 0.000000 -0.654862 0.610666 -0.445245 0.000000 7.481132 -6.507640 5.343665 1.000000]
-    }
-    World {
-        BackgroundShader {
-            Type [Color]
-            Color [rec709, 0.050876 0.050876 0.050876]
-        }
-    }
-    Shaders {
-        SurfaceShader $Material {
-            Type [Lambert]
-            Color [rec709, 0.800000 0.800000 0.800000]
-        }
-    }
-    Objects {
-        MeshSurface $__Plane_ {
-            SurfaceShaderBind [$Material]
-            Vertices [-1.000000 -1.000000 0.000000 1.000000 -1.000000 0.000000 -1.000000 1.000000 0.000000 1.000000 1.000000 0.000000]
-            FaceVertCounts [4 ]
-            FaceVertIndices [0 1 3 2 ]
-        }
-        MeshSurface $__Cube_ {
-            SurfaceShaderBind [$Material]
-            Vertices [1.000000 1.000000 -1.000000 1.000000 -1.000000 -1.000000 -1.000000 -1.000000 -1.000000 -1.000000 1.000000 -1.000000 1.000000 0.999999 1.000000 0.999999 -1.000001 1.000000 -1.000000 -1.000000 1.000000 -1.000000 1.000000 1.000000 ]
-            FaceVertCounts [4 4 4 4 4 4 ]
-            FaceVertIndices [0 1 2 3 4 7 6 5 0 4 5 1 1 5 6 2 2 6 7 3 4 0 3 7 ]
-        }
-        SphereLight $__Lamp {
-            Color [rec709, 50.000000 50.000000 50.000000]
-            Radius [0.100000]
-        }
-    }
-    Assembly {
-        Instance {
-            Data [$__Plane_]
-            Transform [0.078868 -0.000000 0.000000 -0.000000 -0.000000 0.078868 -0.000000 0.000000 0.000000 -0.000000 0.078868 -0.000000 -0.000000 0.000000 -0.000000 1.000000]
-        }
-        Instance {
-            Data [$__Cube_]
-            Transform [1.000000 -0.000000 0.000000 -0.000000 -0.000000 1.000000 -0.000000 0.000000 0.000000 -0.000000 1.000000 -0.000000 -0.000000 0.000000 -1.000000 1.000000]
-        }
-        Instance {
-            Data [$__Lamp]
-            Transform [0.019856 -0.060763 0.000000 -0.000000 0.015191 0.079422 -0.000000 0.000000 0.000000 -0.000000 1.000000 -0.000000 -0.026851 -0.125233 -4.432303 1.000000]
-        }
+ClearScene {}
+
+Output {
+    Path ["test_renders/cube.png"]
+}
+
+RenderSettings {
+    Resolution [960 540]
+    SamplesPerPixel [16]
+    Seed [1]
+}
+
+Camera {
+    Fov [49.134342]
+    FocalDistance [9.559999]
+    ApertureRadius [0.250000]
+    Transform [0.685881 0.727634 -0.010817 -0.317370 0.312469 0.895343 -0.654862 0.610666 -0.445245 7.481132 -6.507640 5.343665]
+}
+
+World {
+    BackgroundShader {
+        Type [Color]
+        Color [rec709, 0.050876 0.050876 0.050876]
+    }
+}
+
+AddShader {
+    SurfaceShader {
+        Name [Material]
+        Type [Lambert]
+        Color [rec709, 0.800000 0.800000 0.800000]
+    }
+}
+
+AddObject {
+    MeshSurface {
+        Name [__Plane_]
+        SurfaceShaderBind [Material]
+        Vertices [-1.000000 -1.000000 0.000000 1.000000 -1.000000 0.000000 -1.000000 1.000000 0.000000 1.000000 1.000000 0.000000]
+        FaceVertCounts [4 ]
+        FaceVertIndices [0 1 3 2 ]
+    }
+    MeshSurface {
+        Name [__Cube_]
+        SurfaceShaderBind [Material]
+        Vertices [1.000000 1.000000 -1.000000 1.000000 -1.000000 -1.000000 -1.000000 -1.000000 -1.000000 -1.000000 1.000000 -1.000000 1.000000 0.999999 1.000000 0.999999 -1.000001 1.000000 -1.000000 -1.000000 1.000000 -1.000000 1.000000 1.000000 ]
+        FaceVertCounts [4 4 4 4 4 4 ]
+        FaceVertIndices [0 1 2 3 4 7 6 5 0 4 5 1 1 5 6 2 2 6 7 3 4 0 3 7 ]
+    }
+    SphereLight {
+        Name [__Lamp]
+        Color [rec709, 50.000000 50.000000 50.000000]
+        Radius [0.100000]
+    }
+}
+
+AddInstance {
+    Instance {
+        Name [__Plane_]
+        Data [__Plane_]
+        Transform [0.078868 0.000000 0.000000 0.000000 0.078868 0.000000 0.000000 0.000000 0.078868 0.000000 0.000000 0.000000]
+    }
+    Instance {
+        Name [__Cube_]
+        Data [__Cube_]
+        Transform [1.000000 0.000000 0.000000 0.000000 1.000000 0.000000 0.000000 0.000000 1.000000 0.000000 0.000000 -1.000000]
+    }
+    Instance {
+        Name [__Lamp]
+        Data [__Lamp]
+        Transform [0.019856 -0.060763 0.000000 0.015191 0.079422 0.000000 0.000000 0.000000 1.000000 -0.026851 -0.125233 -4.432303]
     }
 }

psychoblend/__init__.py

@@ -1,22 +1,17 @@
 bl_info = {
     "name": "PsychoBlend",
-    "version": (0, 1),
+    "version": (0, 1, 0),
     "author": "Nathan Vegdahl",
-    "blender": (2, 70, 0),
+    "blender": (3, 1, 0),
     "description": "Psychopath renderer integration",
     "location": "",
-    "wiki_url": "https://github.com/cessen/psychopath/wiki",
-    "tracker_url": "https://github.com/cessen/psychopath/issues",
-    "category": "Render"}
+    # "wiki_url": "https://github.com/cessen/psychopath/wiki",
+    # "tracker_url": "https://github.com/cessen/psychopath/issues",
+    "category": "Render"
+}
 
 
-if "bpy" in locals():
-    import imp
-    imp.reload(ui)
-    imp.reload(psy_export)
-    imp.reload(render)
-else:
-    from . import ui, psy_export, render
+from . import ui, psy_export, render
 
 import bpy
 from bpy.types import (AddonPreferences,
@@ -33,53 +28,46 @@ from bpy.props import (StringProperty,
                        )
 
 
-# Custom Scene settings
+# Custom Scene settings.
 class RenderPsychopathSettingsScene(PropertyGroup):
-    spp = IntProperty(
+    spp: IntProperty(
         name="Samples Per Pixel", description="Total number of samples to take per pixel",
         min=1, max=65536, default=16
         )
 
-    max_samples_per_bucket = IntProperty(
-        name="Max Samples Per Bucket", description="How many samples to simultaneously calculate per thread; indirectly determines bucket size",
-        min=1, max=2**28, soft_max=2**16, default=4096
+    bucket_size: IntProperty(
+        name="Bucket Size", description="The height and width of each render bucket in pixels.",
+        min=1, max=4096, soft_max=256, default=32
         )
 
-    dicing_rate = FloatProperty(
+    dicing_rate: FloatProperty(
         name="Dicing Rate", description="The target microgeometry width in pixels",
         min=0.0001, max=100.0, soft_min=0.125, soft_max=1.0, default=0.25
         )
 
-    motion_blur_segments = IntProperty(
+    motion_blur_segments: IntProperty(
         name="Motion Segments", description="The number of segments to use in motion blur.  Zero means no motion blur.  Will be rounded down to the nearest power of two.",
         min=0, max=256, default=0
         )
 
-    shutter_start = FloatProperty(
+    shutter_start: FloatProperty(
         name="Shutter Open", description="The time during the frame that the shutter opens, for motion blur",
         min=-1.0, max=1.0, soft_min=0.0, soft_max=1.0, default=0.0
         )
 
-    shutter_end = FloatProperty(
+    shutter_end: FloatProperty(
         name="Shutter Close", description="The time during the frame that the shutter closes, for motion blur",
         min=-1.0, max=1.0, soft_min=0.0, soft_max=1.0, default=0.5
         )
 
-    export_path = StringProperty(
+    export_path: StringProperty(
         name="Export Path", description="The path to where the .psy files should be exported when rendering.  If left blank, /tmp or the equivalent is used.",
         subtype='FILE_PATH'
         )
 
-# Custom Camera properties
-class PsychopathCamera(bpy.types.PropertyGroup):
-    aperture_radius = FloatProperty(
-        name="Aperture Radius", description="Size of the camera's aperture, for DoF",
-        min=0.0, max=10000.0, soft_min=0.0, soft_max=2.0, default=0.0
-        )
-
-# Psychopath material
+# Psychopath material.
 class PsychopathLight(bpy.types.PropertyGroup):
-    color_type = EnumProperty(
+    color_type: EnumProperty(
         name="Color Type", description="",
         items=[
             ('Rec709', 'Rec709', ""),
@@ -89,27 +77,27 @@ class PsychopathLight(bpy.types.PropertyGroup):
         default="Rec709"
         )
 
-    color_blackbody_temp = FloatProperty(
+    color_blackbody_temp: FloatProperty(
         name="Temperature", description="Blackbody temperature in kelvin",
         min=0.0, soft_min=800.0, soft_max=6500.0, default=1200.0
         )
 
-# Custom Mesh properties
+# Custom Mesh properties.
 class PsychopathMesh(bpy.types.PropertyGroup):
-    is_subdivision_surface = BoolProperty(
+    is_subdivision_surface: BoolProperty(
         name="Is Subdivision Surface", description="Whether this is a sibdivision surface or just a normal mesh",
         default=False
         )
 
 # Psychopath material
 class PsychopathMaterial(bpy.types.PropertyGroup):
-    surface_shader_type = EnumProperty(
+    surface_shader_type: EnumProperty(
         name="Surface Shader Type", description="",
         items=[('Emit', 'Emit', ""), ('Lambert', 'Lambert', ""), ('GGX', 'GGX', "")],
         default="Lambert"
         )
 
-    color_type = EnumProperty(
+    color_type: EnumProperty(
         name="Color Type", description="",
         items=[
             ('Rec709', 'Rec709', ""),
@@ -119,39 +107,47 @@ class PsychopathMaterial(bpy.types.PropertyGroup):
         default="Rec709"
         )
 
-    color = FloatVectorProperty(
+    color: FloatVectorProperty(
         name="Color", description="",
         subtype='COLOR',
         min=0.0, soft_min=0.0, soft_max = 1.0,
         default=[0.8,0.8,0.8]
         )
 
-    color_blackbody_temp = FloatProperty(
+    color_blackbody_temp: FloatProperty(
         name="Temperature", description="Blackbody temperature in kelvin",
         min=0.0, soft_min=800.0, soft_max=6500.0, default=1200.0
         )
 
-    roughness = FloatProperty(
+    roughness: FloatProperty(
         name="Roughness", description="",
         min=-1.0, max=1.0, soft_min=0.0, soft_max=1.0, default=0.1
         )
 
-    tail_shape = FloatProperty(
+    tail_shape: FloatProperty(
         name="Tail Shape", description="",
         min=0.0, max=8.0, soft_min=1.0, soft_max=3.0, default=2.0
         )
 
-    fresnel = FloatProperty(
+    fresnel: FloatProperty(
         name="Fresnel", description="",
         min=0.0, max=1.0, soft_min=0.0, soft_max=1.0, default=0.9
         )
 
+# Psychopath world.
+class PsychopathWorld(bpy.types.PropertyGroup):
+    background_color: FloatVectorProperty(
+        name="Background Color", description="",
+        subtype='COLOR',
+        min=0.0, soft_min=0.0, soft_max = 1.0,
+        default=[0.8,0.8,0.8]
+        )
 
-# Addon Preferences
+# Addon Preferences.
 class PsychopathPreferences(AddonPreferences):
     bl_idname = __name__
 
-    filepath_psychopath = StringProperty(
+    filepath_psychopath: StringProperty(
                 name="Psychopath Location",
                 description="Path to renderer executable",
                 subtype='DIR_PATH',
@@ -166,15 +162,15 @@ class PsychopathPreferences(AddonPreferences):
 def register():
     bpy.utils.register_class(PsychopathPreferences)
     bpy.utils.register_class(RenderPsychopathSettingsScene)
-    bpy.utils.register_class(PsychopathCamera)
     bpy.utils.register_class(PsychopathLight)
     bpy.utils.register_class(PsychopathMesh)
     bpy.utils.register_class(PsychopathMaterial)
+    bpy.utils.register_class(PsychopathWorld)
     bpy.types.Scene.psychopath = PointerProperty(type=RenderPsychopathSettingsScene)
-    bpy.types.Camera.psychopath = PointerProperty(type=PsychopathCamera)
-    bpy.types.Lamp.psychopath = PointerProperty(type=PsychopathLight)
+    bpy.types.Light.psychopath = PointerProperty(type=PsychopathLight)
     bpy.types.Mesh.psychopath = PointerProperty(type=PsychopathMesh)
     bpy.types.Material.psychopath = PointerProperty(type=PsychopathMaterial)
+    bpy.types.World.psychopath = PointerProperty(type=PsychopathWorld)
     render.register()
     ui.register()
 
@@ -182,14 +178,14 @@ def register():
 def unregister():
     bpy.utils.unregister_class(PsychopathPreferences)
     bpy.utils.unregister_class(RenderPsychopathSettingsScene)
-    bpy.utils.unregister_class(PsychopathCamera)
     bpy.utils.unregister_class(PsychopathLight)
     bpy.utils.unregister_class(PsychopathMesh)
     bpy.utils.unregister_class(PsychopathMaterial)
+    bpy.utils.unregister_class(PsychopathWorld)
     del bpy.types.Scene.psychopath
-    del bpy.types.Camera.psychopath
-    del bpy.types.Lamp.psychopath
+    del bpy.types.Light.psychopath
     del bpy.types.Mesh.psychopath
     del bpy.types.Material.psychopath
+    del bpy.types.World.psychopath
     render.unregister()
     ui.unregister()

psychoblend/assembly.py

@@ -1,398 +0,0 @@
-import bpy
-
-from .util import escape_name, mat2str, needs_def_mb, needs_xform_mb, ExportCancelled
-
-class Assembly:
-    def __init__(self, render_engine, objects, visible_layers, group_prefix="", translation_offset=(0,0,0)):
-        self.name = group_prefix
-        self.translation_offset = translation_offset
-        self.render_engine = render_engine
-        
-        self.materials = []
-        self.objects = []
-        self.instances = []
-
-        self.material_names = set()
-        self.mesh_names = set()
-        self.assembly_names = set()
-
-        # Collect all the objects, materials, instances, etc.
-        for ob in objects:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-
-            # Check if the object is visible for rendering
-            vis_layer = False
-            for i in range(len(ob.layers)):
-                vis_layer = vis_layer or (ob.layers[i] and visible_layers[i])
-            if ob.hide_render or not vis_layer:
-                continue
-
-            # Store object data
-            name = None
-
-            if ob.type == 'EMPTY':
-                if ob.dupli_type == 'GROUP':
-                    name = group_prefix + "__" + escape_name(ob.dupli_group.name)
-                    if name not in self.assembly_names:
-                        self.assembly_names.add(name)
-                        self.objects += [Assembly(self.render_engine, ob.dupli_group.objects, ob.dupli_group.layers, name, ob.dupli_group.dupli_offset*-1)]
-            elif ob.type == 'MESH':
-                name = self.get_mesh(ob, group_prefix)
-            elif ob.type == 'LAMP' and ob.data.type == 'POINT':
-                name = self.get_sphere_lamp(ob, group_prefix)
-            elif ob.type == 'LAMP' and ob.data.type == 'AREA':
-                name = self.get_rect_lamp(ob, group_prefix)
-            
-            # Store instance
-            if name != None:
-                self.instances += [Instance(render_engine, ob, name)]
-
-    def export(self, render_engine, w):
-        if self.name == "":
-            w.write("Assembly {\n")
-        else:
-            w.write("Assembly $%s {\n" % self.name)
-        w.indent()
-
-        for mat in self.materials:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            mat.export(render_engine, w)
-
-        for ob in self.objects:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            ob.export(render_engine, w)
-
-        for inst in self.instances:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            inst.export(render_engine, w)
-
-        w.unindent()
-        w.write("}\n")
-    
-    #----------------
-
-    def take_sample(self, render_engine, scene, time):
-        for mat in self.materials:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            mat.take_sample(render_engine, scene, time)
-
-        for ob in self.objects:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            ob.take_sample(render_engine, scene, time)
-
-        for inst in self.instances:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            inst.take_sample(render_engine, time, self.translation_offset)
-    
-    def cleanup(self):
-        for mat in self.materials:
-            mat.cleanup()
-        for ob in self.objects:
-            ob.cleanup()
-
-    def get_mesh(self, ob, group_prefix):
-        # Figure out if we need to export or not and figure out what name to
-        # export with.
-        has_modifiers = len(ob.modifiers) > 0
-        deform_mb = needs_def_mb(ob)
-        if has_modifiers or deform_mb:
-            mesh_name = group_prefix + escape_name("__" + ob.name + "__" + ob.data.name + "_")
-        else:
-            mesh_name = group_prefix + escape_name("__" + ob.data.name + "_")
-        has_faces = len(ob.data.polygons) > 0
-        should_export_mesh = has_faces and (mesh_name not in self.mesh_names)
-        
-        # Get mesh
-        if should_export_mesh:
-            self.mesh_names.add(mesh_name)
-            self.objects += [Mesh(self.render_engine, ob, mesh_name)]
-
-            # Get materials
-            for ms in ob.material_slots:
-                if ms != None:
-                    if ms.material.name not in self.material_names:
-                        self.material_names.add(ms.material.name)
-                        self.materials += [Material(self.render_engine, ms.material)]
-
-            return mesh_name
-        else:
-            return None
-
-
-    def get_sphere_lamp(self, ob, group_prefix):
-        name = group_prefix + "__" + escape_name(ob.name)
-        self.objects += [SphereLamp(self.render_engine, ob, name)]
-        return name
-
-    def get_rect_lamp(self, ob, group_prefix):
-        name = group_prefix + "__" + escape_name(ob.name)
-        self.objects += [RectLamp(self.render_engine, ob, name)]
-        return name
-
-
-#=========================================================================
-
-
-class Mesh:
-    """ Holds data for a mesh to be exported.
-    """
-    def __init__(self, render_engine, ob, name):
-        self.ob = ob
-        self.name = name
-        self.needs_mb = needs_def_mb(self.ob)
-        self.time_meshes = []
-
-    def take_sample(self, render_engine, scene, time):
-        if len(self.time_meshes) == 0 or self.needs_mb:
-            render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.ob.name, time))
-            self.time_meshes += [self.ob.to_mesh(scene, True, 'RENDER')]
-    
-    def cleanup(self):
-        for mesh in self.time_meshes:
-            bpy.data.meshes.remove(mesh)
-
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
-
-        if self.ob.data.psychopath.is_subdivision_surface == False:
-            # Exporting normal mesh
-            w.write("MeshSurface $%s {\n" % self.name)
-            w.indent()
-        else:
-            # Exporting subdivision surface cage
-            w.write("SubdivisionSurface $%s {\n" % self.name)
-            w.indent()
-
-        # Write vertices and (if it's smooth shaded) normals
-        for ti in range(len(self.time_meshes)):
-            w.write("Vertices [")
-            w.write(" ".join([("%f" % i) for vert in self.time_meshes[ti].vertices for i in vert.co]), False)
-            w.write("]\n", False)
-            if self.time_meshes[0].polygons[0].use_smooth and self.ob.data.psychopath.is_subdivision_surface == False:
-                w.write("Normals [")
-                w.write(" ".join([("%f" % i) for vert in self.time_meshes[ti].vertices for i in vert.normal]), False)
-                w.write("]\n", False)
-
-        # Write face vertex counts
-        w.write("FaceVertCounts [")
-        w.write(" ".join([("%d" % len(p.vertices)) for p in self.time_meshes[0].polygons]), False)
-        w.write("]\n", False)
-
-        # Write face vertex indices
-        w.write("FaceVertIndices [")
-        w.write(" ".join([("%d"%v) for p in self.time_meshes[0].polygons for v in p.vertices]), False)
-        w.write("]\n", False)
-
-        # MeshSurface/SubdivisionSurface section end
-        w.unindent()
-        w.write("}\n")
-
-
-class SphereLamp:
-    """ Holds data for a sphere light to be exported.
-    """
-    def __init__(self, render_engine, ob, name):
-        self.ob = ob
-        self.name = name
-        self.time_col = []
-        self.time_rad = []
-
-    def take_sample(self, render_engine, scene, time):
-        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.ob.name, time))
-
-        if self.ob.data.psychopath.color_type == 'Rec709':
-            self.time_col += [('Rec709', self.ob.data.color * self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'Blackbody':
-            self.time_col += [('Blackbody', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'ColorTemperature':
-            self.time_col += [('ColorTemperature', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-
-        self.time_rad += [self.ob.data.shadow_soft_size]
-
-    def cleanup(self):
-        pass
-    
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
-
-        w.write("SphereLight $%s {\n" % self.name)
-        w.indent()
-        for col in self.time_col:
-            if col[0] == 'Rec709':
-                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
-            elif col[0] == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
-            elif col[0] == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
-        for rad in self.time_rad:
-            w.write("Radius [%f]\n" % rad)
-
-        w.unindent()
-        w.write("}\n")
-
-
-class RectLamp:
-    """ Holds data for a rectangular light to be exported.
-    """
-    def __init__(self, render_engine, ob, name):
-        self.ob = ob
-        self.name = name
-        self.time_col = []
-        self.time_dim = []
-
-    def take_sample(self, render_engine, scene, time):
-        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.ob.name, time))
-
-        if self.ob.data.psychopath.color_type == 'Rec709':
-            self.time_col += [('Rec709', self.ob.data.color * self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'Blackbody':
-            self.time_col += [('Blackbody', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'ColorTemperature':
-            self.time_col += [('ColorTemperature', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-
-        if self.ob.data.shape == 'RECTANGLE':
-            self.time_dim += [(self.ob.data.size, self.ob.data.size_y)]
-        else:
-            self.time_dim += [(self.ob.data.size, self.ob.data.size)]
-    
-    def cleanup(self):
-        pass
-    
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
-
-        w.write("RectangleLight $%s {\n" % self.name)
-        w.indent()
-        for col in self.time_col:
-            if col[0] == 'Rec709':
-                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
-            elif col[0] == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
-            elif col[0] == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
-        for dim in self.time_dim:
-            w.write("Dimensions [%f %f]\n" % dim)
-
-        w.unindent()
-        w.write("}\n")
-
-
-class Instance:
-    def __init__(self, render_engine, ob, data_name):
-        self.ob = ob
-        self.data_name = data_name
-        self.needs_mb = needs_xform_mb(self.ob)
-        self.time_xforms = []
-
-    def take_sample(self, render_engine, time, translation_offset):
-        if len(self.time_xforms) == 0 or self.needs_mb:
-            render_engine.update_stats("", "Psychopath: Collecting '{}' xforms at time {}".format(self.ob.name, time))
-            mat = self.ob.matrix_world.copy()
-            mat[0][3] += translation_offset[0]
-            mat[1][3] += translation_offset[1]
-            mat[2][3] += translation_offset[2]
-            self.time_xforms += [mat]
-
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
-
-        w.write("Instance {\n")
-        w.indent()
-        w.write("Data [$%s]\n" % self.data_name)
-        for mat in self.time_xforms:
-            w.write("Transform [%s]\n" % mat2str(mat.inverted()))
-        for ms in self.ob.material_slots:
-            if ms != None:
-                w.write("SurfaceShaderBind [$%s]\n" % escape_name(ms.material.name))
-                break
-        w.unindent()
-        w.write("}\n")
-
-
-class Material:
-    def __init__(self, render_engine, material):
-        self.mat = material
-
-    def take_sample(self, render_engine, time, translation_offset):
-        # TODO: motion blur of material settings
-        pass
-
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.mat.name)
-
-        w.write("SurfaceShader $%s {\n" % escape_name(self.mat.name))
-        w.indent()
-        if self.mat.psychopath.surface_shader_type == 'Emit':
-            w.write("Type [Emit]\n")
-            if self.mat.psychopath.color_type == 'Rec709':
-                col = self.mat.psychopath.color
-                w.write("Color [rec709, %f %f %f]\n" % (
-                    col[0], col[1], col[2],
-                ))
-            elif self.mat.psychopath.color_type == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-            elif self.mat.psychopath.color_type == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-        elif self.mat.psychopath.surface_shader_type == 'Lambert':
-            w.write("Type [Lambert]\n")
-            if self.mat.psychopath.color_type == 'Rec709':
-                col = self.mat.psychopath.color
-                w.write("Color [rec709, %f %f %f]\n" % (
-                    col[0], col[1], col[2],
-                ))
-            elif self.mat.psychopath.color_type == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-            elif self.mat.psychopath.color_type == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-        elif self.mat.psychopath.surface_shader_type == 'GGX':
-            w.write("Type [GGX]\n")
-            if self.mat.psychopath.color_type == 'Rec709':
-                col = self.mat.psychopath.color
-                w.write("Color [rec709, %f %f %f]\n" % (
-                    col[0], col[1], col[2],
-                ))
-            elif self.mat.psychopath.color_type == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-            elif self.mat.psychopath.color_type == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (
-                    self.mat.psychopath.color_blackbody_temp,
-                    1.0,
-                ))
-            w.write("Roughness [%f]\n" % self.mat.psychopath.roughness)
-            w.write("Fresnel [%f]\n" % self.mat.psychopath.fresnel)
-        else:
-            raise "Unsupported surface shader type '%s'" % self.mat.psychopath.surface_shader_type
-        w.unindent()
-        w.write("}\n")
-
-    def cleanup(self):
-        pass

psychoblend/material.py

@@ -0,0 +1,77 @@
+import bpy
+
+from .util import escape_name, mat2str, needs_def_mb, needs_xform_mb, ExportCancelled
+
+class Material:
+    def __init__(self, render_engine, depsgraph, material):
+        self.mat = material
+
+    def take_sample(self, render_engine, depsgraph, time):
+        # TODO: motion blur of material settings
+        pass
+
+    def export(self, render_engine, w):
+        render_engine.update_stats("", "Psychopath: Exporting %s" % self.mat.name)
+
+        w.write("SurfaceShader $%s {\n" % escape_name(self.mat.name))
+        w.indent()
+        if self.mat.psychopath.surface_shader_type == 'Emit':
+            w.write("Type [Emit]\n")
+            if self.mat.psychopath.color_type == 'Rec709':
+                col = self.mat.psychopath.color
+                w.write("Color [rec709, %f %f %f]\n" % (
+                    col[0], col[1], col[2],
+                ))
+            elif self.mat.psychopath.color_type == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+            elif self.mat.psychopath.color_type == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+        elif self.mat.psychopath.surface_shader_type == 'Lambert':
+            w.write("Type [Lambert]\n")
+            if self.mat.psychopath.color_type == 'Rec709':
+                col = self.mat.psychopath.color
+                w.write("Color [rec709, %f %f %f]\n" % (
+                    col[0], col[1], col[2],
+                ))
+            elif self.mat.psychopath.color_type == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+            elif self.mat.psychopath.color_type == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+        elif self.mat.psychopath.surface_shader_type == 'GGX':
+            w.write("Type [GGX]\n")
+            if self.mat.psychopath.color_type == 'Rec709':
+                col = self.mat.psychopath.color
+                w.write("Color [rec709, %f %f %f]\n" % (
+                    col[0], col[1], col[2],
+                ))
+            elif self.mat.psychopath.color_type == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+            elif self.mat.psychopath.color_type == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (
+                    self.mat.psychopath.color_blackbody_temp,
+                    1.0,
+                ))
+            w.write("Roughness [%f]\n" % self.mat.psychopath.roughness)
+            w.write("Fresnel [%f]\n" % self.mat.psychopath.fresnel)
+        else:
+            raise "Unsupported surface shader type '%s'" % self.mat.psychopath.surface_shader_type
+        w.unindent()
+        w.write("}\n")
+
+    def cleanup(self):
+        pass

psychoblend/objects.py

@@ -0,0 +1,206 @@
+import bpy
+
+from .util import escape_name, mat2str, needs_def_mb, needs_xform_mb, ExportCancelled
+from mathutils import Vector, Matrix
+
+def make_object_data_cache(render_engine, depsgraph, ob, name):
+    if ob.type == 'MESH':
+        return Mesh(render_engine, depsgraph, ob, name)
+    elif ob.type == 'LIGHT':
+        if ob.data.type == 'POINT':
+            return SphereLamp(render_engine, depsgraph, ob, name)
+        elif ob.data.type == 'AREA':
+            return RectLamp(render_engine, depsgraph, ob, name)
+        elif ob.data.type == 'AREA':
+            return RectLamp(render_engine, depsgraph, ob, name)
+
+class Mesh:
+    """ Holds data for a mesh to be exported.
+    """
+    def __init__(self, render_engine, depsgraph, ob, name):
+        self.name = name
+        self.material_name = None
+        if len(ob.material_slots) >= 1 and ob.material_slots[0].material != None:
+            self.material_name = ob.material_slots[0].material.name
+        self.is_subdiv = ob.data.psychopath.is_subdivision_surface
+        self.needs_mb = needs_def_mb(ob)
+        self.time_meshes = []
+
+    def take_sample(self, render_engine, depsgraph, ob, time):
+        if len(self.time_meshes) == 0 or self.needs_mb:
+            render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.name, time))
+            self.time_meshes += [ob.to_mesh(depsgraph=depsgraph).copy()]
+    
+    def cleanup(self):
+        for mesh in self.time_meshes:
+            bpy.data.meshes.remove(mesh)
+
+    def export(self, render_engine, w):
+        render_engine.update_stats("", "Psychopath: Exporting %s" % self.name)
+
+        if self.is_subdiv == False:
+            # Exporting normal mesh
+            w.write("MeshSurface $%s {\n" % escape_name(self.name))
+            w.indent()
+        else:
+            # Exporting subdivision surface cage
+            w.write("SubdivisionSurface $%s {\n" % escape_name(self.name))
+            w.indent()
+
+        # Material bindings.
+        if self.material_name != None:
+            w.write("SurfaceShaderBind [${}]\n".format(escape_name(self.material_name)))
+
+        # Write vertices and (if it's smooth shaded) normals
+        for ti in range(len(self.time_meshes)):
+            w.write("Vertices [")
+            w.write(" ".join([("%f" % i) for vert in self.time_meshes[ti].vertices for i in vert.co]), False)
+            w.write("]\n", False)
+            if self.time_meshes[0].polygons[0].use_smooth and self.is_subdiv == False:
+                w.write("Normals [")
+                w.write(" ".join([("%f" % i) for vert in self.time_meshes[ti].vertices for i in vert.normal]), False)
+                w.write("]\n", False)
+
+        # Write face vertex counts
+        w.write("FaceVertCounts [")
+        w.write(" ".join([("%d" % len(p.vertices)) for p in self.time_meshes[0].polygons]), False)
+        w.write("]\n", False)
+
+        # Write face vertex indices
+        w.write("FaceVertIndices [")
+        w.write(" ".join([("%d"%v) for p in self.time_meshes[0].polygons for v in p.vertices]), False)
+        w.write("]\n", False)
+
+        # MeshSurface/SubdivisionSurface section end
+        w.unindent()
+        w.write("}\n")
+
+
+class SphereLamp:
+    """ Holds data for a sphere light to be exported.
+    """
+    def __init__(self, render_engine, depsgraph, ob, name):
+        self.name = name
+        self.time_col = []
+        self.time_rad = []
+
+    def take_sample(self, render_engine, depsgraph, ob, time):
+        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(ob.name, time))
+
+        if ob.data.psychopath.color_type == 'Rec709':
+            self.time_col += [('Rec709', ob.data.color * ob.data.energy)]
+        elif ob.data.psychopath.color_type == 'Blackbody':
+            self.time_col += [('Blackbody', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+        elif ob.data.psychopath.color_type == 'ColorTemperature':
+            self.time_col += [('ColorTemperature', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+
+        self.time_rad += [ob.data.shadow_soft_size]
+
+    def cleanup(self):
+        pass
+    
+    def export(self, render_engine, w):
+        render_engine.update_stats("", "Psychopath: Exporting %s" % self.name)
+
+        w.write("SphereLight $%s {\n" % escape_name(self.name))
+        w.indent()
+        for col in self.time_col:
+            if col[0] == 'Rec709':
+                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
+            elif col[0] == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
+            elif col[0] == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
+        for rad in self.time_rad:
+            w.write("Radius [%f]\n" % rad)
+
+        w.unindent()
+        w.write("}\n")
+
+
+class RectLamp:
+    """ Holds data for a rectangular light to be exported.
+    """
+    def __init__(self, render_engine, depsgraph, ob, name):
+        self.name = name
+        self.time_col = []
+        self.time_dim = []
+
+    def take_sample(self, render_engine, depsgraph, ob, time):
+        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.name, time))
+
+        if ob.data.psychopath.color_type == 'Rec709':
+            self.time_col += [('Rec709', ob.data.color * ob.data.energy / 2)]
+        elif ob.data.psychopath.color_type == 'Blackbody':
+            self.time_col += [('Blackbody', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+        elif ob.data.psychopath.color_type == 'ColorTemperature':
+            self.time_col += [('ColorTemperature', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+
+        if ob.data.shape == 'RECTANGLE':
+            self.time_dim += [(ob.data.size, ob.data.size_y)]
+        else:
+            self.time_dim += [(ob.data.size, ob.data.size)]
+
+    def cleanup(self):
+        pass
+    
+    def export(self, render_engine, w):
+        render_engine.update_stats("", "Psychopath: Exporting %s" % self.name)
+
+        w.write("RectangleLight $%s {\n" % escape_name(self.name))
+        w.indent()
+        for col in self.time_col:
+            if col[0] == 'Rec709':
+                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
+            elif col[0] == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
+            elif col[0] == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
+        for dim in self.time_dim:
+            w.write("Dimensions [%f %f]\n" % dim)
+
+        w.unindent()
+        w.write("}\n")
+
+
+class DistantDiskLamp:
+    def __init__(self, render_engine, depsgraph, ob, name):
+        self.name = name
+        self.time_col = []
+        self.time_dir = []
+        self.time_rad = []
+
+    def take_sample(self, render_engine, depsgraph, ob, time):
+        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.name, time))
+        self.time_dir += [tuple(ob.matrix_world.to_3x3() @ Vector((0, 0, -1)))]
+
+        if ob.data.psychopath.color_type == 'Rec709':
+            self.time_col += [('Rec709', ob.data.color * ob.data.energy)]
+        elif ob.data.psychopath.color_type == 'Blackbody':
+            self.time_col += [('Blackbody', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+        elif ob.data.psychopath.color_type == 'ColorTemperature':
+            self.time_col += [('ColorTemperature', ob.data.psychopath.color_blackbody_temp, ob.data.energy)]
+
+        self.time_rad += [ob.data.shadow_soft_size]
+
+    def cleanup(self):
+        pass
+
+    def export(self, render_engine, w):
+        render_engine.update_stats("", "Psychopath: Exporting %s" % escape_name(self.name))
+        w.write("DistantDiskLight $%s {\n" % self.name)
+        w.indent()
+        for direc in self.time_dir:
+            w.write("Direction [%f %f %f]\n" % (direc[0], direc[1], direc[2]))
+        for col in self.time_col:
+            if col[0] == 'Rec709':
+                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
+            elif col[0] == 'Blackbody':
+                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
+            elif col[0] == 'ColorTemperature':
+                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
+        for rad in self.time_rad:
+            w.write("Radius [%f]\n" % rad)
+
+        w.unindent()
+        w.write("}\n")

psychoblend/psy_export.py

@@ -2,9 +2,11 @@ import bpy
 
 from math import log
 
-from .assembly import Assembly
+from .material import Material
+from .objects import make_object_data_cache, Mesh, DistantDiskLamp
 from .util import escape_name, mat2str, ExportCancelled
-from .world import World
+from .world import World, Camera
+from . import bl_info
 
 
 class IndentedWriter:
@@ -29,25 +31,43 @@ class IndentedWriter:
 
 
 class PsychoExporter:
-    def __init__(self, f, render_engine, scene):
+    def __init__(self, f, render_engine, depsgraph):
         self.w = IndentedWriter(f)
         self.render_engine = render_engine
-        self.scene = scene
+        self.depsgraph = depsgraph
+        self.scene = depsgraph.scene
+        self.view_layer = depsgraph.view_layer
 
-        self.mesh_names = {}
-        self.group_names = {}
+        # For camera data.
+        res_x = int(self.scene.render.resolution_x * (self.scene.render.resolution_percentage / 100))
+        res_y = int(self.scene.render.resolution_y * (self.scene.render.resolution_percentage / 100))
+        self.camera = Camera(render_engine, depsgraph.scene.camera, float(res_x) / float(res_y))
 
-        # Motion blur segments are rounded down to a power of two
-        if scene.psychopath.motion_blur_segments > 0:
-            self.time_samples = (2**int(log(scene.psychopath.motion_blur_segments, 2))) + 1
+        # For world data.
+        self.world = World(render_engine, depsgraph)
+
+        # For all objects except sun lamps.
+        self.object_data = {} # name -> cached_data
+        self.instances = {} # instance_id -> [object_data_name, transform_list]
+
+        # For all sun lamps.
+        self.sun_lamp_data = {} # name -> cached_data
+        self.sun_lamp_instances = {} # instance_id -> [sun_lamp_data_name, transform_list]
+
+        # For all materials.
+        self.materials = {} # name -> cached_data
+
+        # Motion blur segments are rounded down to a power of two.
+        if self.scene.psychopath.motion_blur_segments > 0:
+            self.time_samples = (2**int(log(self.scene.psychopath.motion_blur_segments, 2))) + 1
         else:
             self.time_samples = 1
 
-        # pre-calculate useful values for exporting motion blur
-        self.shutter_start = scene.psychopath.shutter_start
-        self.shutter_diff = (scene.psychopath.shutter_end - scene.psychopath.shutter_start) / max(1, (self.time_samples-1))
+        # pre-calculate useful values for exporting motion blur.
+        self.shutter_start = self.scene.psychopath.shutter_start
+        self.shutter_diff = (self.scene.psychopath.shutter_end - self.scene.psychopath.shutter_start) / max(1, (self.time_samples-1))
 
-        self.fr = scene.frame_current
+        self.fr = self.scene.frame_current
 
 
     def set_frame(self, frame, fraction):
@@ -70,25 +90,31 @@ class PsychoExporter:
 
     def _export_psy(self):
         # Info
-        self.w.write("# Exported from Blender 2.7x\n")
+        self.w.write("# Exported from Blender {} with PsychoBlend {}.{}.{}\n".format(
+            bpy.app.version_string,
+            bl_info["version"][0],
+            bl_info["version"][1],
+            bl_info["version"][2],
+        ))
 
         # Scene begin
         self.w.write("\n\nScene $%s_fr%d {\n" % (escape_name(self.scene.name), self.fr))
         self.w.indent()
 
-        #######################
-        # Output section begin
+        #------------------------------------------------------
+        # Output section.
+
         self.w.write("Output {\n")
         self.w.indent()
 
         self.w.write('Path [""]\n')
 
-        # Output section end
         self.w.unindent()
         self.w.write("}\n")
 
-        ###############################
-        # RenderSettings section begin
+        #------------------------------------------------------
+        # RenderSettings section.
+
         self.w.write("RenderSettings {\n")
         self.w.indent()
 
@@ -99,34 +125,145 @@ class PsychoExporter:
         self.w.write("DicingRate [%f]\n" % self.scene.psychopath.dicing_rate)
         self.w.write('Seed [%d]\n' % self.fr)
 
-        # RenderSettings section end
         self.w.unindent()
         self.w.write("}\n")
 
-        ###############################
-        # Export world and object data
-        world = None
-        root_assembly = None
+        #------------------------------------------------------
+        # Collect materials.
+
+        # TODO: handle situations where there are more than one
+        # material with the same name.  This can happen through
+        # library linking.
+
+        for inst in self.depsgraph.object_instances:
+            ob = inst.object
+            if ob.type in ['MESH']:
+                for ms in ob.material_slots:
+                    if ms.material != None:
+                        if ms.material.name not in self.materials:
+                            self.materials[ms.material.name] = Material(self.render_engine, self.depsgraph, ms.material)
+
+        #------------------------------------------------------
+        # Collect world and object data.
+
         try:
-            # Prep for data collection
-            world = World(self.render_engine, self.scene, self.scene.layers, float(res_x) / float(res_y))
-            root_assembly = Assembly(self.render_engine, self.scene.objects, self.scene.layers)
-
-            # Collect data for each time sample
             for i in range(self.time_samples):
-                time = self.fr + self.shutter_start + (self.shutter_diff*i)
-                self.set_frame(self.fr, self.shutter_start + (self.shutter_diff*i))
-                world.take_sample(self.render_engine, self.scene, time)
-                root_assembly.take_sample(self.render_engine, self.scene, time)
+                # Check if render is cancelled
+                if self.render_engine.test_break():
+                    raise ExportCancelled()
 
-            # Export collected data
-            world.export(self.render_engine, self.w)
-            root_assembly.export(self.render_engine, self.w)
-        finally:
-            if world != None:
-                world.cleanup()
-            if root_assembly != None:
-                root_assembly.cleanup()
+                subframe = self.shutter_start + (self.shutter_diff*i)
+                time = self.fr + subframe
+                self.depsgraph.scene.frame_set(self.fr, subframe=subframe)
+                self.depsgraph.update()
+
+                # Collect camera and world data.
+                self.camera.take_sample(self.render_engine, self.depsgraph, time)
+                self.world.take_sample(self.render_engine, self.depsgraph, time)
+
+                # Collect renderable objects.
+                collected_objs = set() # Names of the objects whose data has already been collected.
+                for inst in self.depsgraph.object_instances:
+                    # Check if render is cancelled
+                    if self.render_engine.test_break():
+                        raise ExportCancelled()
+
+                    if inst.object.type not in ['MESH', 'LIGHT']:
+                        continue
+
+                    # We use this a couple of times, so make a shorthand.
+                    is_sun_lamp = inst.object.type == 'LIGHT' and inst.object.data.type == 'SUN'
+
+                    # TODO: handle situations where there are more than one
+                    # object with the same name.  This can happen through
+                    # library linking.
+
+                    # Get a unique id for the instance.  This is surprisingly
+                    # tricky, because the instance's "persistent_id" property
+                    # isn't globally unique, as I would have expected from
+                    # the documentation.
+                    id = None
+                    if inst.is_instance:
+                        id = (
+                            hash((inst.object.name, inst.parent.name)),
+                            # Has to be turned into a tuple, otherwise it doesn't
+                            # work as part of the ID for some reason.
+                            tuple(inst.persistent_id),
+                        )
+                    else:
+                        id = inst.object.name
+
+                    # Save the instance transforms.
+                    if is_sun_lamp:
+                        if id not in self.sun_lamp_instances:
+                            self.sun_lamp_instances[id] = [inst.object.name, [inst.matrix_world.copy()]]
+                        else:
+                            self.sun_lamp_instances[id][1] += [inst.matrix_world.copy()]
+                    else:
+                        if id not in self.instances:
+                            self.instances[id] = [inst.object.name, [inst.matrix_world.copy()]]
+                        else:
+                            self.instances[id][1] += [inst.matrix_world.copy()]
+
+                    # Save the object data if it hasn't already been saved.
+                    if inst.object.name not in collected_objs:
+                        collected_objs.add(inst.object.name)
+                        if is_sun_lamp:
+                            if inst.object.name not in self.sun_lamp_data:
+                                self.sun_lamp_data[inst.object.name] = DistantDiskLamp(self.render_engine, self.depsgraph, inst.object, inst.object.name)
+                            self.sun_lamp_data[inst.object.name].take_sample(self.render_engine, self.depsgraph, inst.object, time)
+                        else:
+                            if inst.object.name not in self.object_data:
+                                self.object_data[inst.object.name] = make_object_data_cache(self.render_engine, self.depsgraph, inst.object, inst.object.name)
+                            self.object_data[inst.object.name].take_sample(self.render_engine, self.depsgraph, inst.object, time)
+
+            #------------------------------------------------------
+            # Export world and object data.
+
+            self.camera.export(self.render_engine, self.w)
+            self.world.export(self.render_engine, self.w)
+
+            self.w.write("Assembly {\n")
+            self.w.indent()
+
+            # Export materials.
+            for name in self.materials:
+                self.materials[name].export(self.render_engine, self.w)
+
+            # Export objects.
+            for name in self.object_data:
+                self.object_data[name].export(self.render_engine, self.w)
+
+            # Export instances.
+            for id in self.instances:
+                [obj_name, xforms] = self.instances[id]
+                self.render_engine.update_stats("", "Psychopath: Exporting %s instance" % obj_name)
+
+                prefix = str(hex(hash(id)))
+                name = "inst_{}__{}".format(prefix, escape_name(obj_name))
+
+                self.w.write("Instance {\n")
+                self.w.indent()
+                self.w.write("Data [${}]\n".format(escape_name(obj_name)))
+                for mat in xforms:
+                    self.w.write("Transform [{}]\n".format(mat2str(mat)))
+                self.w.unindent()
+                self.w.write("}\n")
+
+            self.w.unindent()
+            self.w.write("}\n")
+        finally:            
+            #------------------------------------------------------
+            # Cleanup collected data.
+
+            self.camera.cleanup()
+            self.world.cleanup()
+
+            for data in self.sun_lamp_data:
+                self.sun_lamp_data[data].cleanup()
+
+            for data in self.object_data:
+                self.object_data[data].cleanup()
 
         # Scene end
         self.w.unindent()

psychoblend/render.py

@@ -9,11 +9,37 @@ from . import psy_export
 class PsychopathRender(bpy.types.RenderEngine):
     bl_idname = 'PSYCHOPATH_RENDER'
     bl_label = "Psychopath"
-    DELAY = 1.0
+    bl_use_preview = False
+
+    def __init__(self):
+        pass
+
+    def __del__(self):
+        pass
+
+    def update(self, data, depsgraph):
+        pass
+
+    def render(self, depsgraph):
+        self._process = None
+        try:
+            self._render(depsgraph)
+        except:
+            if self._process != None:
+                self._process.terminate()
+            raise
+
+    def view_update(self, context, depsgraph):
+        pass
+
+    def view_draw(self, context, depsgraph):
+        pass
+
+    #----------------------------------------------------------
 
     @staticmethod
     def _locate_binary():
-        addon_prefs = bpy.context.user_preferences.addons[__package__].preferences
+        addon_prefs = bpy.context.preferences.addons[__package__].preferences
 
         # Use the system preference if its set.
         psy_binary = addon_prefs.filepath_psychopath
@@ -23,7 +49,7 @@ class PsychopathRender(bpy.types.RenderEngine):
             else:
                 print("User Preference to psychopath %r NOT FOUND, checking $PATH" % psy_binary)
 
-        # search the path all os's
+        # Search for the path.
         psy_binary_default = "psychopath"
 
         os_path_ls = os.getenv("PATH").split(':') + [""]
@@ -45,13 +71,13 @@ class PsychopathRender(bpy.types.RenderEngine):
         if crop != None:
             args += ["--crop", str(crop[0]), str(self.size_y - crop[3]), str(crop[2] - 1), str(self.size_y - crop[1] - 1)]
         if use_stdin:
-            args += ["--spb", str(scene.psychopath.max_samples_per_bucket), "--serialized_output", "--use_stdin"]
+            args += ["--bucket_size", str(scene.psychopath.bucket_size), "--serialized_output", "--use_stdin"]
         else:
-            args += ["--spb", str(scene.psychopath.max_samples_per_bucket), "--serialized_output", "-i", psy_filepath]
+            args += ["--bucket_size", str(scene.psychopath.bucket_size), "--serialized_output", "-i", psy_filepath]
 
         # Start Rendering!
         try:
-            self._process = subprocess.Popen([psy_binary] + args, bufsize=1, stdin=subprocess.PIPE, stdout=subprocess.PIPE)
+            self._process = subprocess.Popen([psy_binary] + args, stdin=subprocess.PIPE, stdout=subprocess.PIPE)
         except OSError:
             # TODO, report api
             print("Psychopath: could not execute '%s'" % psy_binary)
@@ -73,28 +99,21 @@ class PsychopathRender(bpy.types.RenderEngine):
         height = bucket_info[3] - bucket_info[1]
 
         # Decode pixel data
-        pixels = [p for p in struct.iter_unpack("ffff", base64.b64decode(pixels_encoded))]
-        pixels_flipped = []
+        pixels_unpacked = [p for p in struct.iter_unpack("ffff", base64.b64decode(pixels_encoded))]
+        pixels = []
         for i in range(height):
             n = height - i - 1
-            pixels_flipped += pixels[n*width:(n+1)*width]
+            pixels += pixels_unpacked[n*width:(n+1)*width]
 
         # Write pixel data to render image
         result = self.begin_result(x, y, width, height)
         lay = result.layers[0].passes["Combined"]
-        lay.rect = pixels_flipped
+        lay.rect = pixels
         self.end_result(result)
 
-    def render(self, scene):
-        self._process = None
-        try:
-            self._render(scene)
-        except:
-            if self._process != None:
-                self._process.terminate()
-            raise
+    def _render(self, depsgraph):
+        scene = depsgraph.scene
 
-    def _render(self, scene):
         # has to be called to update the frame on exporting animations
         scene.frame_set(scene.frame_current)
 
@@ -130,8 +149,8 @@ class PsychopathRender(bpy.types.RenderEngine):
                 return
 
             self.update_stats("", "Psychopath: Collecting...")
-            # Export to Psychopath's stdin
-            if not psy_export.PsychoExporter(self._process.stdin, self, scene).export_psy():
+            # Export to Psychopath's stdin.
+            if not psy_export.PsychoExporter(self._process.stdin, self, depsgraph).export_psy():
                 # Render cancelled in the middle of exporting,
                 # so just return.
                 self._process.terminate()
@@ -142,7 +161,7 @@ class PsychopathRender(bpy.types.RenderEngine):
             # Export to file
             self.update_stats("", "Psychopath: Exporting data from Blender")
             with open(export_path, 'w+b') as f:
-                if not psy_export.PsychoExporter(f, self, scene).export_psy():
+                if not psy_export.PsychoExporter(f, self, depsgraph).export_psy():
                     # Render cancelled in the middle of exporting,
                     # so just return.
                     return
@@ -183,7 +202,7 @@ class PsychopathRender(bpy.types.RenderEngine):
             # Get render output from stdin
             tmp = self._process.stdout.read1(2**16)
             if len(tmp) == 0:
-                time.sleep(0.0001) # Don't spin on the CPU
+                time.sleep(0.001) # Don't spin on the CPU
                 if render_process_finished:
                     all_output_consumed = True
                 continue

psychoblend/ui.py

@@ -1,18 +1,291 @@
 import bpy
 
-# Use some of the existing buttons.
-from bl_ui import properties_render
-properties_render.RENDER_PT_render.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-properties_render.RENDER_PT_dimensions.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-properties_render.RENDER_PT_output.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-del properties_render
+#--------------------------------------------------------------
+# Specify which existing Blender UI panels Psychopath
+# uses/is compatible with.
 
-from bl_ui import properties_data_camera
-properties_data_camera.DATA_PT_lens.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-properties_data_camera.DATA_PT_camera.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-properties_data_camera.DATA_PT_camera_display.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-properties_data_camera.DATA_PT_custom_props_camera.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')
-del properties_data_camera
+import bl_ui
+
+def register_engine_with_panels(area, list):
+    # TODO: reverse this, so we're checking if the list item is
+    # in the real panels, and throw an error if it's not.  That
+    # way things don't just silently fail.
+    for p in list:
+        eval("bl_ui.{}.{}.COMPAT_ENGINES.add('PSYCHOPATH_RENDER')".format(area, p))
+
+register_engine_with_panels(
+    "properties_render",
+    [
+        "RENDER_PT_color_management",
+        "RENDER_PT_color_management_curves",
+    ]
+)
+
+register_engine_with_panels(
+    "properties_output",
+    [
+        "RENDER_PT_encoding",
+        "RENDER_PT_encoding_audio",
+        "RENDER_PT_encoding_video",
+        "RENDER_PT_format",
+        "RENDER_PT_frame_range",
+        "RENDER_PT_output",
+        "RENDER_PT_output_views",
+        "RENDER_PT_post_processing",
+        "RENDER_PT_stamp",
+        "RENDER_PT_stamp_burn",
+        "RENDER_PT_stamp_note",
+        # "RENDER_PT_stereoscopy",
+        "RENDER_PT_time_stretching",
+    ]
+)
+
+register_engine_with_panels(
+    "properties_view_layer",
+    [
+        "VIEWLAYER_PT_layer",
+    ]
+)
+
+register_engine_with_panels(
+    "properties_data_camera",
+    [
+        "DATA_PT_context_camera",
+        "DATA_PT_custom_props_camera",
+        "DATA_PT_camera",
+        "DATA_PT_lens",
+        "DATA_PT_camera_dof",
+        "DATA_PT_camera_dof_aperture",
+        "DATA_PT_camera_display",
+        "DATA_PT_camera_display_composition_guides",
+        "DATA_PT_camera_safe_areas",
+        "DATA_PT_camera_safe_areas_center_cut",
+        "DATA_PT_camera_background_image",
+    ]
+)
+
+register_engine_with_panels(
+    "properties_data_mesh",
+    [
+        "DATA_PT_context_mesh",
+        "DATA_PT_custom_props_mesh",
+        "DATA_PT_customdata",
+        "DATA_PT_face_maps",
+        "DATA_PT_mesh_attributes",
+        "DATA_PT_normals",
+        "DATA_PT_remesh",
+        "DATA_PT_shape_keys",
+        "DATA_PT_texture_space",
+        "DATA_PT_uv_texture",
+        "DATA_PT_vertex_colors",
+        "DATA_PT_vertex_groups",
+    ]
+)
+
+register_engine_with_panels(
+    "properties_particle",
+    [
+        "PARTICLE_MT_context_menu",
+        "PARTICLE_PT_boidbrain",
+        "PARTICLE_PT_cache",
+        "PARTICLE_PT_children",
+        "PARTICLE_PT_children_clumping",
+        "PARTICLE_PT_children_clumping_noise",
+        "PARTICLE_PT_children_kink",
+        "PARTICLE_PT_children_parting",
+        "PARTICLE_PT_children_roughness",
+        "PARTICLE_PT_context_particles",
+        "PARTICLE_PT_custom_props",
+        "PARTICLE_PT_draw",
+        "PARTICLE_PT_emission",
+        "PARTICLE_PT_emission_source",
+        "PARTICLE_PT_field_weights",
+        "PARTICLE_PT_force_fields",
+        "PARTICLE_PT_force_fields_type1",
+        "PARTICLE_PT_force_fields_type1_falloff",
+        "PARTICLE_PT_force_fields_type2",
+        "PARTICLE_PT_force_fields_type2_falloff",
+        "PARTICLE_PT_hair_dynamics",
+        "PARTICLE_PT_hair_dynamics_collision",
+        "PARTICLE_PT_hair_dynamics_presets",
+        "PARTICLE_PT_hair_dynamics_structure",
+        "PARTICLE_PT_hair_dynamics_volume",
+        "PARTICLE_PT_hair_shape",
+        "PARTICLE_PT_physics",
+        "PARTICLE_PT_physics_boids_battle",
+        "PARTICLE_PT_physics_boids_misc",
+        "PARTICLE_PT_physics_boids_movement",
+        "PARTICLE_PT_physics_deflection",
+        "PARTICLE_PT_physics_fluid_advanced",
+        "PARTICLE_PT_physics_fluid_interaction",
+        "PARTICLE_PT_physics_fluid_springs",
+        "PARTICLE_PT_physics_fluid_springs_advanced",
+        "PARTICLE_PT_physics_fluid_springs_viscoelastic",
+        "PARTICLE_PT_physics_forces",
+        "PARTICLE_PT_physics_integration",
+        "PARTICLE_PT_physics_relations",
+        "PARTICLE_PT_render",
+        "PARTICLE_PT_render_collection",
+        "PARTICLE_PT_render_collection_use_count",
+        "PARTICLE_PT_render_extra",
+        "PARTICLE_PT_render_object",
+        "PARTICLE_PT_render_path",
+        "PARTICLE_PT_render_path_timing",
+        "PARTICLE_PT_rotation",
+        "PARTICLE_PT_rotation_angular_velocity",
+        "PARTICLE_PT_textures",
+        "PARTICLE_PT_velocity",
+        "PARTICLE_PT_vertexgroups",
+    ]
+)
+
+# Physics.
+# (Why these require renderer compatibility settings
+# is beyond me.  But they do.)
+register_engine_with_panels(
+    "properties_physics_cloth",
+    [
+        "PHYSICS_PT_cloth",
+        "PHYSICS_PT_cloth_cache",
+        "PHYSICS_PT_cloth_collision",
+        "PHYSICS_PT_cloth_damping",
+        "PHYSICS_PT_cloth_field_weights",
+        "PHYSICS_PT_cloth_internal_springs",
+        "PHYSICS_PT_cloth_object_collision",
+        "PHYSICS_PT_cloth_physical_properties",
+        "PHYSICS_PT_cloth_pressure",
+        "PHYSICS_PT_cloth_property_weights",
+        "PHYSICS_PT_cloth_self_collision",
+        "PHYSICS_PT_cloth_shape",
+        "PHYSICS_PT_cloth_stiffness", 
+    ],
+)
+register_engine_with_panels(
+    "properties_physics_common",
+    [
+        "PHYSICS_PT_add",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_dynamicpaint",
+    [
+        "PHYSICS_PT_dp_brush_source",
+        "PHYSICS_PT_dp_brush_source_color_ramp",
+        "PHYSICS_PT_dp_brush_velocity",
+        "PHYSICS_PT_dp_brush_velocity_color_ramp",
+        "PHYSICS_PT_dp_brush_velocity_smudge",
+        "PHYSICS_PT_dp_brush_wave",
+        "PHYSICS_PT_dp_cache",
+        "PHYSICS_PT_dp_canvas_initial_color",
+        "PHYSICS_PT_dp_canvas_output",
+        "PHYSICS_PT_dp_canvas_output_paintmaps",
+        "PHYSICS_PT_dp_canvas_output_wetmaps",
+        "PHYSICS_PT_dp_effects",
+        "PHYSICS_PT_dp_effects_drip",
+        "PHYSICS_PT_dp_effects_drip_weights",
+        "PHYSICS_PT_dp_effects_shrink",
+        "PHYSICS_PT_dp_effects_spread",
+        "PHYSICS_PT_dp_surface_canvas",
+        "PHYSICS_PT_dp_surface_canvas_paint_dissolve",
+        "PHYSICS_PT_dp_surface_canvas_paint_dry",
+        "PHYSICS_PT_dynamic_paint",
+        "PHYSICS_PT_dynamic_paint_settings",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_field",
+    [
+        "PHYSICS_PT_collision",
+        "PHYSICS_PT_collision_particle",
+        "PHYSICS_PT_collision_softbody",
+        "PHYSICS_PT_field",
+        "PHYSICS_PT_field_falloff",
+        "PHYSICS_PT_field_falloff_angular",
+        "PHYSICS_PT_field_falloff_radial",
+        "PHYSICS_PT_field_settings",
+        "PHYSICS_PT_field_settings_kink",
+        "PHYSICS_PT_field_settings_texture_select",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_fluid",
+    [
+        "PHYSICS_PT_adaptive_domain",
+        "PHYSICS_PT_borders",
+        "PHYSICS_PT_cache",
+        "PHYSICS_PT_collections",
+        "PHYSICS_PT_diffusion",
+        "PHYSICS_PT_export",
+        "PHYSICS_PT_field_weights",
+        "PHYSICS_PT_fire",
+        "PHYSICS_PT_flow_initial_velocity",
+        "PHYSICS_PT_flow_source",
+        "PHYSICS_PT_flow_texture",
+        "PHYSICS_PT_fluid",
+        "PHYSICS_PT_guide",
+        "PHYSICS_PT_liquid",
+        "PHYSICS_PT_mesh",
+        "PHYSICS_PT_noise",
+        "PHYSICS_PT_particles",
+        "PHYSICS_PT_settings",
+        "PHYSICS_PT_smoke",
+        "PHYSICS_PT_smoke_dissolve",
+        "PHYSICS_PT_viscosity",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_rigidbody",
+    [
+        "PHYSICS_PT_rigid_body",
+        "PHYSICS_PT_rigid_body_collisions",
+        "PHYSICS_PT_rigid_body_collisions_collections",
+        "PHYSICS_PT_rigid_body_collisions_sensitivity",
+        "PHYSICS_PT_rigid_body_collisions_surface",
+        "PHYSICS_PT_rigid_body_dynamics",
+        "PHYSICS_PT_rigid_body_dynamics_deactivation",
+        "PHYSICS_PT_rigid_body_settings",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_rigidbody_constraint",
+    [
+        "PHYSICS_PT_rigid_body_constraint",
+        "PHYSICS_PT_rigid_body_constraint_limits",
+        "PHYSICS_PT_rigid_body_constraint_limits_angular",
+        "PHYSICS_PT_rigid_body_constraint_limits_linear",
+        "PHYSICS_PT_rigid_body_constraint_motor",
+        "PHYSICS_PT_rigid_body_constraint_motor_angular",
+        "PHYSICS_PT_rigid_body_constraint_motor_linear",
+        "PHYSICS_PT_rigid_body_constraint_objects",
+        "PHYSICS_PT_rigid_body_constraint_override_iterations",
+        "PHYSICS_PT_rigid_body_constraint_settings",
+        "PHYSICS_PT_rigid_body_constraint_springs",
+        "PHYSICS_PT_rigid_body_constraint_springs_angular",
+        "PHYSICS_PT_rigid_body_constraint_springs_linear",
+    ]
+)
+register_engine_with_panels(
+    "properties_physics_softbody",
+    [
+        "PHYSICS_PT_softbody",
+        "PHYSICS_PT_softbody_cache",
+        "PHYSICS_PT_softbody_collision",
+        "PHYSICS_PT_softbody_edge",
+        "PHYSICS_PT_softbody_edge_aerodynamics",
+        "PHYSICS_PT_softbody_edge_stiffness",
+        "PHYSICS_PT_softbody_field_weights",
+        "PHYSICS_PT_softbody_goal",
+        "PHYSICS_PT_softbody_goal_settings",
+        "PHYSICS_PT_softbody_goal_strengths",
+        "PHYSICS_PT_softbody_object",
+        "PHYSICS_PT_softbody_simulation",
+        "PHYSICS_PT_softbody_solver",
+        "PHYSICS_PT_softbody_solver_diagnostics",
+        "PHYSICS_PT_softbody_solver_helpers",
+    ]
+)
+
+#--------------------------------------------------------------
 
 class PsychopathPanel():
     COMPAT_ENGINES = {'PSYCHOPATH_RENDER'}
@@ -20,7 +293,7 @@ class PsychopathPanel():
     @classmethod
     def poll(cls, context):
         rd = context.scene.render
-        return (rd.use_game_engine is False) and (rd.engine in cls.COMPAT_ENGINES)
+        return rd.engine in cls.COMPAT_ENGINES
 
 
 class RENDER_PT_psychopath_render_settings(PsychopathPanel, bpy.types.Panel):
@@ -47,7 +320,7 @@ class RENDER_PT_psychopath_render_settings(PsychopathPanel, bpy.types.Panel):
         col.prop(scene.psychopath, "shutter_end")
 
         col.label(text="Performance")
-        col.prop(scene.psychopath, "max_samples_per_bucket")
+        col.prop(scene.psychopath, "bucket_size")
 
 
 class RENDER_PT_psychopath_export_settings(PsychopathPanel, bpy.types.Panel):
@@ -78,29 +351,7 @@ class WORLD_PT_psychopath_background(PsychopathPanel, bpy.types.Panel):
         layout = self.layout
 
         world = context.world
-        layout.prop(world, "horizon_color", text="Color")
-
-
-class DATA_PT_psychopath_camera_dof(PsychopathPanel, bpy.types.Panel):
-    bl_label = "Depth of Field"
-    bl_space_type = 'PROPERTIES'
-    bl_region_type = 'WINDOW'
-    bl_context = "data"
-
-    @classmethod
-    def poll(cls, context):
-        engine = context.scene.render.engine
-        return context.camera and PsychopathPanel.poll(context)
-
-    def draw(self, context):
-        ob = context.active_object
-        layout = self.layout
-
-        col = layout.column()
-
-        col.prop(ob.data, "dof_object")
-        col.prop(ob.data, "dof_distance")
-        col.prop(ob.data.psychopath, "aperture_radius")
+        layout.prop(world.psychopath, "background_color", text="Color")
 
 
 class DATA_PT_psychopath_lamp(PsychopathPanel, bpy.types.Panel):
@@ -112,7 +363,7 @@ class DATA_PT_psychopath_lamp(PsychopathPanel, bpy.types.Panel):
     @classmethod
     def poll(cls, context):
         engine = context.scene.render.engine
-        return context.lamp and PsychopathPanel.poll(context)
+        return context.active_object.type == 'LIGHT' and PsychopathPanel.poll(context)
 
     def draw(self, context):
         ob = context.active_object
@@ -144,22 +395,23 @@ class DATA_PT_psychopath_area_lamp(PsychopathPanel, bpy.types.Panel):
 
     @classmethod
     def poll(cls, context):
-        lamp = context.lamp
         engine = context.scene.render.engine
-        return (lamp and lamp.type == 'AREA') and (engine in cls.COMPAT_ENGINES)
+        return context.active_object.type == 'LIGHT' \
+            and context.active_object.data.type == 'AREA' \
+            and (engine in cls.COMPAT_ENGINES)
 
     def draw(self, context):
         layout = self.layout
 
-        lamp = context.lamp
+        lamp = context.active_object.data
 
         col = layout.column()
         col.row().prop(lamp, "shape", expand=True)
         sub = col.row(align=True)
 
-        if lamp.shape == 'SQUARE':
+        if lamp.shape == 'SQUARE' or lamp.shape == 'DISK':
             sub.prop(lamp, "size")
-        elif lamp.shape == 'RECTANGLE':
+        elif lamp.shape == 'RECTANGLE' or lamp.shape == 'ELLIPSE':
             sub.prop(lamp, "size", text="Size X")
             sub.prop(lamp, "size_y", text="Size Y")
 
@@ -208,10 +460,10 @@ class MATERIAL_PT_psychopath_context_material(PsychopathPanel, bpy.types.Panel):
             row.template_list("MATERIAL_UL_matslots", "", ob, "material_slots", ob, "active_material_index", rows=1)
 
             col = row.column(align=True)
-            col.operator("object.material_slot_add", icon='ZOOMIN', text="")
-            col.operator("object.material_slot_remove", icon='ZOOMOUT', text="")
+            col.operator("object.material_slot_add", icon='ADD', text="")
+            col.operator("object.material_slot_remove", icon='REMOVE', text="")
 
-            col.menu("MATERIAL_MT_specials", icon='DOWNARROW_HLT', text="")
+            col.menu("MATERIAL_MT_context_menu", icon='DOWNARROW_HLT', text="")
 
             if ob.mode == 'EDIT':
                 row = layout.row(align=True)
@@ -219,7 +471,7 @@ class MATERIAL_PT_psychopath_context_material(PsychopathPanel, bpy.types.Panel):
                 row.operator("object.material_slot_select", text="Select")
                 row.operator("object.material_slot_deselect", text="Deselect")
 
-        split = layout.split(percentage=0.65)
+        split = layout.split(factor=0.65)
 
         if ob:
             split.template_ID(ob, "active_material", new="material.new")
@@ -271,7 +523,6 @@ def register():
     bpy.utils.register_class(RENDER_PT_psychopath_render_settings)
     bpy.utils.register_class(RENDER_PT_psychopath_export_settings)
     bpy.utils.register_class(WORLD_PT_psychopath_background)
-    bpy.utils.register_class(DATA_PT_psychopath_camera_dof)
     bpy.utils.register_class(DATA_PT_psychopath_mesh)
     bpy.utils.register_class(DATA_PT_psychopath_lamp)
     bpy.utils.register_class(DATA_PT_psychopath_area_lamp)
@@ -282,8 +533,7 @@ def unregister():
     bpy.utils.unregister_class(RENDER_PT_psychopath_render_settings)
     bpy.utils.unregister_class(RENDER_PT_psychopath_export_settings)
     bpy.utils.unregister_class(WORLD_PT_psychopath_background)
-    bpy.utils.unregister_class(DATA_PT_psychopath_camera_dof)
-    bpy.utils.register_class(DATA_PT_psychopath_mesh)
+    bpy.utils.unregister_class(DATA_PT_psychopath_mesh)
     bpy.utils.unregister_class(DATA_PT_psychopath_lamp)
     bpy.utils.unregister_class(DATA_PT_psychopath_area_lamp)
     bpy.utils.unregister_class(MATERIAL_PT_psychopath_context_material)

psychoblend/world.py

@@ -1,52 +1,33 @@
 import bpy
 
-from math import degrees, tan, atan
+from math import degrees, sin, asin, tan, atan
 from mathutils import Vector, Matrix
 
 from .util import escape_name, mat2str, ExportCancelled
 
 class World:
-    def __init__(self, render_engine, scene, visible_layers, aspect_ratio):
+    def __init__(self, render_engine, depsgraph):
+        scene = depsgraph.scene
         self.background_shader = BackgroundShader(render_engine, scene.world)
-        self.camera = Camera(render_engine, scene.camera, aspect_ratio)
-        self.lights = []
 
-        # Collect infinite-extent light sources.
-        # TODO: also get sun lamps inside group instances.
-        for ob in scene.objects:
-            if ob.type == 'LAMP' and ob.data.type == 'SUN':
-                name = escape_name(ob.name)
-                self.lights += [DistantDiskLamp(ob, name)]
 
-    def take_sample(self, render_engine, scene, time):
-        self.camera.take_sample(render_engine, scene, time)
+    def take_sample(self, render_engine, depsgraph, time):
+        if render_engine.test_break():
+            raise ExportCancelled()
+        self.background_shader.take_sample(render_engine, depsgraph, time)
 
-        for light in self.lights:
-            # Check if render is cancelled
-            if render_engine.test_break():
-                raise ExportCancelled()
-            light.take_sample(render_engine, scene, time)
+    def cleanup(self):
+        pass
 
     def export(self, render_engine, w):
-        self.camera.export(render_engine, w)
-
         w.write("World {\n")
         w.indent()
 
         self.background_shader.export(render_engine, w)
 
-        for light in self.lights:
-            light.export(render_engine, w)
-
         w.unindent()
         w.write("}\n")
 
-    def cleanup(self):
-        # For future use.  This is run by the calling code when finished,
-        # even if export did not succeed.
-        pass
-
-#================================================================
 
 class Camera:
     def __init__(self, render_engine, ob, aspect_ratio):
@@ -58,31 +39,43 @@ class Camera:
         self.focal_distances = []
         self.xforms = []
 
-    def take_sample(self, render_engine, scene, time):
+    def take_sample(self, render_engine, depsgraph, time):
         render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.ob.name, time))
 
-        # Fov
-        if self.aspect_ratio >= 1.0:
-            self.fovs += [degrees(self.ob.data.angle)]
+        # Fov.
+        # TODO: account for the various ways sensor size can be specified.
+        x_extent = depsgraph.scene.render.resolution_x / depsgraph.scene.render.pixel_aspect_x
+        y_extent = depsgraph.scene.render.resolution_y / depsgraph.scene.render.pixel_aspect_y
+        aspect_ratio = x_extent / y_extent
+        if aspect_ratio >= 1.0:
+            self.fovs += [degrees(self.ob.data.angle_x)]
         else:
-            self.fovs += [degrees(2.0 * atan(tan(self.ob.data.angle * 0.5) * self.aspect_ratio))]
+            self.fovs += [degrees(2.0 * atan(tan(self.ob.data.angle_x * 0.5) * aspect_ratio))]
 
-        # Aperture radius
-        self.aperture_radii += [self.ob.data.psychopath.aperture_radius]
+        if self.ob.data.dof.use_dof:
+            # Aperture radius.
+            radius = self.ob.data.lens / 2000.0 / self.ob.data.dof.aperture_fstop
+            self.aperture_radii += [radius]
 
-        # Dof distance
-        if self.ob.data.dof_object == None:
-            self.focal_distances += [self.ob.data.dof_distance]
+            # Dof distance
+            if self.ob.data.dof.focus_object == None:
+                self.focal_distances += [self.ob.data.dof.focus_distance]
+            else:
+                # TODO: implement DoF object tracking here
+                self.focal_distances += [0.0]
+                print("WARNING: DoF object tracking not yet implemented.")
         else:
-            # TODO: implement DoF object tracking here
-            self.focal_distances += [0.0]
-            print("WARNING: DoF object tracking not yet implemented.")
+            self.aperture_radii += [0.0]
+            self.focal_distances += [1.0]
 
         # Transform
         mat = self.ob.matrix_world.copy()
         matz = Matrix()
         matz[2][2] = -1
-        self.xforms += [mat * matz]
+        self.xforms += [(mat @ matz).inverted()]
+
+    def cleanup(self):
+        pass
 
     def export(self, render_engine, w):
         render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
@@ -108,55 +101,24 @@ class Camera:
 class BackgroundShader:
     def __init__(self, render_engine, world):
         self.world = world
+        self.color = []
+
+    def take_sample(self, render_engine, depsgraph, time):
         if self.world != None:
-            self.color = (world.horizon_color[0], world.horizon_color[1], world.horizon_color[2])
+            self.color += [(
+                self.world.psychopath.background_color[0],
+                self.world.psychopath.background_color[1],
+                self.world.psychopath.background_color[2],
+            )]
 
     def export(self, render_engine, w):
         if self.world != None:
             w.write("BackgroundShader {\n")
             w.indent();
             w.write("Type [Color]\n")
-            w.write("Color [rec709, %f %f %f]\n" % self.color)
+            for c in self.color:
+                w.write("Color [rec709, %f %f %f]\n" % c)
             w.unindent()
             w.write("}\n")
 
 
-class DistantDiskLamp:
-    def __init__(self, ob, name):
-        self.ob = ob
-        self.name = name
-        self.time_col = []
-        self.time_dir = []
-        self.time_rad = []
-
-    def take_sample(self, render_engine, scene, time):
-        render_engine.update_stats("", "Psychopath: Collecting '{}' at time {}".format(self.ob.name, time))
-        self.time_dir += [tuple(self.ob.matrix_world.to_3x3() * Vector((0, 0, -1)))]
-
-        if self.ob.data.psychopath.color_type == 'Rec709':
-            self.time_col += [('Rec709', self.ob.data.color * self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'Blackbody':
-            self.time_col += [('Blackbody', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-        elif self.ob.data.psychopath.color_type == 'ColorTemperature':
-            self.time_col += [('ColorTemperature', self.ob.data.psychopath.color_blackbody_temp, self.ob.data.energy)]
-
-        self.time_rad += [self.ob.data.shadow_soft_size]
-
-    def export(self, render_engine, w):
-        render_engine.update_stats("", "Psychopath: Exporting %s" % self.ob.name)
-        w.write("DistantDiskLight $%s {\n" % self.name)
-        w.indent()
-        for direc in self.time_dir:
-            w.write("Direction [%f %f %f]\n" % (direc[0], direc[1], direc[2]))
-        for col in self.time_col:
-            if col[0] == 'Rec709':
-                w.write("Color [rec709, %f %f %f]\n" % (col[1][0], col[1][1], col[1][2]))
-            elif col[0] == 'Blackbody':
-                w.write("Color [blackbody, %f %f]\n" % (col[1], col[2]))
-            elif col[0] == 'ColorTemperature':
-                w.write("Color [color_temperature, %f %f]\n" % (col[1], col[2]))
-        for rad in self.time_rad:
-            w.write("Radius [%f]\n" % rad)
-
-        w.unindent()
-        w.write("}\n")

src/accel/bvh.rs

@@ -67,7 +67,91 @@ impl<'a> BVH<'a> {
         self.depth
     }
 
-    pub fn traverse<T, F>(&self, rays: &mut [AccelRay], objects: &[T], mut obj_ray_test: F)
+    pub fn traverse<T, F>(&self, ray: &mut AccelRay, objects: &[T], mut obj_ray_test: F)
+    where
+        F: FnMut(&T, &mut AccelRay),
+    {
+        if self.root.is_none() {
+            return;
+        }
+
+        let mut timer = Timer::new();
+        let mut trav_time: f64 = 0.0;
+        let mut node_tests: u64 = 0;
+
+        let ray_sign = [
+            ray.dir_inv.x() >= 0.0,
+            ray.dir_inv.y() >= 0.0,
+            ray.dir_inv.z() >= 0.0,
+        ];
+
+        // +2 of max depth for root and last child
+        let mut node_stack = [self.root.unwrap(); BVH_MAX_DEPTH + 2];
+        let mut stack_ptr = 1;
+
+        while stack_ptr > 0 && !ray.is_done {
+            node_tests += 1;
+            match *node_stack[stack_ptr] {
+                BVHNode::Internal {
+                    children,
+                    bounds_start,
+                    bounds_len,
+                    split_axis,
+                } => {
+                    let bounds =
+                        unsafe { std::slice::from_raw_parts(bounds_start, bounds_len as usize) };
+                    let is_hit = lerp_slice(bounds, ray.time).intersect_accel_ray(&ray);
+
+                    if is_hit {
+                        if ray_sign[split_axis as usize] {
+                            node_stack[stack_ptr] = children.1;
+                            node_stack[stack_ptr + 1] = children.0;
+                        } else {
+                            node_stack[stack_ptr] = children.0;
+                            node_stack[stack_ptr + 1] = children.1;
+                        }
+                        stack_ptr += 1;
+                    } else {
+                        stack_ptr -= 1;
+                    }
+                }
+
+                BVHNode::Leaf {
+                    object_range,
+                    bounds_start,
+                    bounds_len,
+                } => {
+                    let bounds =
+                        unsafe { std::slice::from_raw_parts(bounds_start, bounds_len as usize) };
+                    let is_hit = lerp_slice(bounds, ray.time).intersect_accel_ray(&ray);
+
+                    trav_time += timer.tick() as f64;
+
+                    if is_hit {
+                        for obj in &objects[object_range.0..object_range.1] {
+                            obj_ray_test(obj, ray);
+                        }
+                    }
+
+                    timer.tick();
+
+                    stack_ptr -= 1;
+                }
+            }
+        }
+
+        trav_time += timer.tick() as f64;
+        ACCEL_TRAV_TIME.with(|att| {
+            let v = att.get();
+            att.set(v + trav_time);
+        });
+        ACCEL_NODE_RAY_TESTS.with(|anv| {
+            let v = anv.get();
+            anv.set(v + node_tests);
+        });
+    }
+
+    pub fn traverse_multi<T, F>(&self, rays: &mut [AccelRay], objects: &[T], mut obj_ray_test: F)
     where
         F: FnMut(&T, &mut [AccelRay]),
     {

src/accel/bvh4.rs

@@ -1,13 +1,7 @@
-//! This BVH4 implementation is based on the ideas from the paper
-//! "Efficient Ray Tracing Kernels for Modern CPU Architectures"
-//! by Fuetterling et al.
-
 #![allow(dead_code)]
 
 use std::mem::{transmute, MaybeUninit};
 
-use glam::BVec4A;
-
 use kioku::Arena;
 
 use crate::{
@@ -16,7 +10,7 @@ use crate::{
     boundable::Boundable,
     lerp::lerp_slice,
     math::Vector,
-    ray::{RayBatch, RayStack},
+    ray::{LocalRay, Ray},
 };
 
 use super::{
@@ -25,6 +19,7 @@ use super::{
 };
 
 use bvh_order::{calc_traversal_code, SplitAxes, TRAVERSAL_TABLE};
+use rmath::wide4::Float4;
 
 pub fn ray_code(dir: Vector) -> usize {
     let ray_sign_is_neg = [dir.x() < 0.0, dir.y() < 0.0, dir.z() < 0.0];
@@ -33,6 +28,8 @@ pub fn ray_code(dir: Vector) -> usize {
         + ((ray_sign_is_neg[2] as usize) << 2)
 }
 
+//-------------------------------------------------------------
+
 #[derive(Copy, Clone, Debug)]
 pub struct BVH4<'a> {
     root: Option<&'a BVH4Node<'a>>,
@@ -98,9 +95,9 @@ impl<'a> BVH4<'a> {
         self.depth
     }
 
-    pub fn traverse<F>(&self, rays: &mut RayBatch, ray_stack: &mut RayStack, mut obj_ray_test: F)
+    pub fn traverse<F>(&self, ray: &mut Ray, local_ray: &LocalRay, mut obj_ray_test: F)
     where
-        F: FnMut(std::ops::Range<usize>, &mut RayBatch, &mut RayStack),
+        F: FnMut(std::ops::Range<usize>, &mut Ray),
     {
         if self.root.is_none() {
             return;
@@ -108,55 +105,48 @@ impl<'a> BVH4<'a> {
 
         let mut node_tests: u64 = 0;
 
-        let traversal_table =
-            &TRAVERSAL_TABLE[ray_code(rays.dir_inv_local(ray_stack.next_task_ray_idx(0)))];
+        // SIMD-ready ray data.
+        let orig4 = [
+            local_ray.orig.0.aaaa(),
+            local_ray.orig.0.bbbb(),
+            local_ray.orig.0.cccc(),
+        ];
+        let dir_inv4 = [
+            local_ray.dir_inv.0.aaaa(),
+            local_ray.dir_inv.0.bbbb(),
+            local_ray.dir_inv.0.cccc(),
+        ];
+        let mut max_t4 = Float4::splat(ray.max_t);
 
         // +2 of max depth for root and last child
         let mut node_stack = [self.root.unwrap(); (BVH_MAX_DEPTH * 3) + 2];
         let mut stack_ptr = 1;
 
-        while stack_ptr > 0 {
+        let traversal_table = &TRAVERSAL_TABLE[ray_code(local_ray.dir_inv)];
+
+        while stack_ptr > 0 && !ray.is_done() {
             match *node_stack[stack_ptr] {
                 BVH4Node::Internal {
                     bounds,
                     children,
                     traversal_code,
                 } => {
-                    node_tests += ray_stack.ray_count_in_next_task() as u64;
-                    let mut all_hits = BVec4A::default();
+                    node_tests += 1;
 
-                    // Ray testing
-                    ray_stack.pop_do_next_task_and_push_rays(children.len(), |ray_idx| {
-                        if rays.is_done(ray_idx) {
-                            BVec4A::default()
-                        } else {
-                            let hits = if bounds.len() == 1 {
-                                bounds[0].intersect_ray(
-                                    rays.orig_local(ray_idx),
-                                    rays.dir_inv_local(ray_idx),
-                                    rays.max_t(ray_idx),
-                                )
-                            } else {
-                                lerp_slice(bounds, rays.time(ray_idx)).intersect_ray(
-                                    rays.orig_local(ray_idx),
-                                    rays.dir_inv_local(ray_idx),
-                                    rays.max_t(ray_idx),
-                                )
-                            };
-                            all_hits |= hits;
-                            hits
-                        }
-                    });
+                    let hits = if bounds.len() == 1 {
+                        bounds[0].intersect_ray(orig4, dir_inv4, max_t4)
+                    } else {
+                        lerp_slice(bounds, ray.time).intersect_ray(orig4, dir_inv4, max_t4)
+                    };
 
-                    // If there were any intersections, create tasks.
-                    if all_hits.any() {
+                    // Push child nodes onto the stack if there were any hits.
+                    if hits.any() {
                         let order_code = traversal_table[traversal_code as usize];
+                        let hits = hits.to_bools();
                         let mut lane_count = 0;
-                        let mut i = children.len() as u8;
-                        while i > 0 {
-                            i -= 1;
+                        for i in (0..children.len() as u8).rev() {
                             let child_i = ((order_code >> (i * 2)) & 3) as usize;
-                            if ray_stack.push_lane_to_task(child_i) {
+                            if hits[child_i] {
                                 node_stack[stack_ptr + lane_count] = &children[child_i];
                                 lane_count += 1;
                             }
@@ -169,8 +159,10 @@ impl<'a> BVH4<'a> {
                 }
 
                 BVH4Node::Leaf { object_range } => {
-                    // Do the ray tests.
-                    obj_ray_test(object_range.0..object_range.1, rays, ray_stack);
+                    obj_ray_test(object_range.0..object_range.1, ray);
+
+                    // Update SIMD max_t in case there was a hit.
+                    max_t4 = Float4::splat(ray.max_t);
 
                     stack_ptr -= 1;
                 }

src/algorithm.rs

@@ -5,10 +5,9 @@ use std::{
     mem::MaybeUninit,
 };
 
-use crate::{
-    hash::hash_u64,
-    lerp::{lerp_slice, Lerp},
-};
+use rrand::mix64_seed;
+
+use crate::lerp::{lerp_slice, Lerp};
 
 /// Selects an item from a slice based on a weighting function and a
 /// number (n) between 0.0 and 1.0.  Returns the index of the selected
@@ -209,7 +208,7 @@ where
     let mut seed = n as u64;
 
     loop {
-        let i = left + (hash_u64(right as u64, seed) as usize % (right - left));
+        let i = left + (mix64_seed(right as u64, seed) as usize % (right - left));
 
         slc.swap(i, right - 1);
         let ii = left + {

src/bbox.rs

@@ -7,7 +7,7 @@ use std::{
 
 use crate::{
     lerp::{lerp, lerp_slice, Lerp},
-    math::{Point, Transform, Vector},
+    math::{fast_minf32, Point, Vector, Xform},
 };
 
 const BBOX_MAXT_ADJUST: f32 = 1.000_000_24;
@@ -41,21 +41,23 @@ impl BBox {
     // Returns whether the given ray intersects with the bbox.
     pub fn intersect_ray(&self, orig: Point, dir_inv: Vector, max_t: f32) -> bool {
         // Calculate slab intersections
-        let t1 = (self.min.co - orig.co) * dir_inv.co;
-        let t2 = (self.max.co - orig.co) * dir_inv.co;
+        let t1 = (self.min.0 - orig.0) * dir_inv.0;
+        let t2 = (self.max.0 - orig.0) * dir_inv.0;
 
         // Find the far and near intersection
-        let far_t = t1.max(t2).extend(std::f32::INFINITY);
-        let near_t = t1.min(t2).extend(0.0);
-        let far_hit_t = (far_t.min_element() * BBOX_MAXT_ADJUST).min(max_t);
+        let far_t = t1.max(t2).set_d(std::f32::INFINITY);
+        let near_t = t1.min(t2).set_d(0.0);
+        let far_hit_t = fast_minf32(far_t.min_element() * BBOX_MAXT_ADJUST, max_t);
         let near_hit_t = near_t.max_element();
 
         // Did we hit?
         near_hit_t <= far_hit_t
     }
 
-    // Creates a new BBox transformed into a different space.
-    pub fn transformed(&self, xform: Transform) -> BBox {
+    // Creates a new BBox transformed from its local space to the
+    // given space.
+    #[must_use]
+    pub fn xform(&self, xform: &Xform) -> BBox {
         // BBox corners
         let vs = [
             Point::new(self.min.x(), self.min.y(), self.min.z()),
@@ -71,7 +73,7 @@ impl BBox {
         // Transform BBox corners and make new bbox
         let mut b = BBox::new();
         for v in &vs {
-            let v = *v * xform;
+            let v = v.xform(xform);
             b.min = v.min(b.min);
             b.max = v.max(b.max);
         }
@@ -103,12 +105,8 @@ impl BitOr for BBox {
 
     fn bitor(self, rhs: BBox) -> BBox {
         BBox::from_points(
-            Point {
-                co: self.min.co.min(rhs.min.co),
-            },
-            Point {
-                co: self.max.co.max(rhs.max.co),
-            },
+            Point(self.min.0.min(rhs.min.0)),
+            Point(self.max.0.max(rhs.max.0)),
         )
     }
 }
@@ -124,14 +122,7 @@ impl BitOr<Point> for BBox {
     type Output = BBox;
 
     fn bitor(self, rhs: Point) -> BBox {
-        BBox::from_points(
-            Point {
-                co: self.min.co.min(rhs.co),
-            },
-            Point {
-                co: self.max.co.max(rhs.co),
-            },
-        )
+        BBox::from_points(Point(self.min.0.min(rhs.0)), Point(self.max.0.max(rhs.0)))
     }
 }
 
@@ -150,7 +141,7 @@ impl Lerp for BBox {
     }
 }
 
-pub fn transform_bbox_slice_from(bbs_in: &[BBox], xforms: &[Transform], bbs_out: &mut Vec<BBox>) {
+pub fn transform_bbox_slice_from(bbs_in: &[BBox], xforms: &[Xform], bbs_out: &mut Vec<BBox>) {
     bbs_out.clear();
 
     // Transform the bounding boxes
@@ -158,17 +149,17 @@ pub fn transform_bbox_slice_from(bbs_in: &[BBox], xforms: &[Transform], bbs_out:
         bbs_out.extend_from_slice(bbs_in);
     } else if bbs_in.len() == xforms.len() {
         for (bb, xf) in Iterator::zip(bbs_in.iter(), xforms.iter()) {
-            bbs_out.push(bb.transformed(xf.inverse()));
+            bbs_out.push(bb.xform(&xf));
         }
     } else if bbs_in.len() > xforms.len() {
         let s = (bbs_in.len() - 1) as f32;
         for (i, bb) in bbs_in.iter().enumerate() {
-            bbs_out.push(bb.transformed(lerp_slice(xforms, i as f32 / s).inverse()));
+            bbs_out.push(bb.xform(&lerp_slice(xforms, i as f32 / s)));
         }
     } else if bbs_in.len() < xforms.len() {
         let s = (xforms.len() - 1) as f32;
         for (i, xf) in xforms.iter().enumerate() {
-            bbs_out.push(lerp_slice(bbs_in, i as f32 / s).transformed(xf.inverse()));
+            bbs_out.push(lerp_slice(bbs_in, i as f32 / s).xform(&xf));
         }
     }
 }

src/bbox4.rs

@@ -6,19 +6,18 @@ use std::ops::{BitOr, BitOrAssign};
 use crate::{
     bbox::BBox,
     lerp::{lerp, Lerp},
-    math::{Point, Vector},
 };
 
-use glam::{BVec4A, Vec4};
+use rmath::wide4::{Bool4, Float4};
 
 const BBOX_MAXT_ADJUST: f32 = 1.000_000_24;
 
 /// A SIMD set of 4 3D axis-aligned bounding boxes.
 #[derive(Debug, Copy, Clone)]
 pub struct BBox4 {
-    pub x: (Vec4, Vec4), // (min, max)
-    pub y: (Vec4, Vec4), // (min, max)
-    pub z: (Vec4, Vec4), // (min, max)
+    pub x: (Float4, Float4), // (min, max)
+    pub y: (Float4, Float4), // (min, max)
+    pub z: (Float4, Float4), // (min, max)
 }
 
 impl BBox4 {
@@ -26,16 +25,16 @@ impl BBox4 {
     pub fn new() -> BBox4 {
         BBox4 {
             x: (
-                Vec4::splat(std::f32::INFINITY),
-                Vec4::splat(std::f32::NEG_INFINITY),
+                Float4::splat(std::f32::INFINITY),
+                Float4::splat(std::f32::NEG_INFINITY),
             ),
             y: (
-                Vec4::splat(std::f32::INFINITY),
-                Vec4::splat(std::f32::NEG_INFINITY),
+                Float4::splat(std::f32::INFINITY),
+                Float4::splat(std::f32::NEG_INFINITY),
             ),
             z: (
-                Vec4::splat(std::f32::INFINITY),
-                Vec4::splat(std::f32::NEG_INFINITY),
+                Float4::splat(std::f32::INFINITY),
+                Float4::splat(std::f32::NEG_INFINITY),
             ),
         }
     }
@@ -45,38 +44,30 @@ impl BBox4 {
     pub fn from_bboxes(b1: BBox, b2: BBox, b3: BBox, b4: BBox) -> BBox4 {
         BBox4 {
             x: (
-                Vec4::new(b1.min.x(), b2.min.x(), b3.min.x(), b4.min.x()),
-                Vec4::new(b1.max.x(), b2.max.x(), b3.max.x(), b4.max.x()),
+                Float4::new(b1.min.x(), b2.min.x(), b3.min.x(), b4.min.x()),
+                Float4::new(b1.max.x(), b2.max.x(), b3.max.x(), b4.max.x()),
             ),
             y: (
-                Vec4::new(b1.min.y(), b2.min.y(), b3.min.y(), b4.min.y()),
-                Vec4::new(b1.max.y(), b2.max.y(), b3.max.y(), b4.max.y()),
+                Float4::new(b1.min.y(), b2.min.y(), b3.min.y(), b4.min.y()),
+                Float4::new(b1.max.y(), b2.max.y(), b3.max.y(), b4.max.y()),
             ),
             z: (
-                Vec4::new(b1.min.z(), b2.min.z(), b3.min.z(), b4.min.z()),
-                Vec4::new(b1.max.z(), b2.max.z(), b3.max.z(), b4.max.z()),
+                Float4::new(b1.min.z(), b2.min.z(), b3.min.z(), b4.min.z()),
+                Float4::new(b1.max.z(), b2.max.z(), b3.max.z(), b4.max.z()),
             ),
         }
     }
 
     // Returns whether the given ray intersects with the bboxes.
-    pub fn intersect_ray(&self, orig: Point, dir_inv: Vector, max_t: f32) -> BVec4A {
-        // Get the ray data into SIMD format.
-        let ro_x = Vec4::splat(orig.co[0]);
-        let ro_y = Vec4::splat(orig.co[1]);
-        let ro_z = Vec4::splat(orig.co[2]);
-        let rdi_x = Vec4::splat(dir_inv.co[0]);
-        let rdi_y = Vec4::splat(dir_inv.co[1]);
-        let rdi_z = Vec4::splat(dir_inv.co[2]);
-        let max_t = Vec4::splat(max_t);
-
+    #[inline(always)]
+    pub fn intersect_ray(&self, orig: [Float4; 3], dir_inv: [Float4; 3], max_t: Float4) -> Bool4 {
         // Slab tests
-        let t1_x = (self.x.0 - ro_x) * rdi_x;
-        let t1_y = (self.y.0 - ro_y) * rdi_y;
-        let t1_z = (self.z.0 - ro_z) * rdi_z;
-        let t2_x = (self.x.1 - ro_x) * rdi_x;
-        let t2_y = (self.y.1 - ro_y) * rdi_y;
-        let t2_z = (self.z.1 - ro_z) * rdi_z;
+        let t1_x = (self.x.0 - orig[0]) * dir_inv[0];
+        let t1_y = (self.y.0 - orig[1]) * dir_inv[1];
+        let t1_z = (self.z.0 - orig[2]) * dir_inv[2];
+        let t2_x = (self.x.1 - orig[0]) * dir_inv[0];
+        let t2_y = (self.y.1 - orig[1]) * dir_inv[1];
+        let t2_z = (self.z.1 - orig[2]) * dir_inv[2];
 
         // Get the far and near t hits for each axis.
         let t_far_x = t1_x.max(t2_x);
@@ -87,10 +78,11 @@ impl BBox4 {
         let t_near_z = t1_z.min(t2_z);
 
         // Calculate over-all far t hit.
-        let far_t = (t_far_x.min(t_far_y.min(t_far_z)) * Vec4::splat(BBOX_MAXT_ADJUST)).min(max_t);
+        let far_t =
+            (t_far_x.min(t_far_y.min(t_far_z)) * Float4::splat(BBOX_MAXT_ADJUST)).min(max_t);
 
         // Calculate over-all near t hit.
-        let near_t = t_near_x.max(t_near_y).max(t_near_z.max(Vec4::splat(0.0)));
+        let near_t = t_near_x.max(t_near_y).max(t_near_z.max(Float4::splat(0.0)));
 
         // Hit results
         near_t.cmplt(far_t)

src/camera.rs

@@ -1,31 +1,28 @@
 #![allow(dead_code)]
 
-use kioku::Arena;
-
 use crate::{
     lerp::lerp_slice,
-    math::{Point, Transform, Vector},
+    math::{Point, Vector, Xform},
     ray::Ray,
     sampling::square_to_circle,
 };
 
-#[derive(Copy, Clone, Debug)]
-pub struct Camera<'a> {
-    transforms: &'a [Transform],
-    fovs: &'a [f32],
-    tfovs: &'a [f32],
-    aperture_radii: &'a [f32],
-    focus_distances: &'a [f32],
+#[derive(Debug, Clone)]
+pub struct Camera {
+    transforms: Vec<Xform>,
+    fovs: Vec<f32>,
+    tfovs: Vec<f32>,
+    aperture_radii: Vec<f32>,
+    focus_distances: Vec<f32>,
 }
 
-impl<'a> Camera<'a> {
+impl Camera {
     pub fn new(
-        arena: &'a Arena,
-        transforms: &[Transform],
+        transforms: &[Xform],
         fovs: &[f32],
         mut aperture_radii: &[f32],
         mut focus_distances: &[f32],
-    ) -> Camera<'a> {
+    ) -> Camera {
         assert!(!transforms.is_empty(), "Camera has no transform(s)!");
         assert!(!fovs.is_empty(), "Camera has no fov(s)!");
 
@@ -63,20 +60,20 @@ impl<'a> Camera<'a> {
             .collect();
 
         Camera {
-            transforms: arena.copy_slice(&transforms),
-            fovs: arena.copy_slice(&fovs),
-            tfovs: arena.copy_slice(&tfovs),
-            aperture_radii: arena.copy_slice(&aperture_radii),
-            focus_distances: arena.copy_slice(&focus_distances),
+            transforms: transforms.into(),
+            fovs: fovs.into(),
+            tfovs: tfovs.into(),
+            aperture_radii: aperture_radii.into(),
+            focus_distances: focus_distances.into(),
         }
     }
 
     pub fn generate_ray(&self, x: f32, y: f32, time: f32, wavelength: f32, u: f32, v: f32) -> Ray {
         // Get time-interpolated camera settings
-        let transform = lerp_slice(self.transforms, time);
-        let tfov = lerp_slice(self.tfovs, time);
-        let aperture_radius = lerp_slice(self.aperture_radii, time);
-        let focus_distance = lerp_slice(self.focus_distances, time);
+        let transform = lerp_slice(&self.transforms, time).to_full_fast().unwrap();
+        let tfov = lerp_slice(&self.tfovs, time);
+        let aperture_radius = lerp_slice(&self.aperture_radii, time);
+        let focus_distance = lerp_slice(&self.focus_distances, time);
 
         // Ray origin
         let orig = {
@@ -92,12 +89,13 @@ impl<'a> Camera<'a> {
         )
         .normalized();
 
-        Ray {
-            orig: orig * transform,
-            dir: dir * transform,
-            time: time,
-            wavelength: wavelength,
-            max_t: std::f32::INFINITY,
-        }
+        Ray::new(
+            orig.xform_inv_fast(&transform),
+            dir.xform_inv_fast(&transform),
+            time,
+            wavelength,
+            std::f32::INFINITY,
+            false,
+        )
     }
 }

src/color.rs

@@ -1,13 +1,8 @@
 use std::ops::{Add, AddAssign, Div, DivAssign, Mul, MulAssign};
 
-pub use color::{
-    rec709_e_to_xyz, rec709_to_xyz, xyz_to_aces_ap0, xyz_to_aces_ap0_e, xyz_to_rec709,
-    xyz_to_rec709_e,
-};
+use crate::math::Float4;
 use compact::fluv::fluv32;
-use glam::Vec4;
 use half::f16;
-use spectral_upsampling::meng::{spectrum_xyz_to_p_4, EQUAL_ENERGY_REFLECTANCE};
 
 use crate::{lerp::Lerp, math::fast_exp};
 
@@ -31,10 +26,10 @@ fn nth_wavelength(hero_wavelength: f32, n: usize) -> f32 {
     }
 }
 
-/// Returns all wavelengths of a hero wavelength set as a Vec4
+/// Returns all wavelengths of a hero wavelength set as a Float4
 #[inline(always)]
-fn wavelengths(hero_wavelength: f32) -> Vec4 {
-    Vec4::new(
+fn wavelengths(hero_wavelength: f32) -> Float4 {
+    Float4::new(
         nth_wavelength(hero_wavelength, 0),
         nth_wavelength(hero_wavelength, 1),
         nth_wavelength(hero_wavelength, 2),
@@ -94,7 +89,7 @@ impl Color {
             } => {
                 SpectralSample::from_parts(
                     // TODO: make this SIMD
-                    Vec4::new(
+                    Float4::new(
                         plancks_law(temperature, wls[0]) * factor,
                         plancks_law(temperature, wls[1]) * factor,
                         plancks_law(temperature, wls[2]) * factor,
@@ -109,7 +104,7 @@ impl Color {
             } => {
                 SpectralSample::from_parts(
                     // TODO: make this SIMD
-                    Vec4::new(
+                    Float4::new(
                         plancks_law_normalized(temperature, wls[0]) * factor,
                         plancks_law_normalized(temperature, wls[1]) * factor,
                         plancks_law_normalized(temperature, wls[2]) * factor,
@@ -386,7 +381,7 @@ fn plancks_law_normalized(temperature: f32, wavelength: f32) -> f32 {
 
 #[derive(Copy, Clone, Debug)]
 pub struct SpectralSample {
-    pub e: Vec4,
+    pub e: Float4,
     hero_wavelength: f32,
 }
 
@@ -394,7 +389,7 @@ impl SpectralSample {
     pub fn new(wavelength: f32) -> SpectralSample {
         debug_assert!(wavelength >= WL_MIN && wavelength <= WL_MAX);
         SpectralSample {
-            e: Vec4::splat(0.0),
+            e: Float4::splat(0.0),
             hero_wavelength: wavelength,
         }
     }
@@ -403,12 +398,12 @@ impl SpectralSample {
     pub fn from_value(value: f32, wavelength: f32) -> SpectralSample {
         debug_assert!(wavelength >= WL_MIN && wavelength <= WL_MAX);
         SpectralSample {
-            e: Vec4::splat(value),
+            e: Float4::splat(value),
             hero_wavelength: wavelength,
         }
     }
 
-    pub fn from_parts(e: Vec4, wavelength: f32) -> SpectralSample {
+    pub fn from_parts(e: Float4, wavelength: f32) -> SpectralSample {
         debug_assert!(wavelength >= WL_MIN && wavelength <= WL_MAX);
         SpectralSample {
             e: e,
@@ -599,30 +594,61 @@ impl DivAssign<f32> for XYZ {
 /// the method in the paper "Physically Meaningful Rendering using Tristimulus
 /// Colours" by Meng et al.
 #[inline(always)]
-fn xyz_to_spectrum_4(xyz: (f32, f32, f32), wavelengths: Vec4) -> Vec4 {
-    spectrum_xyz_to_p_4(wavelengths, xyz) * Vec4::splat(1.0 / EQUAL_ENERGY_REFLECTANCE)
-    // aces_to_spectrum_p4(wavelengths, xyz_to_aces_ap0_e(xyz))
+fn xyz_to_spectrum_4(xyz: (f32, f32, f32), wavelengths: Float4) -> Float4 {
+    use spectral_upsampling as su;
+
+    // su::meng::spectrum_xyz_to_p_4(wavelengths, xyz)
+    //     * Float4::splat(1.0 / su::meng::EQUAL_ENERGY_REFLECTANCE)
+
+    su::jakob::rec2020_to_spectrum_p4(wavelengths, color::xyz_to_rec2020_e(xyz))
+
+    // su::jakob::rec709_to_spectrum_p4(wavelengths, color::xyz_to_rec709_e(xyz))
 }
 
 /// Close analytic approximations of the CIE 1931 XYZ color curves.
 /// From the paper "Simple Analytic Approximations to the CIE XYZ Color Matching
 /// Functions" by Wyman et al.
 pub fn x_1931(wavelength: f32) -> f32 {
-    let t1 = (wavelength - 442.0) * (if wavelength < 442.0 { 0.0624 } else { 0.0374 });
-    let t2 = (wavelength - 599.8) * (if wavelength < 599.8 { 0.0264 } else { 0.0323 });
-    let t3 = (wavelength - 501.1) * (if wavelength < 501.1 { 0.0490 } else { 0.0382 });
-    (0.362 * fast_exp(-0.5 * t1 * t1)) + (1.056 * fast_exp(-0.5 * t2 * t2))
-        - (0.065 * fast_exp(-0.5 * t3 * t3))
+    use colorbox::tables::cie_1931_xyz::{MAX_WAVELENGTH, MIN_WAVELENGTH, X};
+
+    let norm = 1.0 / (MAX_WAVELENGTH - MIN_WAVELENGTH);
+    let n = (wavelength - MIN_WAVELENGTH) * norm;
+
+    if n < 0.0 {
+        X[0]
+    } else if n > 1.0 {
+        *X.last().unwrap()
+    } else {
+        crate::lerp::lerp_slice(X, n)
+    }
 }
 
 pub fn y_1931(wavelength: f32) -> f32 {
-    let t1 = (wavelength - 568.8) * (if wavelength < 568.8 { 0.0213 } else { 0.0247 });
-    let t2 = (wavelength - 530.9) * (if wavelength < 530.9 { 0.0613 } else { 0.0322 });
-    (0.821 * fast_exp(-0.5 * t1 * t1)) + (0.286 * fast_exp(-0.5 * t2 * t2))
+    use colorbox::tables::cie_1931_xyz::{MAX_WAVELENGTH, MIN_WAVELENGTH, Y};
+
+    let norm = 1.0 / (MAX_WAVELENGTH - MIN_WAVELENGTH);
+    let n = (wavelength - MIN_WAVELENGTH) * norm;
+
+    if n < 0.0 {
+        Y[0]
+    } else if n > 1.0 {
+        *Y.last().unwrap()
+    } else {
+        crate::lerp::lerp_slice(Y, n)
+    }
 }
 
 pub fn z_1931(wavelength: f32) -> f32 {
-    let t1 = (wavelength - 437.0) * (if wavelength < 437.0 { 0.0845 } else { 0.0278 });
-    let t2 = (wavelength - 459.0) * (if wavelength < 459.0 { 0.0385 } else { 0.0725 });
-    (1.217 * fast_exp(-0.5 * t1 * t1)) + (0.681 * fast_exp(-0.5 * t2 * t2))
+    use colorbox::tables::cie_1931_xyz::{MAX_WAVELENGTH, MIN_WAVELENGTH, Z};
+
+    let norm = 1.0 / (MAX_WAVELENGTH - MIN_WAVELENGTH);
+    let n = (wavelength - MIN_WAVELENGTH) * norm;
+
+    if n < 0.0 {
+        Z[0]
+    } else if n > 1.0 {
+        *Z.last().unwrap()
+    } else {
+        crate::lerp::lerp_slice(Z, n)
+    }
 }

src/fp_utils.rs

@@ -4,6 +4,7 @@
 //! From Theory to Implementation" 3rd edition by Pharr et al.
 
 use crate::math::{dot, Normal, Point, Vector};
+pub use rmath::utils::{decrement_ulp, increment_ulp};
 
 #[inline(always)]
 pub fn fp_gamma(n: u32) -> f32 {
@@ -12,36 +13,6 @@ pub fn fp_gamma(n: u32) -> f32 {
     (e * n as f32) / (1.0 - (e * n as f32))
 }
 
-pub fn increment_ulp(v: f32) -> f32 {
-    if v.is_finite() {
-        if v > 0.0 {
-            f32::from_bits(v.to_bits() + 1)
-        } else if v < -0.0 {
-            f32::from_bits(v.to_bits() - 1)
-        } else {
-            f32::from_bits(0x00_00_00_01)
-        }
-    } else {
-        // Infinity or NaN.
-        v
-    }
-}
-
-pub fn decrement_ulp(v: f32) -> f32 {
-    if v.is_finite() {
-        if v > 0.0 {
-            f32::from_bits(v.to_bits() - 1)
-        } else if v < -0.0 {
-            f32::from_bits(v.to_bits() + 1)
-        } else {
-            f32::from_bits(0x80_00_00_01)
-        }
-    } else {
-        // Infinity or NaN.
-        v
-    }
-}
-
 pub fn robust_ray_origin(pos: Point, pos_err: f32, nor: Normal, ray_dir: Vector) -> Point {
     // Get surface normal pointing in the same
     // direction as ray_dir.
@@ -81,51 +52,7 @@ pub fn robust_ray_origin(pos: Point, pos_err: f32, nor: Normal, ray_dir: Vector)
     Point::new(x, y, z)
 }
 
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn inc_ulp() {
-        assert!(increment_ulp(1.0) > 1.0);
-        assert!(increment_ulp(-1.0) > -1.0);
-    }
-
-    #[test]
-    fn dec_ulp() {
-        assert!(decrement_ulp(1.0) < 1.0);
-        assert!(decrement_ulp(-1.0) < -1.0);
-    }
-
-    #[test]
-    fn inc_ulp_zero() {
-        assert!(increment_ulp(0.0) > 0.0);
-        assert!(increment_ulp(0.0) > -0.0);
-        assert!(increment_ulp(-0.0) > 0.0);
-        assert!(increment_ulp(-0.0) > -0.0);
-    }
-
-    #[test]
-    fn dec_ulp_zero() {
-        assert!(decrement_ulp(0.0) < 0.0);
-        assert!(decrement_ulp(0.0) < -0.0);
-        assert!(decrement_ulp(-0.0) < 0.0);
-        assert!(decrement_ulp(-0.0) < -0.0);
-    }
-
-    #[test]
-    fn inc_dec_ulp() {
-        assert_eq!(decrement_ulp(increment_ulp(1.0)), 1.0);
-        assert_eq!(decrement_ulp(increment_ulp(-1.0)), -1.0);
-        assert_eq!(decrement_ulp(increment_ulp(1.2)), 1.2);
-        assert_eq!(decrement_ulp(increment_ulp(-1.2)), -1.2);
-    }
-
-    #[test]
-    fn dec_inc_ulp() {
-        assert_eq!(increment_ulp(decrement_ulp(1.0)), 1.0);
-        assert_eq!(increment_ulp(decrement_ulp(-1.0)), -1.0);
-        assert_eq!(increment_ulp(decrement_ulp(1.2)), 1.2);
-        assert_eq!(increment_ulp(decrement_ulp(-1.2)), -1.2);
-    }
-}
+// #[cfg(test)]
+// mod tests {
+//     use super::*;
+// }

src/hash.rs

@@ -1,29 +0,0 @@
-pub fn hash_u32(n: u32, seed: u32) -> u32 {
-    let mut hash = n;
-    for _ in 0..3 {
-        hash = hash.wrapping_mul(0x736caf6f);
-        hash ^= hash.wrapping_shr(16);
-        hash ^= seed;
-    }
-
-    hash
-}
-
-pub fn hash_u64(n: u64, seed: u64) -> u64 {
-    let mut hash = n;
-    for _ in 0..4 {
-        hash = hash.wrapping_mul(32_416_190_071 * 314_604_959);
-        hash ^= hash.wrapping_shr(32);
-        hash ^= seed;
-    }
-
-    hash
-}
-
-/// Returns a random float in [0, 1] based on 'n' and a seed.
-/// Generally use n for getting a bunch of different random
-/// numbers, and use seed to vary between runs.
-pub fn hash_u32_to_f32(n: u32, seed: u32) -> f32 {
-    const INV_MAX: f32 = 1.0 / std::u32::MAX as f32;
-    hash_u32(n, seed) as f32 * INV_MAX
-}

src/hilbert.rs

@@ -1,77 +0,0 @@
-#![allow(dead_code)]
-
-const N: u32 = 1 << 16;
-
-// Utility function used by the functions below.
-fn hil_rot(n: u32, rx: u32, ry: u32, x: &mut u32, y: &mut u32) {
-    use std::mem;
-    if ry == 0 {
-        if rx == 1 {
-            *x = (n - 1).wrapping_sub(*x);
-            *y = (n - 1).wrapping_sub(*y);
-        }
-        mem::swap(x, y);
-    }
-}
-
-/// Convert (x,y) to hilbert curve index.
-///
-/// x: The x coordinate.  Must be a positive integer no greater than 2^16-1.
-/// y: The y coordinate.  Must be a positive integer no greater than 2^16-1.
-///
-/// Returns the hilbert curve index corresponding to the (x,y) coordinates given.
-pub fn xy2d(x: u32, y: u32) -> u32 {
-    assert!(x < N);
-    assert!(y < N);
-
-    let (mut x, mut y) = (x, y);
-    let mut d = 0;
-    let mut s = N >> 1;
-    while s > 0 {
-        let rx = if (x & s) > 0 { 1 } else { 0 };
-        let ry = if (y & s) > 0 { 1 } else { 0 };
-        d += s * s * ((3 * rx) ^ ry);
-        hil_rot(s, rx, ry, &mut x, &mut y);
-
-        s >>= 1
-    }
-
-    d
-}
-
-/// Convert hilbert curve index to (x,y).
-///
-/// d: The hilbert curve index.
-///
-/// Returns the (x, y) coords at the given index.
-pub fn d2xy(d: u32) -> (u32, u32) {
-    let (mut x, mut y) = (0, 0);
-    let mut s = 1;
-    let mut t = d;
-    while s < N {
-        let rx = 1 & (t >> 1);
-        let ry = 1 & (t ^ rx);
-        hil_rot(s, rx, ry, &mut x, &mut y);
-        x += s * rx;
-        y += s * ry;
-        t >>= 2;
-
-        s <<= 1;
-    }
-
-    (x, y)
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn reversible() {
-        let d = 54;
-        let (x, y) = d2xy(d);
-        let d2 = xy2d(x, y);
-
-        assert_eq!(d, d2);
-    }
-}

src/image.rs

@@ -13,7 +13,9 @@ use std::{
 
 use half::f16;
 
-use crate::color::{xyz_to_rec709_e, XYZ};
+pub use color::{rec709_e_to_xyz, xyz_to_rec709_e};
+
+use crate::color::XYZ;
 
 #[derive(Debug)]
 #[allow(clippy::type_complexity)]
@@ -236,22 +238,35 @@ impl<'a> Bucket<'a> {
     where
         F: Fn((f32, f32, f32)) -> (f32, f32, f32),
     {
-        use std::slice;
-        let mut data = Vec::with_capacity(
-            (4 * (self.max.0 - self.min.0) * (self.max.1 - self.min.1)) as usize,
+        let data_u8 = self.rgba_raw(color_convert);
+        base64::encode(&data_u8)
+    }
+
+    /// Returns the bucket's contents as a binary string.
+    ///
+    /// The data is laid out as four-floats-per-pixel in scanline order.
+    /// The fourth channel is alpha, and is set to 1.0 for all pixels.
+    ///
+    /// `color_convert` lets you do a colorspace conversion before base64
+    /// encoding if desired.
+    pub fn rgba_raw<F>(&mut self, color_convert: F) -> Vec<u8>
+    where
+        F: Fn((f32, f32, f32)) -> (f32, f32, f32),
+    {
+        let mut data: Vec<u8> = Vec::with_capacity(
+            std::mem::size_of::<f32>()
+                * (4 * (self.max.0 - self.min.0) * (self.max.1 - self.min.1)) as usize,
         );
         for y in self.min.1..self.max.1 {
             for x in self.min.0..self.max.0 {
                 let color = color_convert(self.get(x, y).to_tuple());
-                data.push(color.0);
-                data.push(color.1);
-                data.push(color.2);
-                data.push(1.0);
+                data.extend_from_slice(&color.0.to_ne_bytes());
+                data.extend_from_slice(&color.1.to_ne_bytes());
+                data.extend_from_slice(&color.2.to_ne_bytes());
+                data.extend_from_slice(&1.0f32.to_ne_bytes());
             }
         }
-        let data_u8 =
-            unsafe { slice::from_raw_parts(&data[0] as *const f32 as *const u8, data.len() * 4) };
-        base64::encode(data_u8)
+        data
     }
 }
 

src/lerp.rs

@@ -1,6 +1,6 @@
 #![allow(dead_code)]
 
-use math3d::{Normal, Point, Transform, Vector};
+use rmath::{wide4::Float4, Normal, Point, Vector, Xform};
 
 /// Trait for allowing a type to be linearly interpolated.
 pub trait Lerp: Copy {
@@ -100,36 +100,34 @@ impl<T: Lerp> Lerp for [T; 4] {
     }
 }
 
-impl Lerp for glam::Vec4 {
-    fn lerp(self, other: glam::Vec4, alpha: f32) -> glam::Vec4 {
+impl Lerp for Float4 {
+    fn lerp(self, other: Self, alpha: f32) -> Self {
         (self * (1.0 - alpha)) + (other * alpha)
     }
 }
 
-impl Lerp for Transform {
-    fn lerp(self, other: Transform, alpha: f32) -> Transform {
+impl Lerp for Xform {
+    fn lerp(self, other: Self, alpha: f32) -> Self {
         (self * (1.0 - alpha)) + (other * alpha)
     }
 }
 
 impl Lerp for Normal {
-    fn lerp(self, other: Normal, alpha: f32) -> Normal {
+    fn lerp(self, other: Self, alpha: f32) -> Self {
         (self * (1.0 - alpha)) + (other * alpha)
     }
 }
 
 impl Lerp for Point {
-    fn lerp(self, other: Point, alpha: f32) -> Point {
-        let s = self;
-        let o = other;
-        Point {
-            co: (s.co * (1.0 - alpha)) + (o.co * alpha),
-        }
+    fn lerp(self, other: Self, alpha: f32) -> Self {
+        let a = self.0;
+        let b = other.0;
+        Point((a * (1.0 - alpha)) + (b * alpha))
     }
 }
 
 impl Lerp for Vector {
-    fn lerp(self, other: Vector, alpha: f32) -> Vector {
+    fn lerp(self, other: Self, alpha: f32) -> Self {
         (self * (1.0 - alpha)) + (other * alpha)
     }
 }
@@ -215,20 +213,18 @@ mod tests {
 
     #[test]
     fn lerp_matrix() {
-        let a = Transform::new_from_values(
-            0.0, 2.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0,
-        );
-        let b = Transform::new_from_values(
+        let a = Xform::new(0.0, 2.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0);
+        let b = Xform::new(
             -1.0, 1.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0,
         );
 
-        let c1 = Transform::new_from_values(
+        let c1 = Xform::new(
             -0.25, 1.75, 2.25, 3.25, 4.25, 5.25, 6.25, 7.25, 8.25, 9.25, 10.25, 11.25,
         );
-        let c2 = Transform::new_from_values(
+        let c2 = Xform::new(
             -0.5, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5, 11.5,
         );
-        let c3 = Transform::new_from_values(
+        let c3 = Xform::new(
             -0.75, 1.25, 2.75, 3.75, 4.75, 5.75, 6.75, 7.75, 8.75, 9.75, 10.75, 11.75,
         );
 

src/light/distant_disk_light.rs

@@ -1,7 +1,5 @@
 use std::f64::consts::PI as PI_64;
 
-use kioku::Arena;
-
 use crate::{
     color::{Color, SpectralSample},
     lerp::lerp_slice,
@@ -13,24 +11,19 @@ use super::WorldLightSource;
 
 // TODO: handle case where radius = 0.0.
 
-#[derive(Copy, Clone, Debug)]
-pub struct DistantDiskLight<'a> {
-    radii: &'a [f32],
-    directions: &'a [Vector],
-    colors: &'a [Color],
+#[derive(Debug, Clone)]
+pub struct DistantDiskLight {
+    radii: Vec<f32>,
+    directions: Vec<Vector>,
+    colors: Vec<Color>,
 }
 
-impl<'a> DistantDiskLight<'a> {
-    pub fn new(
-        arena: &'a Arena,
-        radii: &[f32],
-        directions: &[Vector],
-        colors: &[Color],
-    ) -> DistantDiskLight<'a> {
+impl DistantDiskLight {
+    pub fn new(radii: &[f32], directions: &[Vector], colors: &[Color]) -> DistantDiskLight {
         DistantDiskLight {
-            radii: arena.copy_slice(&radii),
-            directions: arena.copy_slice(&directions),
-            colors: arena.copy_slice(&colors),
+            radii: radii.into(),
+            directions: directions.into(),
+            colors: colors.into(),
         }
     }
 
@@ -55,7 +48,7 @@ impl<'a> DistantDiskLight<'a> {
     // }
 }
 
-impl<'a> WorldLightSource for DistantDiskLight<'a> {
+impl WorldLightSource for DistantDiskLight {
     fn sample_from_point(
         &self,
         u: f32,
@@ -64,9 +57,9 @@ impl<'a> WorldLightSource for DistantDiskLight<'a> {
         time: f32,
     ) -> (SpectralSample, Vector, f32) {
         // Calculate time interpolated values
-        let radius: f64 = lerp_slice(self.radii, time) as f64;
-        let direction = lerp_slice(self.directions, time);
-        let col = lerp_slice(self.colors, time);
+        let radius: f64 = lerp_slice(&self.radii, time) as f64;
+        let direction = lerp_slice(&self.directions, time);
+        let col = lerp_slice(&self.colors, time);
         let solid_angle_inv = 1.0 / (2.0 * PI_64 * (1.0 - radius.cos()));
 
         // Create a coordinate system from the vector pointing at the center of

src/light/mod.rs

@@ -6,7 +6,7 @@ use std::fmt::Debug;
 
 use crate::{
     color::SpectralSample,
-    math::{Normal, Point, Transform, Vector},
+    math::{Normal, Point, Vector, XformFull},
     surface::Surface,
 };
 
@@ -34,7 +34,7 @@ pub trait SurfaceLight: Surface {
     /// - The pdf of the sample.
     fn sample_from_point(
         &self,
-        space: &Transform,
+        space: &XformFull,
         arr: Point,
         u: f32,
         v: f32,

src/light/rectangle_light.rs

@@ -5,8 +5,8 @@ use crate::{
     boundable::Boundable,
     color::{Color, SpectralSample},
     lerp::lerp_slice,
-    math::{cross, dot, Normal, Point, Transform, Vector},
-    ray::{RayBatch, RayStack},
+    math::{cross, dot, Normal, Point, Vector, XformFull},
+    ray::{LocalRay, Ray},
     sampling::{
         spherical_triangle_solid_angle, triangle_surface_area, uniform_sample_spherical_triangle,
         uniform_sample_triangle,
@@ -51,7 +51,7 @@ impl<'a> RectangleLight<'a> {
     // more efficiently by inlining it there.
     fn sample_pdf(
         &self,
-        space: &Transform,
+        space: &XformFull,
         arr: Point,
         sample_dir: Vector,
         hit_point: Point,
@@ -64,11 +64,10 @@ impl<'a> RectangleLight<'a> {
         let dim = lerp_slice(self.dimensions, time);
 
         // Get the four corners of the rectangle, transformed into world space
-        let space_inv = space.inverse();
-        let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0) * space_inv;
-        let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0) * space_inv;
-        let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0) * space_inv;
-        let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0) * space_inv;
+        let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0).xform(space);
+        let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0).xform(space);
 
         // Get the four corners of the rectangle, projected on to the unit
         // sphere centered around arr.
@@ -82,7 +81,7 @@ impl<'a> RectangleLight<'a> {
         let area_2 = spherical_triangle_solid_angle(sp4, sp1, sp3);
 
         // World-space surface normal
-        let normal = Normal::new(0.0, 0.0, 1.0) * space_inv;
+        let normal = Normal::new(0.0, 0.0, 1.0).xform_fast(space);
 
         // PDF
         if (area_1 + area_2) < SIMPLE_SAMPLING_THRESHOLD {
@@ -97,7 +96,7 @@ impl<'a> RectangleLight<'a> {
 
     // fn outgoing(
     //     &self,
-    //     space: &Transform,
+    //     space: &XformFull,
     //     dir: Vector,
     //     u: f32,
     //     v: f32,
@@ -120,7 +119,7 @@ impl<'a> RectangleLight<'a> {
 impl<'a> SurfaceLight for RectangleLight<'a> {
     fn sample_from_point(
         &self,
-        space: &Transform,
+        space: &XformFull,
         arr: Point,
         u: f32,
         v: f32,
@@ -135,11 +134,10 @@ impl<'a> SurfaceLight for RectangleLight<'a> {
         let surface_area_inv: f64 = 1.0 / surface_area;
 
         // Get the four corners of the rectangle, transformed into world space
-        let space_inv = space.inverse();
-        let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0) * space_inv;
-        let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0) * space_inv;
-        let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0) * space_inv;
-        let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0) * space_inv;
+        let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0).xform(space);
+        let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0).xform(space);
 
         // Get the four corners of the rectangle relative to arr.
         let lp1 = p1 - arr;
@@ -158,7 +156,7 @@ impl<'a> SurfaceLight for RectangleLight<'a> {
         let area_2 = spherical_triangle_solid_angle(sp4, sp1, sp3);
 
         // Calculate world-space surface normal
-        let normal = Normal::new(0.0, 0.0, 1.0) * space_inv;
+        let normal = Normal::new(0.0, 0.0, 1.0).xform_fast(space);
 
         if (area_1 + area_2) < SIMPLE_SAMPLING_THRESHOLD {
             // Simple sampling for more distant lights
@@ -215,18 +213,16 @@ impl<'a> SurfaceLight for RectangleLight<'a> {
             };
 
             // Project shadow_vec back onto the light's surface
-            let arr_local = arr * *space;
-            let shadow_vec_local = shadow_vec * *space;
+            let arr_local = arr.xform_inv(space);
+            let shadow_vec_local = shadow_vec.xform_inv(space);
             let shadow_vec_local = shadow_vec_local * (-arr_local.z() / shadow_vec_local.z());
             let mut sample_point_local = arr_local + shadow_vec_local;
             {
                 let x = sample_point_local.x().max(dim.0 * -0.5).min(dim.0 * 0.5);
                 let y = sample_point_local.y().max(dim.1 * -0.5).min(dim.1 * 0.5);
-                sample_point_local.set_x(x);
-                sample_point_local.set_y(y);
-                sample_point_local.set_z(0.0);
+                sample_point_local = Point::new(x, y, 0.0);
             }
-            let sample_point = sample_point_local * space_inv;
+            let sample_point = sample_point_local.xform(space);
             let point_err = 0.0001; // TODO: this is a hack, do properly.
 
             // Calculate pdf and light energy
@@ -255,85 +251,77 @@ impl<'a> SurfaceLight for RectangleLight<'a> {
 }
 
 impl<'a> Surface for RectangleLight<'a> {
-    fn intersect_rays(
+    fn intersect_ray(
         &self,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
-        isects: &mut [SurfaceIntersection],
-        shader: &dyn SurfaceShader,
-        space: &[Transform],
+        ray: &mut Ray,
+        _local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
+        _shaders: &[&dyn SurfaceShader],
     ) {
-        let _ = shader; // Silence 'unused' warning
+        let time = ray.time;
 
-        ray_stack.pop_do_next_task(|ray_idx| {
-            let time = rays.time(ray_idx);
-            let orig = rays.orig(ray_idx);
-            let dir = rays.dir(ray_idx);
-            let max_t = rays.max_t(ray_idx);
+        // Calculate time interpolated values.
+        let dim = lerp_slice(self.dimensions, time);
 
-            // Calculate time interpolated values
-            let dim = lerp_slice(self.dimensions, time);
-            let xform = lerp_slice(space, time);
+        // Get the four corners of the rectangle, transformed into world space.
+        let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0).xform(space);
+        let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0).xform(space);
+        let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0).xform(space);
 
-            let space_inv = xform.inverse();
+        // Test against two triangles that make up the light.
+        let ray_pre = triangle::RayTriPrecompute::new(ray.dir);
+        for tri in &[(p1, p2, p3), (p3, p4, p1)] {
+            if let Some((t, b0, b1, b2)) =
+                triangle::intersect_ray(ray.orig, ray_pre, ray.max_t, *tri)
+            {
+                if t < ray.max_t {
+                    if ray.is_occlusion() {
+                        *isect = SurfaceIntersection::Occlude;
+                        ray.mark_done();
+                        return;
+                    } else {
+                        let (pos, pos_err) = triangle::surface_point(*tri, (b0, b1, b2));
+                        let normal = cross(tri.0 - tri.1, tri.0 - tri.2).into_normal();
 
-            // Get the four corners of the rectangle, transformed into world space
-            let p1 = Point::new(dim.0 * 0.5, dim.1 * 0.5, 0.0) * space_inv;
-            let p2 = Point::new(dim.0 * -0.5, dim.1 * 0.5, 0.0) * space_inv;
-            let p3 = Point::new(dim.0 * -0.5, dim.1 * -0.5, 0.0) * space_inv;
-            let p4 = Point::new(dim.0 * 0.5, dim.1 * -0.5, 0.0) * space_inv;
+                        let intersection_data = SurfaceIntersectionData {
+                            incoming: ray.dir,
+                            t: t,
+                            pos: pos,
+                            pos_err: pos_err,
+                            nor: normal,
+                            nor_g: normal,
+                            local_space: *space,
+                            sample_pdf: self.sample_pdf(
+                                space,
+                                ray.orig,
+                                ray.dir,
+                                pos,
+                                ray.wavelength,
+                                time,
+                            ),
+                        };
 
-            // Test against two triangles that make up the light
-            let ray_pre = triangle::RayTriPrecompute::new(dir);
-            for tri in &[(p1, p2, p3), (p3, p4, p1)] {
-                if let Some((t, b0, b1, b2)) = triangle::intersect_ray(orig, ray_pre, max_t, *tri) {
-                    if t < max_t {
-                        if rays.is_occlusion(ray_idx) {
-                            isects[ray_idx] = SurfaceIntersection::Occlude;
-                            rays.mark_done(ray_idx);
-                        } else {
-                            let (pos, pos_err) = triangle::surface_point(*tri, (b0, b1, b2));
-                            let normal = cross(tri.0 - tri.1, tri.0 - tri.2).into_normal();
+                        let closure = {
+                            let inv_surface_area = (1.0 / (dim.0 as f64 * dim.1 as f64)) as f32;
+                            let color = lerp_slice(self.colors, time) * inv_surface_area;
+                            SurfaceClosure::Emit(color)
+                        };
 
-                            let intersection_data = SurfaceIntersectionData {
-                                incoming: dir,
-                                t: t,
-                                pos: pos,
-                                pos_err: pos_err,
-                                nor: normal,
-                                nor_g: normal,
-                                local_space: xform,
-                                sample_pdf: self.sample_pdf(
-                                    &xform,
-                                    orig,
-                                    dir,
-                                    pos,
-                                    rays.wavelength(ray_idx),
-                                    time,
-                                ),
-                            };
+                        // Fill in intersection.
+                        *isect = SurfaceIntersection::Hit {
+                            intersection_data: intersection_data,
+                            closure: closure,
+                        };
 
-                            let closure = {
-                                let inv_surface_area = (1.0 / (dim.0 as f64 * dim.1 as f64)) as f32;
-                                let color = lerp_slice(self.colors, time) * inv_surface_area;
-                                SurfaceClosure::Emit(color)
-                            };
-
-                            // Fill in intersection
-                            isects[ray_idx] = SurfaceIntersection::Hit {
-                                intersection_data: intersection_data,
-                                closure: closure,
-                            };
-
-                            // Set ray's max t
-                            rays.set_max_t(ray_idx, t);
-                        }
-
-                        break;
+                        ray.max_t = t;
                     }
+
+                    break;
                 }
             }
-        });
+        }
     }
 }
 

src/light/sphere_light.rs

@@ -7,8 +7,8 @@ use crate::{
     boundable::Boundable,
     color::{Color, SpectralSample},
     lerp::lerp_slice,
-    math::{coordinate_system_from_vector, dot, Normal, Point, Transform, Vector},
-    ray::{RayBatch, RayStack},
+    math::{coordinate_system_from_vector, dot, Normal, Point, Vector, XformFull},
+    ray::{LocalRay, Ray},
     sampling::{uniform_sample_cone, uniform_sample_cone_pdf, uniform_sample_sphere},
     shading::surface_closure::SurfaceClosure,
     shading::SurfaceShader,
@@ -50,7 +50,7 @@ impl<'a> SphereLight<'a> {
     // more efficiently by inlining it there.
     fn sample_pdf(
         &self,
-        space: &Transform,
+        space: &XformFull,
         arr: Point,
         sample_dir: Vector,
         sample_u: f32,
@@ -61,7 +61,7 @@ impl<'a> SphereLight<'a> {
         // We're not using these, silence warnings
         let _ = (sample_dir, sample_u, sample_v, wavelength);
 
-        let arr = arr * *space;
+        let arr = arr.xform_inv(space);
         let pos = Point::new(0.0, 0.0, 0.0);
         let radius: f64 = lerp_slice(self.radii, time) as f64;
 
@@ -84,7 +84,7 @@ impl<'a> SphereLight<'a> {
 impl<'a> SurfaceLight for SphereLight<'a> {
     fn sample_from_point(
         &self,
-        space: &Transform,
+        space: &XformFull,
         arr: Point,
         u: f32,
         v: f32,
@@ -92,12 +92,9 @@ impl<'a> SurfaceLight for SphereLight<'a> {
         time: f32,
     ) -> (SpectralSample, (Point, Normal, f32), f32) {
         // TODO: track fp error due to transforms
-        let arr = arr * *space;
+        let arr = arr.xform_inv(space);
         let pos = Point::new(0.0, 0.0, 0.0);
 
-        // Precalculate local->world space transform matrix
-        let inv_space = space.inverse();
-
         // Calculate time interpolated values
         let radius: f64 = lerp_slice(self.radii, time) as f64;
         let col = lerp_slice(self.colors, time);
@@ -115,7 +112,7 @@ impl<'a> SurfaceLight for SphereLight<'a> {
         // TODO: do this properly.  This is a total hack.
         let sample_point_err = {
             let v = Vector::new(radius as f32, radius as f32, radius as f32);
-            let v2 = v * inv_space;
+            let v2 = v.xform(space);
             v2.length() * SAMPLE_POINT_FUDGE
         };
 
@@ -159,8 +156,8 @@ impl<'a> SurfaceLight for SphereLight<'a> {
                 let normal = (arr + sample_vec).into_vector().normalized();
                 let point = normal * radius as f32;
                 (
-                    point.into_point() * inv_space,
-                    normal.into_normal() * inv_space,
+                    point.into_point().xform(space),
+                    normal.into_normal().xform_fast(space),
                 )
             };
             let pdf = uniform_sample_cone_pdf(cos_theta_max);
@@ -177,8 +174,8 @@ impl<'a> SurfaceLight for SphereLight<'a> {
                 let normal = (arr + sample_vec).into_vector().normalized();
                 let point = normal * radius as f32;
                 (
-                    point.into_point() * inv_space,
-                    normal.into_normal() * inv_space,
+                    point.into_point().xform(space),
+                    normal.into_normal().xform_fast(space),
                 )
             };
             let pdf = 1.0 / (4.0 * PI_64);
@@ -204,137 +201,122 @@ impl<'a> SurfaceLight for SphereLight<'a> {
 }
 
 impl<'a> Surface for SphereLight<'a> {
-    fn intersect_rays(
+    fn intersect_ray(
         &self,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
-        isects: &mut [SurfaceIntersection],
-        shader: &dyn SurfaceShader,
-        space: &[Transform],
+        ray: &mut Ray,
+        local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
+        _shaders: &[&dyn SurfaceShader],
     ) {
-        let _ = shader; // Silence 'unused' warning
+        let time = ray.time;
 
-        ray_stack.pop_do_next_task(|ray_idx| {
-            let time = rays.time(ray_idx);
+        // Get the radius of the sphere at the ray's time
+        let radius = lerp_slice(self.radii, time); // Radius of the sphere
 
-            // Get the transform space
-            let xform = lerp_slice(space, time);
+        // Code adapted to Rust from https://github.com/Tecla/Rayito
+        // Ray-sphere intersection can result in either zero, one or two points
+        // of intersection.  It turns into a quadratic equation, so we just find
+        // the solution using the quadratic formula.  Note that there is a
+        // slightly more stable form of it when computing it on a computer, and
+        // we use that method to keep everything accurate.
 
-            // Get the radius of the sphere at the ray's time
-            let radius = lerp_slice(self.radii, time); // Radius of the sphere
+        // Calculate quadratic coeffs
+        let a = local_ray.dir.length2();
+        let b = 2.0 * dot(local_ray.dir, local_ray.orig.into_vector());
+        let c = local_ray.orig.into_vector().length2() - (radius * radius);
 
-            // Get the ray origin and direction in local space
-            let orig = rays.orig_local(ray_idx).into_vector();
-            let dir = rays.dir(ray_idx) * xform;
+        let discriminant = (b * b) - (4.0 * a * c);
+        if discriminant < 0.0 {
+            // Discriminant less than zero?  No solution => no intersection.
+            return;
+        }
+        let discriminant = discriminant.sqrt();
 
-            // Code adapted to Rust from https://github.com/Tecla/Rayito
-            // Ray-sphere intersection can result in either zero, one or two points
-            // of intersection.  It turns into a quadratic equation, so we just find
-            // the solution using the quadratic formula.  Note that there is a
-            // slightly more stable form of it when computing it on a computer, and
-            // we use that method to keep everything accurate.
+        // Compute a more stable form of our param t (t0 = q/a, t1 = c/q)
+        // q = -0.5 * (b - sqrt(b * b - 4.0 * a * c)) if b < 0, or
+        // q = -0.5 * (b + sqrt(b * b - 4.0 * a * c)) if b >= 0
+        let q = if b < 0.0 {
+            -0.5 * (b - discriminant)
+        } else {
+            -0.5 * (b + discriminant)
+        };
 
-            // Calculate quadratic coeffs
-            let a = dir.length2();
-            let b = 2.0 * dot(dir, orig);
-            let c = orig.length2() - (radius * radius);
+        // Get our final parametric values
+        let mut t0 = q / a;
+        let mut t1 = if q != 0.0 { c / q } else { ray.max_t };
 
-            let discriminant = (b * b) - (4.0 * a * c);
-            if discriminant < 0.0 {
-                // Discriminant less than zero?  No solution => no intersection.
-                return;
-            }
-            let discriminant = discriminant.sqrt();
+        // Swap them so they are ordered right
+        if t0 > t1 {
+            use std::mem::swap;
+            swap(&mut t0, &mut t1);
+        }
 
-            // Compute a more stable form of our param t (t0 = q/a, t1 = c/q)
-            // q = -0.5 * (b - sqrt(b * b - 4.0 * a * c)) if b < 0, or
-            // q = -0.5 * (b + sqrt(b * b - 4.0 * a * c)) if b >= 0
-            let q = if b < 0.0 {
-                -0.5 * (b - discriminant)
-            } else {
-                -0.5 * (b + discriminant)
+        // Check our intersection for validity against this ray's extents
+        if t0 > ray.max_t || t1 <= 0.0 {
+            // Didn't hit because sphere is entirely outside of ray's extents
+            return;
+        }
+
+        let t = if t0 > 0.0 {
+            t0
+        } else if t1 <= ray.max_t {
+            t1
+        } else {
+            // Didn't hit because ray is entirely within the sphere, and
+            // therefore doesn't hit its surface.
+            return;
+        };
+
+        // We hit the sphere, so calculate intersection info.
+        if ray.is_occlusion() {
+            *isect = SurfaceIntersection::Occlude;
+            ray.mark_done();
+        } else {
+            // Position is calculated from the local-space ray and t, and then
+            // re-projected onto the surface of the sphere.
+            let t_pos = local_ray.orig + (local_ray.dir * t);
+            let unit_pos = t_pos.into_vector().normalized();
+            let pos = (unit_pos * radius).xform(space).into_point();
+
+            // TODO: proper error bounds.
+            let pos_err = 0.001;
+
+            let normal = unit_pos.into_normal().xform_fast(space);
+
+            let intersection_data = SurfaceIntersectionData {
+                incoming: ray.dir,
+                t: t,
+                pos: pos,
+                pos_err: pos_err,
+                nor: normal,
+                nor_g: normal,
+                local_space: *space,
+                sample_pdf: self.sample_pdf(
+                    space,
+                    ray.orig,
+                    ray.dir,
+                    0.0,
+                    0.0,
+                    ray.wavelength,
+                    time,
+                ),
             };
 
-            // Get our final parametric values
-            let mut t0 = q / a;
-            let mut t1 = if q != 0.0 { c / q } else { rays.max_t(ray_idx) };
-
-            // Swap them so they are ordered right
-            if t0 > t1 {
-                use std::mem::swap;
-                swap(&mut t0, &mut t1);
-            }
-
-            // Check our intersection for validity against this ray's extents
-            if t0 > rays.max_t(ray_idx) || t1 <= 0.0 {
-                // Didn't hit because sphere is entirely outside of ray's extents
-                return;
-            }
-
-            let t = if t0 > 0.0 {
-                t0
-            } else if t1 <= rays.max_t(ray_idx) {
-                t1
-            } else {
-                // Didn't hit because ray is entirely within the sphere, and
-                // therefore doesn't hit its surface.
-                return;
+            let closure = {
+                let inv_surface_area = (1.0 / (4.0 * PI_64 * radius as f64 * radius as f64)) as f32;
+                let color = lerp_slice(self.colors, time) * inv_surface_area;
+                SurfaceClosure::Emit(color)
             };
 
-            // We hit the sphere, so calculate intersection info.
-            if rays.is_occlusion(ray_idx) {
-                isects[ray_idx] = SurfaceIntersection::Occlude;
-                rays.mark_done(ray_idx);
-            } else {
-                let inv_xform = xform.inverse();
+            // Fill in intersection
+            *isect = SurfaceIntersection::Hit {
+                intersection_data: intersection_data,
+                closure: closure,
+            };
 
-                // Position is calculated from the local-space ray and t, and then
-                // re-projected onto the surface of the sphere.
-                let t_pos = orig + (dir * t);
-                let unit_pos = t_pos.normalized();
-                let pos = (unit_pos * radius * inv_xform).into_point();
-
-                // TODO: proper error bounds.
-                let pos_err = 0.001;
-
-                let normal = unit_pos.into_normal() * inv_xform;
-
-                let intersection_data = SurfaceIntersectionData {
-                    incoming: rays.dir(ray_idx),
-                    t: t,
-                    pos: pos,
-                    pos_err: pos_err,
-                    nor: normal,
-                    nor_g: normal,
-                    local_space: xform,
-                    sample_pdf: self.sample_pdf(
-                        &xform,
-                        rays.orig(ray_idx),
-                        rays.dir(ray_idx),
-                        0.0,
-                        0.0,
-                        rays.wavelength(ray_idx),
-                        time,
-                    ),
-                };
-
-                let closure = {
-                    let inv_surface_area =
-                        (1.0 / (4.0 * PI_64 * radius as f64 * radius as f64)) as f32;
-                    let color = lerp_slice(self.colors, time) * inv_surface_area;
-                    SurfaceClosure::Emit(color)
-                };
-
-                // Fill in intersection
-                isects[ray_idx] = SurfaceIntersection::Hit {
-                    intersection_data: intersection_data,
-                    closure: closure,
-                };
-
-                // Set ray's max t
-                rays.set_max_t(ray_idx, t);
-            }
-        });
+            ray.max_t = t;
+        }
     }
 }
 

src/main.rs

@@ -22,8 +22,6 @@ mod boundable;
 mod camera;
 mod color;
 mod fp_utils;
-mod hash;
-mod hilbert;
 mod image;
 mod lerp;
 mod light;
@@ -34,11 +32,13 @@ mod ray;
 mod renderer;
 mod sampling;
 mod scene;
+mod scramble;
 mod shading;
+mod space_fill;
 mod surface;
 mod timer;
 mod tracer;
-mod transform_stack;
+// mod transform_stack;
 
 use std::{fs::File, io, io::Read, mem, path::Path, str::FromStr};
 
@@ -51,7 +51,6 @@ use crate::{
     accel::BVH4Node,
     bbox::BBox,
     parse::{parse_scene, DataTree},
-    renderer::LightPath,
     surface::SurfaceIntersection,
     timer::Timer,
 };
@@ -89,11 +88,11 @@ fn main() {
                 }),
         )
         .arg(
-            Arg::with_name("max_bucket_samples")
+            Arg::with_name("bucket_size")
                 .short("b")
-                .long("spb")
+                .long("bucket_size")
                 .value_name("N")
-                .help("Target number of samples per bucket (determines bucket size)")
+                .help("Height and width of each render bucket in pixels.")
                 .takes_value(true)
                 .validator(|s| {
                     usize::from_str(&s)
@@ -163,7 +162,6 @@ fn main() {
             "SurfaceIntersection size:  {} bytes",
             mem::size_of::<SurfaceIntersection>()
         );
-        println!("LightPath size: {} bytes", mem::size_of::<LightPath>());
         println!("BBox size: {} bytes", mem::size_of::<BBox>());
         // println!("BVHNode size: {} bytes", mem::size_of::<BVHNode>());
         println!("BVH4Node size: {} bytes", mem::size_of::<BVH4Node>());
@@ -259,12 +257,11 @@ fn main() {
                     r.spp = usize::from_str(spp).unwrap();
                 }
 
-                let max_samples_per_bucket =
-                    if let Some(max_samples_per_bucket) = args.value_of("max_bucket_samples") {
-                        u32::from_str(max_samples_per_bucket).unwrap()
-                    } else {
-                        4096
-                    };
+                let bucket_size = if let Some(bucket_size) = args.value_of("bucket_size") {
+                    u32::from_str(bucket_size).unwrap()
+                } else {
+                    32
+                };
 
                 let thread_count = if let Some(threads) = args.value_of("threads") {
                     u32::from_str(threads).unwrap()
@@ -280,7 +277,7 @@ fn main() {
                     println!("Rendering scene with {} threads...", thread_count);
                 }
                 let (mut image, rstats) = r.render(
-                    max_samples_per_bucket,
+                    bucket_size,
                     crop,
                     thread_count,
                     args.is_present("serialized_output"),
@@ -288,30 +285,9 @@ fn main() {
                 // Print render stats
                 if !args.is_present("serialized_output") {
                     let rtime = t.tick();
-                    let ntime = rtime as f64 / rstats.total_time;
                     println!("\tRendered scene in {:.3}s", rtime);
-                    println!(
-                        "\t\tTrace:                  {:.3}s",
-                        ntime * rstats.trace_time
-                    );
                     println!("\t\t\tRays traced:          {}", rstats.ray_count);
-                    println!(
-                        "\t\t\tRays/sec:             {}",
-                        (rstats.ray_count as f64 / (ntime * rstats.trace_time) as f64) as u64
-                    );
                     println!("\t\t\tRay/node tests:       {}", rstats.accel_node_visits);
-                    println!(
-                        "\t\tInitial ray generation: {:.3}s",
-                        ntime * rstats.initial_ray_generation_time
-                    );
-                    println!(
-                        "\t\tRay generation:         {:.3}s",
-                        ntime * rstats.ray_generation_time
-                    );
-                    println!(
-                        "\t\tSample writing:         {:.3}s",
-                        ntime * rstats.sample_writing_time
-                    );
                 }
 
                 // Write to disk

src/math.rs

@@ -2,18 +2,78 @@
 
 use std::f32;
 
-pub use math3d::{cross, dot, CrossProduct, DotProduct, Normal, Point, Transform, Vector};
+pub use rmath::{
+    cross, cross_fast, dot, dot_fast, wide4::Float4, AsXform, CrossProduct, DotProduct, Normal,
+    Point, Vector, Xform, XformFull,
+};
+
+/// Clamps a value between a min and max.
+pub fn clamp<T: PartialOrd>(v: T, lower: T, upper: T) -> T {
+    if v < lower {
+        lower
+    } else if v > upper {
+        upper
+    } else {
+        v
+    }
+}
+
+// The stdlib min function is slower than a simple if statement for some reason.
+pub fn fast_minf32(a: f32, b: f32) -> f32 {
+    if a < b {
+        a
+    } else {
+        b
+    }
+}
+
+// The stdlib max function is slower than a simple if statement for some reason.
+pub fn fast_maxf32(a: f32, b: f32) -> f32 {
+    if a > b {
+        a
+    } else {
+        b
+    }
+}
+
+/// Rounds an integer up to the next power of two.
+pub fn upper_power_of_two(mut v: u32) -> u32 {
+    v -= 1;
+    v |= v >> 1;
+    v |= v >> 2;
+    v |= v >> 4;
+    v |= v >> 8;
+    v |= v >> 16;
+    v + 1
+}
 
 /// Gets the log base 2 of the given integer
-pub fn log2_64(n: u64) -> u64 {
-    // This works by finding the largest non-zero binary digit in the
-    // number.  Its bit position is then the log2 of the integer.
+pub fn log2_64(mut value: u64) -> u64 {
+    // This works by doing a binary search for the largest non-zero binary
+    // digit in the number.  Its bit position is then the log2 of the integer.
 
-    if n == 0 {
-        0
-    } else {
-        (63 - n.leading_zeros()) as u64
+    let mut log = 0;
+
+    const POWERS: [(u64, u64); 6] = [
+        (32, (1 << 32) - 1),
+        (16, (1 << 16) - 1),
+        (8, (1 << 8) - 1),
+        (4, (1 << 4) - 1),
+        (2, (1 << 2) - 1),
+        (1, (1 << 1) - 1),
+    ];
+
+    for &(i, j) in &POWERS {
+        let tmp = value >> i;
+        if tmp != 0 {
+            log += i;
+            value = tmp;
+        } else {
+            value &= j;
+        }
     }
+
+    log
 }
 
 /// Creates a coordinate system from a single vector.

src/parse/psy.rs

@@ -6,14 +6,10 @@ use nom::{combinator::all_consuming, sequence::tuple, IResult};
 
 use kioku::Arena;
 
+use color::rec709_e_to_xyz;
+
 use crate::{
-    camera::Camera,
-    color::{rec709_e_to_xyz, Color},
-    light::WorldLightSource,
-    math::Transform,
-    renderer::Renderer,
-    scene::Scene,
-    scene::World,
+    camera::Camera, color::Color, math::Xform, renderer::Renderer, scene::Scene, scene::World,
 };
 
 use super::{
@@ -153,13 +149,10 @@ pub fn parse_scene<'a>(
     )?;
 
     // Parse camera
-    let camera = parse_camera(
-        arena,
-        tree.iter_children_with_type("Camera").nth(0).unwrap(),
-    )?;
+    let camera = parse_camera(tree.iter_children_with_type("Camera").nth(0).unwrap())?;
 
     // Parse world
-    let world = parse_world(arena, tree.iter_children_with_type("World").nth(0).unwrap())?;
+    let world = parse_world(tree.iter_children_with_type("World").nth(0).unwrap())?;
 
     // Parse root scene assembly
     let assembly = parse_assembly(
@@ -350,7 +343,7 @@ fn parse_render_settings(tree: &DataTree) -> Result<((u32, u32), u32, u32), PsyP
     };
 }
 
-fn parse_camera<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<Camera<'a>, PsyParseError> {
+fn parse_camera<'a>(tree: &'a DataTree) -> Result<Camera, PsyParseError> {
     if let DataTree::Internal { ref children, .. } = *tree {
         let mut mats = Vec::new();
         let mut fovs = Vec::new();
@@ -435,13 +428,7 @@ fn parse_camera<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<Camera<'a>,
             }
         }
 
-        return Ok(Camera::new(
-            arena,
-            &mats,
-            &fovs,
-            &aperture_radii,
-            &focus_distances,
-        ));
+        return Ok(Camera::new(&mats, &fovs, &aperture_radii, &focus_distances));
     } else {
         return Err(PsyParseError::ExpectedInternalNode(
             tree.byte_offset(),
@@ -452,10 +439,10 @@ fn parse_camera<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<Camera<'a>,
     }
 }
 
-fn parse_world<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<World<'a>, PsyParseError> {
+fn parse_world(tree: &DataTree) -> Result<World, PsyParseError> {
     if tree.is_internal() {
         let background_color;
-        let mut lights: Vec<&dyn WorldLightSource> = Vec::new();
+        let mut lights: Vec<_> = Vec::new();
 
         // Parse background shader
         let bgs = {
@@ -531,7 +518,7 @@ fn parse_world<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<World<'a>, Ps
         for child in tree.iter_children() {
             match *child {
                 DataTree::Internal { type_name, .. } if type_name == "DistantDiskLight" => {
-                    lights.push(arena.alloc(parse_distant_disk_light(arena, child)?));
+                    lights.push(parse_distant_disk_light(child)?);
                 }
 
                 _ => {}
@@ -541,7 +528,7 @@ fn parse_world<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<World<'a>, Ps
         // Build and return the world
         return Ok(World {
             background_color: background_color,
-            lights: arena.copy_slice(&lights),
+            lights: lights,
         });
     } else {
         return Err(PsyParseError::ExpectedInternalNode(
@@ -553,17 +540,17 @@ fn parse_world<'a>(arena: &'a Arena, tree: &'a DataTree) -> Result<World<'a>, Ps
     }
 }
 
-pub fn parse_matrix(contents: &str) -> Result<Transform, PsyParseError> {
+pub fn parse_matrix(contents: &str) -> Result<Xform, PsyParseError> {
     if let IResult::Ok((leftover, ns)) = all_consuming(tuple((
         ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32, ws_f32,
         ws_f32, ws_f32, ws_f32, ws_f32, ws_f32,
     )))(contents)
     {
         if leftover.is_empty() {
-            return Ok(Transform::new_from_values(
+            return Ok(Xform::new(
                 // We throw away the last row, since it's not necessarily affine.
                 // TODO: is there a more correct way to handle this?
-                ns.0, ns.4, ns.8, ns.12, ns.1, ns.5, ns.9, ns.13, ns.2, ns.6, ns.10, ns.14,
+                ns.0, ns.1, ns.2, ns.4, ns.5, ns.6, ns.8, ns.9, ns.10, ns.12, ns.13, ns.14,
             ));
         }
     }

src/parse/psy_assembly.rs

@@ -50,23 +50,6 @@ pub fn parse_assembly<'a>(
                         child.iter_leaf_children_with_type("Data").nth(0).unwrap().1
                     };
 
-                    // Get surface shader binding, if any.
-                    let surface_shader_name = if child
-                        .iter_leaf_children_with_type("SurfaceShaderBind")
-                        .count()
-                        > 0
-                    {
-                        Some(
-                            child
-                                .iter_leaf_children_with_type("SurfaceShaderBind")
-                                .nth(0)
-                                .unwrap()
-                                .1,
-                        )
-                    } else {
-                        None
-                    };
-
                     // Get xforms
                     let mut xforms = Vec::new();
                     for (_, contents, _) in child.iter_leaf_children_with_type("Transform") {
@@ -75,7 +58,7 @@ pub fn parse_assembly<'a>(
 
                     // Add instance
                     if builder.name_exists(name) {
-                        builder.add_instance(name, surface_shader_name, Some(&xforms));
+                        builder.add_instance(name, Some(&xforms));
                     } else {
                         return Err(PsyParseError::InstancedMissingData(
                             child.iter_leaf_children_with_type("Data").nth(0).unwrap().2,
@@ -113,7 +96,11 @@ pub fn parse_assembly<'a>(
                     {
                         builder.add_object(
                             ident,
-                            Object::Surface(arena.alloc(parse_mesh_surface(arena, child)?)),
+                            Object::Surface(arena.alloc(parse_mesh_surface(
+                                arena,
+                                child,
+                                &builder.surface_shader_map,
+                            )?)),
                         );
                     } else {
                         // TODO: error condition of some kind, because no ident

src/parse/psy_light.rs

@@ -17,10 +17,7 @@ use super::{
     DataTree,
 };
 
-pub fn parse_distant_disk_light<'a>(
-    arena: &'a Arena,
-    tree: &'a DataTree,
-) -> Result<DistantDiskLight<'a>, PsyParseError> {
+pub fn parse_distant_disk_light<'a>(tree: &'a DataTree) -> Result<DistantDiskLight, PsyParseError> {
     if let DataTree::Internal { ref children, .. } = *tree {
         let mut radii = Vec::new();
         let mut directions = Vec::new();
@@ -77,7 +74,7 @@ pub fn parse_distant_disk_light<'a>(
             }
         }
 
-        return Ok(DistantDiskLight::new(arena, &radii, &directions, &colors));
+        return Ok(DistantDiskLight::new(&radii, &directions, &colors));
     } else {
         return Err(PsyParseError::UnknownError(tree.byte_offset()));
     }

src/parse/psy_mesh_surface.rs

@@ -1,6 +1,6 @@
 #![allow(dead_code)]
 
-use std::result::Result;
+use std::{collections::HashMap, result::Result};
 
 use nom::{sequence::tuple, IResult};
 
@@ -27,7 +27,9 @@ use super::{
 pub fn parse_mesh_surface<'a>(
     arena: &'a Arena,
     tree: &'a DataTree,
+    surface_shader_map: &HashMap<String, usize>,
 ) -> Result<TriangleMesh<'a>, PsyParseError> {
+    let mut shader_idx = None;
     let mut verts = Vec::new(); // Vec of vecs, one for each time sample
     let mut normals = Vec::new(); // Vec of vecs, on for each time sample
     let mut face_vert_counts = Vec::new();
@@ -36,6 +38,20 @@ pub fn parse_mesh_surface<'a>(
     // TODO: make sure there are the right number of various children,
     // and other validation.
 
+    // Get surface shader binding, if any.
+    if tree
+        .iter_leaf_children_with_type("SurfaceShaderBind")
+        .count()
+        > 0
+    {
+        let name = tree
+            .iter_leaf_children_with_type("SurfaceShaderBind")
+            .nth(0)
+            .unwrap()
+            .1;
+        shader_idx = surface_shader_map.get(name).map(|i| *i);
+    }
+
     // Get verts
     for (_, mut text, _) in tree.iter_leaf_children_with_type("Vertices") {
         // Collect verts for this time sample
@@ -116,6 +132,7 @@ pub fn parse_mesh_surface<'a>(
 
     Ok(TriangleMesh::from_verts_and_indices(
         arena,
+        shader_idx,
         &verts,
         &if normals.is_empty() {
             None

src/ray.rs

@@ -1,16 +1,11 @@
 #![allow(dead_code)]
 
-use glam::BVec4A;
+use crate::math::{Point, Vector, XformFull};
 
-use crate::math::{Point, Transform, Vector};
-
-type RayIndexType = u16;
 type FlagType = u8;
 const OCCLUSION_FLAG: FlagType = 1;
 const DONE_FLAG: FlagType = 1 << 1;
 
-/// This is never used directly in ray tracing--it's only used as a convenience
-/// for filling the RayBatch structure.
 #[derive(Debug, Copy, Clone)]
 pub struct Ray {
     pub orig: Point,
@@ -18,384 +13,85 @@ pub struct Ray {
     pub time: f32,
     pub wavelength: f32,
     pub max_t: f32,
+    pub flags: FlagType,
 }
 
-/// The hot (frequently accessed) parts of ray data.
+/// A specifically local-space ray, for passing to functions when we've
+/// already calculated the local-space version of a ray for the object
+/// in question.
+///
+/// Also includes `dir_inv`, which is generally useful to have as well.
 #[derive(Debug, Copy, Clone)]
-struct RayHot {
-    orig_local: Point,     // Local-space ray origin
-    dir_inv_local: Vector, // Local-space 1.0/ray direction
-    max_t: f32,
-    time: f32,
-    flags: FlagType,
+pub struct LocalRay {
+    pub orig: Point,
+    pub dir: Vector,
+    pub dir_inv: Vector,
 }
 
-/// The cold (infrequently accessed) parts of ray data.
-#[derive(Debug, Copy, Clone)]
-struct RayCold {
-    orig: Point, // World-space ray origin
-    dir: Vector, // World-space ray direction
-    wavelength: f32,
-}
-
-/// A batch of rays, separated into hot and cold parts.
-#[derive(Debug)]
-pub struct RayBatch {
-    hot: Vec<RayHot>,
-    cold: Vec<RayCold>,
-}
-
-impl RayBatch {
-    /// Creates a new empty ray batch.
-    pub fn new() -> RayBatch {
-        RayBatch {
-            hot: Vec::new(),
-            cold: Vec::new(),
-        }
-    }
-
-    /// Creates a new empty ray batch, with pre-allocated capacity for
-    /// `n` rays.
-    pub fn with_capacity(n: usize) -> RayBatch {
-        RayBatch {
-            hot: Vec::with_capacity(n),
-            cold: Vec::with_capacity(n),
-        }
-    }
-
-    pub fn push(&mut self, ray: Ray, is_occlusion: bool) {
-        self.hot.push(RayHot {
-            orig_local: ray.orig,   // Bogus, to place-hold.
-            dir_inv_local: ray.dir, // Bogus, to place-hold.
-            max_t: ray.max_t,
-            time: ray.time,
+impl Ray {
+    pub fn new(
+        orig: Point,
+        dir: Vector,
+        time: f32,
+        wavelength: f32,
+        max_t: f32,
+        is_occlusion: bool,
+    ) -> Self {
+        Self {
+            orig: orig,
+            dir: dir,
+            time: time,
+            wavelength: wavelength,
+            max_t: max_t,
             flags: if is_occlusion { OCCLUSION_FLAG } else { 0 },
-        });
-        self.cold.push(RayCold {
-            orig: ray.orig,
-            dir: ray.dir,
-            wavelength: ray.wavelength,
-        });
+        }
     }
 
-    pub fn swap(&mut self, a: usize, b: usize) {
-        self.hot.swap(a, b);
-        self.cold.swap(a, b);
+    /// Creates a local ray from the given transform.
+    pub fn to_local_xform(&self, xform: &XformFull) -> LocalRay {
+        let orig = self.orig.xform_inv(xform);
+        let dir = self.dir.xform_inv(xform);
+
+        LocalRay {
+            orig: orig,
+            dir: dir,
+            dir_inv: dir.recip(),
+        }
     }
 
-    pub fn set_from_ray(&mut self, ray: &Ray, is_occlusion: bool, idx: usize) {
-        self.hot[idx].orig_local = ray.orig;
-        self.hot[idx].dir_inv_local = Vector {
-            co: ray.dir.co.recip(),
-        };
-        self.hot[idx].max_t = ray.max_t;
-        self.hot[idx].time = ray.time;
-        self.hot[idx].flags = if is_occlusion { OCCLUSION_FLAG } else { 0 };
-
-        self.cold[idx].orig = ray.orig;
-        self.cold[idx].dir = ray.dir;
-        self.cold[idx].wavelength = ray.wavelength;
+    /// Creates a local ray with no transform applied.
+    pub fn to_local(&self) -> LocalRay {
+        LocalRay {
+            orig: self.orig,
+            dir: self.dir,
+            dir_inv: self.dir.recip(),
+        }
     }
 
-    pub fn truncate(&mut self, len: usize) {
-        self.hot.truncate(len);
-        self.cold.truncate(len);
-    }
-
-    /// Clear all rays, settings the size of the batch back to zero.
-    ///
-    /// Capacity is maintained.
-    pub fn clear(&mut self) {
-        self.hot.clear();
-        self.cold.clear();
-    }
-
-    pub fn len(&self) -> usize {
-        self.hot.len()
-    }
-
-    /// Updates the accel data of the given ray (at index `idx`) with the
-    /// given world-to-local-space transform matrix.
-    ///
-    /// This should be called when entering (and exiting) traversal of a
-    /// new transform space.
-    pub fn update_local(&mut self, idx: usize, xform: &Transform) {
-        self.hot[idx].orig_local = self.cold[idx].orig * *xform;
-        self.hot[idx].dir_inv_local = Vector {
-            co: (self.cold[idx].dir * *xform).co.recip(),
-        };
-    }
-
-    //==========================================================
-    // Data access
+    //---------------------------------------------------------
+    // Flags.
 
+    /// Returns whether this is an occlusion ray.
     #[inline(always)]
-    pub fn orig(&self, idx: usize) -> Point {
-        self.cold[idx].orig
+    pub fn is_occlusion(&self) -> bool {
+        (self.flags & OCCLUSION_FLAG) != 0
     }
 
+    /// Returns whether this ray has finished traversal.
     #[inline(always)]
-    pub fn dir(&self, idx: usize) -> Vector {
-        self.cold[idx].dir
+    pub fn is_done(&self) -> bool {
+        (self.flags & DONE_FLAG) != 0
     }
 
+    /// Marks this as an occlusion ray.
     #[inline(always)]
-    pub fn orig_local(&self, idx: usize) -> Point {
-        self.hot[idx].orig_local
+    pub fn mark_occlusion(&mut self) {
+        self.flags |= OCCLUSION_FLAG
     }
 
+    /// Marks this as having finished traversal.
     #[inline(always)]
-    pub fn dir_inv_local(&self, idx: usize) -> Vector {
-        self.hot[idx].dir_inv_local
-    }
-
-    #[inline(always)]
-    pub fn time(&self, idx: usize) -> f32 {
-        self.hot[idx].time
-    }
-
-    #[inline(always)]
-    pub fn max_t(&self, idx: usize) -> f32 {
-        self.hot[idx].max_t
-    }
-
-    #[inline(always)]
-    pub fn set_max_t(&mut self, idx: usize, new_max_t: f32) {
-        self.hot[idx].max_t = new_max_t;
-    }
-
-    #[inline(always)]
-    pub fn wavelength(&self, idx: usize) -> f32 {
-        self.cold[idx].wavelength
-    }
-
-    /// Returns whether the given ray (at index `idx`) is an occlusion ray.
-    #[inline(always)]
-    pub fn is_occlusion(&self, idx: usize) -> bool {
-        (self.hot[idx].flags & OCCLUSION_FLAG) != 0
-    }
-
-    /// Returns whether the given ray (at index `idx`) has finished traversal.
-    #[inline(always)]
-    pub fn is_done(&self, idx: usize) -> bool {
-        (self.hot[idx].flags & DONE_FLAG) != 0
-    }
-
-    /// Marks the given ray (at index `idx`) as an occlusion ray.
-    #[inline(always)]
-    pub fn mark_occlusion(&mut self, idx: usize) {
-        self.hot[idx].flags |= OCCLUSION_FLAG
-    }
-
-    /// Marks the given ray (at index `idx`) as having finished traversal.
-    #[inline(always)]
-    pub fn mark_done(&mut self, idx: usize) {
-        self.hot[idx].flags |= DONE_FLAG
+    pub fn mark_done(&mut self) {
+        self.flags |= DONE_FLAG
     }
 }
-
-/// A structure used for tracking traversal of a ray batch through a scene.
-#[derive(Debug)]
-pub struct RayStack {
-    lanes: Vec<Lane>,
-    tasks: Vec<RayTask>,
-}
-
-impl RayStack {
-    pub fn new() -> RayStack {
-        RayStack {
-            lanes: Vec::new(),
-            tasks: Vec::new(),
-        }
-    }
-
-    /// Returns whether the stack is empty of tasks or not.
-    pub fn is_empty(&self) -> bool {
-        self.tasks.is_empty()
-    }
-
-    /// Makes sure there are at least `count` lanes.
-    pub fn ensure_lane_count(&mut self, count: usize) {
-        while self.lanes.len() < count {
-            self.lanes.push(Lane {
-                idxs: Vec::new(),
-                end_len: 0,
-            })
-        }
-    }
-
-    pub fn ray_count_in_next_task(&self) -> usize {
-        let task = self.tasks.last().unwrap();
-        let end = self.lanes[task.lane].end_len;
-        end - task.start_idx
-    }
-
-    pub fn next_task_ray_idx(&self, i: usize) -> usize {
-        let task = self.tasks.last().unwrap();
-        let i = i + task.start_idx;
-        debug_assert!(i < self.lanes[task.lane].end_len);
-        self.lanes[task.lane].idxs[i] as usize
-    }
-
-    /// Clears the lanes and tasks of the RayStack.
-    ///
-    /// Note: this is (importantly) different than calling clear individually
-    /// on the `lanes` and `tasks` members.  Specifically, we don't want to
-    /// clear `lanes` itself, as that would also free all the memory of the
-    /// individual lanes.  Instead, we want to iterate over the individual
-    /// lanes and clear them, but leave `lanes` itself untouched.
-    pub fn clear(&mut self) {
-        for lane in self.lanes.iter_mut() {
-            lane.idxs.clear();
-            lane.end_len = 0;
-        }
-
-        self.tasks.clear();
-    }
-
-    /// Pushes the given ray index onto the end of the specified lane.
-    pub fn push_ray_index(&mut self, ray_idx: usize, lane: usize) {
-        assert!(self.lanes.len() > lane);
-        self.lanes[lane].idxs.push(ray_idx as RayIndexType);
-    }
-
-    /// Pushes any excess indices on the given lane to a new task on the
-    /// task stack.
-    ///
-    /// Returns whether a task was pushed or not.  No task will be pushed
-    /// if there are no excess indices on the end of the lane.
-    pub fn push_lane_to_task(&mut self, lane_idx: usize) -> bool {
-        if self.lanes[lane_idx].end_len < self.lanes[lane_idx].idxs.len() {
-            self.tasks.push(RayTask {
-                lane: lane_idx,
-                start_idx: self.lanes[lane_idx].end_len,
-            });
-            self.lanes[lane_idx].end_len = self.lanes[lane_idx].idxs.len();
-            true
-        } else {
-            false
-        }
-    }
-
-    /// Takes the given list of lane indices, and pushes any excess indices on
-    /// the end of each into a new task, in the order provided.
-    pub fn push_lanes_to_tasks(&mut self, lane_idxs: &[usize]) {
-        for &l in lane_idxs {
-            self.push_lane_to_task(l);
-        }
-    }
-
-    pub fn duplicate_next_task(&mut self) {
-        let task = self.tasks.last().unwrap();
-        let l = task.lane;
-        let start = task.start_idx;
-        let end = self.lanes[l].end_len;
-
-        // Extend the indices vector
-        self.lanes[l].idxs.reserve(end - start);
-        let old_len = self.lanes[l].idxs.len();
-        let new_len = old_len + end - start;
-        unsafe {
-            self.lanes[l].idxs.set_len(new_len);
-        }
-
-        // Copy elements
-        copy_in_place::copy_in_place(&mut self.lanes[l].idxs, start..end, end);
-
-        // Push the new task onto the stack
-        self.tasks.push(RayTask {
-            lane: l,
-            start_idx: end,
-        });
-
-        self.lanes[l].end_len = self.lanes[l].idxs.len();
-    }
-
-    // Pops the next task off the stack.
-    pub fn pop_task(&mut self) {
-        let task = self.tasks.pop().unwrap();
-        self.lanes[task.lane].end_len = task.start_idx;
-        self.lanes[task.lane].idxs.truncate(task.start_idx);
-    }
-
-    // Executes a task without popping it from the task stack.
-    pub fn do_next_task<F>(&mut self, mut handle_ray: F)
-    where
-        F: FnMut(usize),
-    {
-        let task = self.tasks.last().unwrap();
-        let task_range = (task.start_idx, self.lanes[task.lane].end_len);
-
-        // Execute task.
-        for i in task_range.0..task_range.1 {
-            let ray_idx = self.lanes[task.lane].idxs[i];
-            handle_ray(ray_idx as usize);
-        }
-    }
-
-    /// Pops the next task off the stack, and executes the provided closure for
-    /// each ray index in the task.
-    #[inline(always)]
-    pub fn pop_do_next_task<F>(&mut self, handle_ray: F)
-    where
-        F: FnMut(usize),
-    {
-        self.do_next_task(handle_ray);
-        self.pop_task();
-    }
-
-    /// Pops the next task off the stack, executes the provided closure for
-    /// each ray index in the task, and pushes the ray indices back onto the
-    /// indicated lanes.
-    pub fn pop_do_next_task_and_push_rays<F>(&mut self, output_lane_count: usize, mut handle_ray: F)
-    where
-        F: FnMut(usize) -> BVec4A,
-    {
-        // Pop the task and do necessary bookkeeping.
-        let task = self.tasks.pop().unwrap();
-        let task_range = (task.start_idx, self.lanes[task.lane].end_len);
-        self.lanes[task.lane].end_len = task.start_idx;
-
-        // SAFETY: this is probably evil, and depends on behavior of Vec that
-        // are not actually promised.  But we're essentially truncating the lane
-        // to the start of our task range, but will continue to access it's
-        // elements beyond that range via `get_unchecked()` below.  Because the
-        // memory is not freed nor altered, this is safe.  However, again, the
-        // Vec apis don't promise this behavior.  So:
-        //
-        // TODO: build a slightly different lane abstraction to get this same
-        // efficiency without depending on implicit Vec behavior.
-        unsafe {
-            self.lanes[task.lane].idxs.set_len(task.start_idx);
-        }
-
-        // Execute task.
-        for i in task_range.0..task_range.1 {
-            let ray_idx = *unsafe { self.lanes[task.lane].idxs.get_unchecked(i) };
-            let push_mask = handle_ray(ray_idx as usize).bitmask();
-            for l in 0..output_lane_count {
-                if (push_mask & (1 << l)) != 0 {
-                    self.lanes[l as usize].idxs.push(ray_idx);
-                }
-            }
-        }
-    }
-}
-
-/// A lane within a RayStack.
-#[derive(Debug)]
-struct Lane {
-    idxs: Vec<RayIndexType>,
-    end_len: usize,
-}
-
-/// A task within a RayStack.
-//
-// Specifies the lane that the relevant ray pointers are in, and the
-// starting index within that lane.  The relevant pointers are always
-// `&[start_idx..]` within the given lane.
-#[derive(Debug)]
-struct RayTask {
-    lane: usize,
-    start_idx: usize,
-}

src/sampling/monte_carlo.rs

@@ -2,7 +2,7 @@
 
 use std::{f32::consts::FRAC_PI_4 as QPI_32, f32::consts::PI as PI_32, f64::consts::PI as PI_64};
 
-use crate::math::{cross, dot, Point, Vector};
+use crate::math::{cross_fast, dot_fast, Point, Vector};
 
 /// Maps the unit square to the unit circle.
 /// NOTE: x and y should be distributed within [-1, 1],
@@ -90,7 +90,7 @@ pub fn uniform_sample_triangle(va: Vector, vb: Vector, vc: Vector, i: f32, j: f3
 
 /// Calculates the surface area of a triangle.
 pub fn triangle_surface_area(p0: Point, p1: Point, p2: Point) -> f32 {
-    0.5 * cross(p1 - p0, p2 - p0).length()
+    0.5 * cross_fast(p1 - p0, p2 - p0).length()
 }
 
 /// Calculates the projected solid angle of a spherical triangle.
@@ -98,9 +98,9 @@ pub fn triangle_surface_area(p0: Point, p1: Point, p2: Point) -> f32 {
 /// A, B, and C are the points of the triangle on a unit sphere.
 pub fn spherical_triangle_solid_angle(va: Vector, vb: Vector, vc: Vector) -> f32 {
     // Calculate sines and cosines of the spherical triangle's edge lengths
-    let cos_a: f64 = dot(vb, vc).max(-1.0).min(1.0) as f64;
-    let cos_b: f64 = dot(vc, va).max(-1.0).min(1.0) as f64;
-    let cos_c: f64 = dot(va, vb).max(-1.0).min(1.0) as f64;
+    let cos_a: f64 = dot_fast(vb, vc).max(-1.0).min(1.0) as f64;
+    let cos_b: f64 = dot_fast(vc, va).max(-1.0).min(1.0) as f64;
+    let cos_c: f64 = dot_fast(va, vb).max(-1.0).min(1.0) as f64;
     let sin_a: f64 = (1.0 - (cos_a * cos_a)).sqrt();
     let sin_b: f64 = (1.0 - (cos_b * cos_b)).sqrt();
     let sin_c: f64 = (1.0 - (cos_c * cos_c)).sqrt();
@@ -141,9 +141,9 @@ pub fn uniform_sample_spherical_triangle(
     j: f32,
 ) -> Vector {
     // Calculate sines and cosines of the spherical triangle's edge lengths
-    let cos_a: f64 = dot(vb, vc).max(-1.0).min(1.0) as f64;
-    let cos_b: f64 = dot(vc, va).max(-1.0).min(1.0) as f64;
-    let cos_c: f64 = dot(va, vb).max(-1.0).min(1.0) as f64;
+    let cos_a: f64 = dot_fast(vb, vc).max(-1.0).min(1.0) as f64;
+    let cos_b: f64 = dot_fast(vc, va).max(-1.0).min(1.0) as f64;
+    let cos_c: f64 = dot_fast(va, vb).max(-1.0).min(1.0) as f64;
     let sin_a: f64 = (1.0 - (cos_a * cos_a)).sqrt();
     let sin_b: f64 = (1.0 - (cos_b * cos_b)).sqrt();
     let sin_c: f64 = (1.0 - (cos_c * cos_c)).sqrt();
@@ -191,10 +191,10 @@ pub fn uniform_sample_spherical_triangle(
     let q_bottom = ((v * s) + (u * t)) * sin_va;
     let q = q_top / q_bottom;
 
-    let vc_2 =
-        (va * q as f32) + ((vc - (va * dot(vc, va))).normalized() * (1.0 - (q * q)).sqrt() as f32);
+    let vc_2 = (va * q as f32)
+        + ((vc - (va * dot_fast(vc, va))).normalized() * (1.0 - (q * q)).sqrt() as f32);
 
-    let z = 1.0 - (j * (1.0 - dot(vc_2, vb)));
+    let z = 1.0 - (j * (1.0 - dot_fast(vc_2, vb)));
 
-    (vb * z) + ((vc_2 - (vb * dot(vc_2, vb))).normalized() * (1.0 - (z * z)).sqrt())
+    (vb * z) + ((vc_2 - (vb * dot_fast(vc_2, vb))).normalized() * (1.0 - (z * z)).sqrt())
 }

src/scene/assembly.rs

@@ -10,10 +10,9 @@ use crate::{
     color::SpectralSample,
     lerp::lerp_slice,
     light::SurfaceLight,
-    math::{Normal, Point, Transform},
+    math::{Normal, Point, Xform, XformFull},
     shading::SurfaceShader,
     surface::{Surface, SurfaceIntersection},
-    transform_stack::TransformStack,
 };
 
 #[derive(Copy, Clone, Debug)]
@@ -21,7 +20,7 @@ pub struct Assembly<'a> {
     // Instance list
     pub instances: &'a [Instance],
     pub light_instances: &'a [Instance],
-    pub xforms: &'a [Transform],
+    pub xforms: &'a [Xform],
 
     // Surface shader list
     pub surface_shaders: &'a [&'a dyn SurfaceShader],
@@ -45,11 +44,11 @@ impl<'a> Assembly<'a> {
     // Returns (light_color, (sample_point, normal, point_err), pdf, selection_pdf)
     pub fn sample_lights(
         &self,
-        xform_stack: &mut TransformStack,
         n: f32,
         uvw: (f32, f32, f32),
         wavelength: f32,
         time: f32,
+        space: &XformFull,
         intr: &SurfaceIntersection,
     ) -> Option<(SpectralSample, (Point, Normal, f32), f32, f32)> {
         if let SurfaceIntersection::Hit {
@@ -57,46 +56,44 @@ impl<'a> Assembly<'a> {
             closure,
         } = *intr
         {
-            let sel_xform = if !xform_stack.top().is_empty() {
-                lerp_slice(xform_stack.top(), time)
-            } else {
-                Transform::new()
-            };
             if let Some((light_i, sel_pdf, whittled_n)) = self.light_accel.select(
-                idata.incoming * sel_xform,
-                idata.pos * sel_xform,
-                idata.nor * sel_xform,
-                idata.nor_g * sel_xform,
+                idata.incoming.xform_inv(space),
+                idata.pos.xform_inv(space),
+                idata.nor.xform_inv_fast(space),
+                idata.nor_g.xform_inv_fast(space),
                 &closure,
                 time,
                 n,
             ) {
                 let inst = self.light_instances[light_i];
+
+                // Handle transforms.
+                let local_space = if let Some((a, b)) = inst.transform_indices {
+                    if let Some(new_space) = lerp_slice(&self.xforms[a..b], time)
+                        .compose(&space.fwd)
+                        .to_full()
+                    {
+                        new_space
+                    } else {
+                        // Invalid transform.  Give up.
+                        return None;
+                    }
+                } else {
+                    *space
+                };
+
                 match inst.instance_type {
                     InstanceType::Object => {
                         match self.objects[inst.data_index] {
                             Object::SurfaceLight(light) => {
-                                // Get the world-to-object space transform of the light
-                                let xform = if let Some((a, b)) = inst.transform_indices {
-                                    let pxforms = xform_stack.top();
-                                    let xform = lerp_slice(&self.xforms[a..b], time);
-                                    if !pxforms.is_empty() {
-                                        lerp_slice(pxforms, time) * xform
-                                    } else {
-                                        xform
-                                    }
-                                } else {
-                                    let pxforms = xform_stack.top();
-                                    if !pxforms.is_empty() {
-                                        lerp_slice(pxforms, time)
-                                    } else {
-                                        Transform::new()
-                                    }
-                                };
-
                                 // Sample the light
                                 let (color, sample_geo, pdf) = light.sample_from_point(
-                                    &xform, idata.pos, uvw.0, uvw.1, wavelength, time,
+                                    &local_space,
+                                    idata.pos,
+                                    uvw.0,
+                                    uvw.1,
+                                    wavelength,
+                                    time,
                                 );
                                 return Some((color, sample_geo, pdf, sel_pdf));
                             }
@@ -106,27 +103,16 @@ impl<'a> Assembly<'a> {
                     }
 
                     InstanceType::Assembly => {
-                        // Push the world-to-object space transforms of the assembly onto
-                        // the transform stack.
-                        if let Some((a, b)) = inst.transform_indices {
-                            xform_stack.push(&self.xforms[a..b]);
-                        }
-
                         // Sample sub-assembly lights
                         let sample = self.assemblies[inst.data_index].sample_lights(
-                            xform_stack,
                             whittled_n,
                             uvw,
                             wavelength,
                             time,
+                            &local_space,
                             intr,
                         );
 
-                        // Pop the assembly's transforms off the transform stack.
-                        if inst.transform_indices.is_some() {
-                            xform_stack.pop();
-                        }
-
                         // Return sample
                         return sample.map(|(ss, v, pdf, spdf)| (ss, v, pdf, spdf * sel_pdf));
                     }
@@ -152,11 +138,11 @@ pub struct AssemblyBuilder<'a> {
 
     // Instance list
     instances: Vec<Instance>,
-    xforms: Vec<Transform>,
+    xforms: Vec<Xform>,
 
     // Shader list
     surface_shaders: Vec<&'a dyn SurfaceShader>,
-    surface_shader_map: HashMap<String, usize>, // map Name -> Index
+    pub surface_shader_map: HashMap<String, usize>, // map Name -> Index
 
     // Object list
     objects: Vec<Object<'a>>,
@@ -220,12 +206,7 @@ impl<'a> AssemblyBuilder<'a> {
         self.assemblies.push(asmb);
     }
 
-    pub fn add_instance(
-        &mut self,
-        name: &str,
-        surface_shader_name: Option<&str>,
-        xforms: Option<&[Transform]>,
-    ) {
+    pub fn add_instance(&mut self, name: &str, xforms: Option<&[Xform]>) {
         // Make sure name exists
         if !self.name_exists(name) {
             panic!("Attempted to add instance with a name that doesn't exist.");
@@ -247,12 +228,6 @@ impl<'a> AssemblyBuilder<'a> {
             Instance {
                 instance_type: InstanceType::Object,
                 data_index: self.object_map[name],
-                surface_shader_index: surface_shader_name.map(|name| {
-                    *self
-                        .surface_shader_map
-                        .get(name)
-                        .unwrap_or_else(|| panic!("Unknown surface shader '{}'.", name))
-                }),
                 id: self.instances.len(),
                 transform_indices: xforms
                     .map(|xf| (self.xforms.len(), self.xforms.len() + xf.len())),
@@ -261,12 +236,6 @@ impl<'a> AssemblyBuilder<'a> {
             Instance {
                 instance_type: InstanceType::Assembly,
                 data_index: self.assembly_map[name],
-                surface_shader_index: surface_shader_name.map(|name| {
-                    *self
-                        .surface_shader_map
-                        .get(name)
-                        .unwrap_or_else(|| panic!("Unknown surface shader '{}'.", name))
-                }),
                 id: self.instances.len(),
                 transform_indices: xforms
                     .map(|xf| (self.xforms.len(), self.xforms.len() + xf.len())),
@@ -405,7 +374,6 @@ pub enum Object<'a> {
 pub struct Instance {
     pub instance_type: InstanceType,
     pub data_index: usize,
-    pub surface_shader_index: Option<usize>,
     pub id: usize,
     pub transform_indices: Option<(usize, usize)>,
 }

src/scene/mod.rs

@@ -6,9 +6,9 @@ use crate::{
     algorithm::weighted_choice,
     camera::Camera,
     color::SpectralSample,
-    math::{Normal, Point, Vector},
+    light::WorldLightSource,
+    math::{Normal, Point, Vector, XformFull},
     surface::SurfaceIntersection,
-    transform_stack::TransformStack,
 };
 
 pub use self::{
@@ -19,19 +19,19 @@ pub use self::{
 #[derive(Debug)]
 pub struct Scene<'a> {
     pub name: Option<String>,
-    pub camera: Camera<'a>,
-    pub world: World<'a>,
+    pub camera: Camera,
+    pub world: World,
     pub root: Assembly<'a>,
 }
 
 impl<'a> Scene<'a> {
     pub fn sample_lights(
         &self,
-        xform_stack: &mut TransformStack,
         n: f32,
         uvw: (f32, f32, f32),
         wavelength: f32,
         time: f32,
+        space: &XformFull,
         intr: &SurfaceIntersection,
     ) -> SceneLightSample {
         // TODO: this just selects between world lights and local lights
@@ -68,7 +68,7 @@ impl<'a> Scene<'a> {
             if n < wl_prob {
                 // World lights
                 let n = n / wl_prob;
-                let (i, p) = weighted_choice(self.world.lights, n, |l| l.approximate_energy());
+                let (i, p) = weighted_choice(&self.world.lights, n, |l| l.approximate_energy());
                 let (ss, sv, pdf) =
                     self.world.lights[i].sample_from_point(uvw.0, uvw.1, wavelength, time);
                 return SceneLightSample::Distant {
@@ -81,9 +81,9 @@ impl<'a> Scene<'a> {
                 // Local lights
                 let n = (n - wl_prob) / (1.0 - wl_prob);
 
-                if let Some((ss, sgeo, pdf, spdf)) =
-                    self.root
-                        .sample_lights(xform_stack, n, uvw, wavelength, time, intr)
+                if let Some((ss, sgeo, pdf, spdf)) = self
+                    .root
+                    .sample_lights(n, uvw, wavelength, time, space, intr)
                 {
                     return SceneLightSample::Surface {
                         color: ss,

src/scene/world.rs

@@ -1,7 +1,7 @@
-use crate::{color::Color, light::WorldLightSource};
+use crate::{color::Color, light::DistantDiskLight};
 
 #[derive(Debug)]
-pub struct World<'a> {
+pub struct World {
     pub background_color: Color,
-    pub lights: &'a [&'a dyn WorldLightSource],
+    pub lights: Vec<DistantDiskLight>,
 }

src/scramble.rs

@@ -0,0 +1,101 @@
+#![allow(dead_code)]
+
+/// Performs a base-2 Owen scramble on an integer.
+pub fn owen2(n: u32, seed: u32) -> u32 {
+    // Multiply by a large random prime and xor by a random number.
+    // This is to ensure that the seed doesn't behave poorly with
+    // e.g. incrementing parameters, and also that zero doesn't
+    // map to zero in the hash function.
+    let seed = seed.wrapping_mul(0x68318d2f) ^ 0x5adbc2a7;
+
+    let mut result = n;
+    for i in 0..32 {
+        result ^= hash((n & (!1 << i)) ^ seed) & (1 << i);
+    }
+
+    result
+}
+
+#[inline(always)]
+pub fn owen4_fast(mut n: u32, seed: u32) -> u32 {
+    let scramble = hash(seed);
+
+    n = n.reverse_bits();
+
+    n ^= n.wrapping_mul(0x3d20adea);
+    n ^= (n >> 1) & (n << 1) & 0x55555555;
+    n = n.wrapping_add(scramble);
+    n = n.wrapping_mul((scramble >> 16) | 1);
+    n ^= (n >> 1) & (n << 1) & 0x55555555;
+    n ^= n.wrapping_mul(0x05526c56);
+    n ^= n.wrapping_mul(0x53a22864);
+
+    n.reverse_bits()
+}
+
+pub fn owen4(n: u32, seed: u32) -> u32 {
+    // Bit-packed permutation table.
+    const PERMUTATION_TABLE: [u8; 24] = [
+        0 | (1 << 2) | (2 << 4) | (3 << 6), // [0, 1, 2, 3],
+        0 | (1 << 2) | (3 << 4) | (2 << 6), // [0, 1, 3, 2],
+        0 | (2 << 2) | (1 << 4) | (3 << 6), // [0, 2, 1, 3],
+        0 | (2 << 2) | (3 << 4) | (1 << 6), // [0, 2, 3, 1],
+        0 | (3 << 2) | (1 << 4) | (2 << 6), // [0, 3, 1, 2],
+        0 | (3 << 2) | (2 << 4) | (1 << 6), // [0, 3, 2, 1],
+        1 | (0 << 2) | (2 << 4) | (3 << 6), // [1, 0, 2, 3],
+        1 | (0 << 2) | (3 << 4) | (2 << 6), // [1, 0, 3, 2],
+        1 | (2 << 2) | (0 << 4) | (3 << 6), // [1, 2, 0, 3],
+        1 | (2 << 2) | (3 << 4) | (0 << 6), // [1, 2, 3, 0],
+        1 | (3 << 2) | (0 << 4) | (2 << 6), // [1, 3, 0, 2],
+        1 | (3 << 2) | (2 << 4) | (0 << 6), // [1, 3, 2, 0],
+        2 | (0 << 2) | (1 << 4) | (3 << 6), // [2, 0, 1, 3],
+        2 | (0 << 2) | (3 << 4) | (1 << 6), // [2, 0, 3, 1],
+        2 | (1 << 2) | (0 << 4) | (3 << 6), // [2, 1, 0, 3],
+        2 | (1 << 2) | (3 << 4) | (0 << 6), // [2, 1, 3, 0],
+        2 | (3 << 2) | (0 << 4) | (1 << 6), // [2, 3, 0, 1],
+        2 | (3 << 2) | (1 << 4) | (0 << 6), // [2, 3, 1, 0],
+        3 | (0 << 2) | (1 << 4) | (2 << 6), // [3, 0, 1, 2],
+        3 | (0 << 2) | (2 << 4) | (1 << 6), // [3, 0, 2, 1],
+        3 | (1 << 2) | (0 << 4) | (2 << 6), // [3, 1, 0, 2],
+        3 | (1 << 2) | (2 << 4) | (0 << 6), // [3, 1, 2, 0],
+        3 | (2 << 2) | (0 << 4) | (1 << 6), // [3, 2, 0, 1],
+        3 | (2 << 2) | (1 << 4) | (0 << 6), // [3, 2, 1, 0],
+    ];
+
+    // Multiply by a large random prime and xor by a random number.
+    // This is to ensure that the seed doesn't behave poorly with
+    // e.g. incrementing parameters, and also that zero doesn't
+    // map to zero in the hash function.
+    let seed = seed.wrapping_mul(0xe8559dcb) ^ 0x372fcdb9;
+
+    let mut result = 0;
+    for i in 0..16 {
+        let mask = !0b11 << (i * 2);
+        let perm_entry = PERMUTATION_TABLE[
+            // The xor with `i` is to ensure runs of zeros in `n` still
+            // result in different shuffles on each iteration.  `i` is
+            // shifted to avoid interacting poorly with an incrementing
+            // `n`.
+            (hash((n & mask) ^ seed ^ (i << 16)) % 24) as usize
+        ];
+        let perm_cell_idx = ((n >> (i * 2)) & 0b11) as usize;
+        result |= (((perm_entry >> (perm_cell_idx * 2)) & 0b11) as u32) << (i * 2);
+    }
+
+    result
+}
+
+//-------------------------------------------------------------
+
+/// Fast bit-mixing hash for use in the functions above.
+#[inline(always)]
+pub fn hash(mut n: u32) -> u32 {
+    // From https://github.com/skeeto/hash-prospector
+    n ^= n >> 16;
+    n = n.wrapping_mul(0x21f0aaad);
+    n ^= n >> 15;
+    n = n.wrapping_mul(0xd35a2d97);
+    n ^= n >> 15;
+
+    n
+}

src/shading/surface_closure.rs

@@ -2,12 +2,10 @@
 
 use std::f32::consts::PI as PI_32;
 
-use glam::Vec4;
-
 use crate::{
     color::{Color, SpectralSample},
     lerp::{lerp, Lerp},
-    math::{dot, zup_to_vec, Normal, Vector},
+    math::{clamp, dot_fast, zup_to_vec, Float4, Normal, Vector},
     sampling::cosine_sample_hemisphere,
 };
 
@@ -289,7 +287,7 @@ mod lambert_closure {
         uv: (f32, f32),
         wavelength: f32,
     ) -> (Vector, SpectralSample, f32) {
-        let (nn, flipped_nor_g) = if dot(nor_g.into_vector(), inc) <= 0.0 {
+        let (nn, flipped_nor_g) = if dot_fast(nor_g.into_vector(), inc) <= 0.0 {
             (nor.normalized().into_vector(), nor_g.into_vector())
         } else {
             (-nor.normalized().into_vector(), -nor_g.into_vector())
@@ -302,7 +300,7 @@ mod lambert_closure {
         let out = zup_to_vec(dir, nn);
 
         // Make sure it's not on the wrong side of the geometric normal.
-        if dot(flipped_nor_g, out) >= 0.0 {
+        if dot_fast(flipped_nor_g, out) >= 0.0 {
             (out, color.to_spectral_sample(wavelength) * pdf, pdf)
         } else {
             (out, SpectralSample::new(0.0), 0.0)
@@ -317,14 +315,14 @@ mod lambert_closure {
         nor_g: Normal,
         wavelength: f32,
     ) -> (SpectralSample, f32) {
-        let (nn, flipped_nor_g) = if dot(nor_g.into_vector(), inc) <= 0.0 {
+        let (nn, flipped_nor_g) = if dot_fast(nor_g.into_vector(), inc) <= 0.0 {
             (nor.normalized().into_vector(), nor_g.into_vector())
         } else {
             (-nor.normalized().into_vector(), -nor_g.into_vector())
         };
 
-        if dot(flipped_nor_g, out) >= 0.0 {
-            let fac = dot(nn, out.normalized()).max(0.0) * INV_PI;
+        if dot_fast(flipped_nor_g, out) >= 0.0 {
+            let fac = dot_fast(nn, out.normalized()).max(0.0) * INV_PI;
             (color.to_spectral_sample(wavelength) * fac, fac)
         } else {
             (SpectralSample::new(0.0), 0.0)
@@ -383,14 +381,14 @@ mod lambert_closure {
             let cos_theta_max = (1.0 - sin_theta_max2).sqrt();
 
             let v = to_light_center.normalized();
-            let nn = if dot(nor_g.into_vector(), inc) <= 0.0 {
+            let nn = if dot_fast(nor_g.into_vector(), inc) <= 0.0 {
                 nor.normalized()
             } else {
                 -nor.normalized()
             }
             .into_vector();
 
-            let cos_nv = dot(nn, v).max(-1.0).min(1.0);
+            let cos_nv = dot_fast(nn, v).max(-1.0).min(1.0);
 
             // Alt implementation from the SPI paper.
             // Worse sampling, but here for reference.
@@ -428,7 +426,7 @@ mod ggx_closure {
         wavelength: f32,
     ) -> (Vector, SpectralSample, f32) {
         // Get normalized surface normal
-        let (nn, flipped_nor_g) = if dot(nor_g.into_vector(), inc) <= 0.0 {
+        let (nn, flipped_nor_g) = if dot_fast(nor_g.into_vector(), inc) <= 0.0 {
             (nor.normalized().into_vector(), nor_g.into_vector())
         } else {
             (-nor.normalized().into_vector(), -nor_g.into_vector())
@@ -442,10 +440,10 @@ mod ggx_closure {
         let mut half_dir = Vector::new(angle.cos() * theta_sin, angle.sin() * theta_sin, theta_cos);
         half_dir = zup_to_vec(half_dir, nn).normalized();
 
-        let out = inc - (half_dir * 2.0 * dot(inc, half_dir));
+        let out = inc - (half_dir * 2.0 * dot_fast(inc, half_dir));
 
         // Make sure it's not on the wrong side of the geometric normal.
-        if dot(flipped_nor_g, out) >= 0.0 {
+        if dot_fast(flipped_nor_g, out) >= 0.0 {
             let (filter, pdf) = evaluate(col, roughness, fresnel, inc, out, nor, nor_g, wavelength);
             (out, filter, pdf)
         } else {
@@ -469,23 +467,23 @@ mod ggx_closure {
         let hh = (aa + bb).normalized(); // Half-way between aa and bb
 
         // Surface normal
-        let (nn, flipped_nor_g) = if dot(nor_g.into_vector(), inc) <= 0.0 {
+        let (nn, flipped_nor_g) = if dot_fast(nor_g.into_vector(), inc) <= 0.0 {
             (nor.normalized().into_vector(), nor_g.into_vector())
         } else {
             (-nor.normalized().into_vector(), -nor_g.into_vector())
         };
 
         // Make sure everything's on the correct side of the surface
-        if dot(nn, aa) < 0.0 || dot(nn, bb) < 0.0 || dot(flipped_nor_g, bb) < 0.0 {
+        if dot_fast(nn, aa) < 0.0 || dot_fast(nn, bb) < 0.0 || dot_fast(flipped_nor_g, bb) < 0.0 {
             return (SpectralSample::new(0.0), 0.0);
         }
 
         // Calculate needed dot products
-        let na = dot(nn, aa).clamp(-1.0, 1.0);
-        let nb = dot(nn, bb).clamp(-1.0, 1.0);
-        let ha = dot(hh, aa).clamp(-1.0, 1.0);
-        let hb = dot(hh, bb).clamp(-1.0, 1.0);
-        let nh = dot(nn, hh).clamp(-1.0, 1.0);
+        let na = clamp(dot_fast(nn, aa), -1.0, 1.0);
+        let nb = clamp(dot_fast(nn, bb), -1.0, 1.0);
+        let ha = clamp(dot_fast(hh, aa), -1.0, 1.0);
+        let hb = clamp(dot_fast(hh, bb), -1.0, 1.0);
+        let nh = clamp(dot_fast(nn, hh), -1.0, 1.0);
 
         // Calculate F - Fresnel
         let col_f = {
@@ -512,7 +510,7 @@ mod ggx_closure {
                 rev_fresnel,
             );
 
-            SpectralSample::from_parts(Vec4::new(c0, c1, c2, c3), wavelength)
+            SpectralSample::from_parts(Float4::new(c0, c1, c2, c3), wavelength)
         };
 
         // Calculate everything else
@@ -556,7 +554,7 @@ mod ggx_closure {
         assert!(cos_theta_max <= 1.0);
 
         // Surface normal
-        let nn = if dot(nor.into_vector(), inc) < 0.0 {
+        let nn = if dot_fast(nor.into_vector(), inc) < 0.0 {
             nor.normalized()
         } else {
             -nor.normalized() // If back-facing, flip normal
@@ -574,9 +572,9 @@ mod ggx_closure {
         //    let vv = Halton::sample(1, i);
         //    let mut samp = uniform_sample_cone(uu, vv, cos_theta_max);
         //    samp = zup_to_vec(samp, bb).normalized();
-        //    if dot(nn, samp) > 0.0 {
+        //    if dot_fast(nn, samp) > 0.0 {
         //        let hh = (aa+samp).normalized();
-        //        fac += ggx_d(dot(nn, hh), roughness);
+        //        fac += ggx_d(dot_fast(nn, hh), roughness);
         //    }
         //}
         //fac /= N * N;
@@ -584,7 +582,7 @@ mod ggx_closure {
         // Approximate method
         let theta = cos_theta_max.acos();
         let hh = (aa + bb).normalized();
-        let nh = dot(nn, hh).clamp(-1.0, 1.0);
+        let nh = clamp(dot_fast(nn, hh), -1.0, 1.0);
         let fac = ggx_d(nh, (1.0f32).min(roughness.sqrt() + (2.0 * theta / PI_32)));
 
         fac * (1.0f32).min(1.0 - cos_theta_max) * INV_PI

src/space_fill.rs

@@ -0,0 +1,216 @@
+//! Space-filling curves and other related functionality.
+
+#![allow(dead_code)]
+
+pub mod hilbert {
+    const N: u32 = 1 << 16;
+
+    /// Convert (x,y) to hilbert curve index.
+    ///
+    /// x: The x coordinate.  Must be no greater than 2^16-1.
+    /// y: The y coordinate.  Must be no greater than 2^16-1.
+    /// n: Basically the "resolution" of the curve, on one side.
+    ///
+    /// Returns the hilbert curve index corresponding to the (x,y) coordinates given.
+    pub fn encode(x: u32, y: u32, n: u32) -> u32 {
+        assert!(x < N);
+        assert!(y < N);
+
+        let (mut x, mut y) = (x, y);
+        let mut d = 0;
+        let mut s = n >> 1;
+        while s > 0 {
+            let rx = if (x & s) > 0 { 1 } else { 0 };
+            let ry = if (y & s) > 0 { 1 } else { 0 };
+            d += s * s * ((3 * rx) ^ ry);
+            (x, y) = hilbert_rotate(s, rx, ry, x, y);
+
+            s >>= 1
+        }
+
+        d
+    }
+
+    /// Convert hilbert curve index to (x,y).
+    ///
+    /// d: The hilbert curve index.
+    /// n: Basically the "resolution" of the curve, on one side.
+    ///
+    /// Returns the (x, y) coords at the given index.
+    pub fn decode(d: u32, n: u32) -> (u32, u32) {
+        let (mut x, mut y) = (0, 0);
+        let mut s = 1;
+        let mut t = d;
+        while s < n {
+            let rx = 1 & (t >> 1);
+            let ry = 1 & (t ^ rx);
+            (x, y) = hilbert_rotate(s, rx, ry, x, y);
+            x += s * rx;
+            y += s * ry;
+            t >>= 2;
+
+            s <<= 1;
+        }
+
+        (x, y)
+    }
+
+    //------------
+    // Utilities.
+
+    fn hilbert_rotate(n: u32, rx: u32, ry: u32, x: u32, y: u32) -> (u32, u32) {
+        if ry == 0 {
+            if rx == 1 {
+                ((n - 1).wrapping_sub(y), (n - 1).wrapping_sub(x))
+            } else {
+                (y, x)
+            }
+        } else {
+            (x, y)
+        }
+    }
+}
+
+pub mod morton {
+    const N: u32 = 1 << 16;
+
+    /// Convert (x,y) to morton curve index.
+    ///
+    /// x: The x coordinate.  Should be no greater than 2^16-1.
+    /// y: The y coordinate.  Should be no greater than 2^16-1.
+    ///
+    /// Returns the morton curve index corresponding to the (x,y) coordinates given.
+    pub fn encode(x: u32, y: u32) -> u32 {
+        debug_assert!(x < N);
+        debug_assert!(y < N);
+
+        part_1_by_1(x) | (part_1_by_1(y) << 1)
+    }
+
+    /// Convert morton curve index to (x,y).
+    ///
+    /// i: The morton curve index.
+    ///
+    /// Returns the (x, y) coords at the given index.
+    pub fn decode(i: u32) -> (u32, u32) {
+        (compact_1_by_1(i), compact_1_by_1(i >> 1))
+    }
+
+    //------------
+    // Utilities.
+
+    #[inline(always)]
+    fn part_1_by_1(mut x: u32) -> u32 {
+        x &= 0x0000ffff;
+        x = (x ^ (x << 8)) & 0x00ff00ff;
+        x = (x ^ (x << 4)) & 0x0f0f0f0f;
+        x = (x ^ (x << 2)) & 0x33333333;
+        x = (x ^ (x << 1)) & 0x55555555;
+        x
+    }
+
+    #[inline(always)]
+    fn compact_1_by_1(mut x: u32) -> u32 {
+        x &= 0x55555555;
+        x = (x ^ (x >> 1)) & 0x33333333;
+        x = (x ^ (x >> 2)) & 0x0f0f0f0f;
+        x = (x ^ (x >> 4)) & 0x00ff00ff;
+        x = (x ^ (x >> 8)) & 0x0000ffff;
+        x
+    }
+}
+
+/// Yields coordinates in outward spiral, but incorporating a Hilbert
+/// curve at the smaller scales.
+pub mod hilbert_spiral {
+    /// Convert from hilbert-spiral index to (x,y).
+    ///
+    /// Note: this returns both negative and positive coordinates.
+    /// It starts at 0,0 and spirals outwards.
+    ///
+    /// i: The hilbert-spiral index.
+    /// hilbert_size: the size of the hulbert blocks on a side.  Will be
+    ///               rounded down to the nearest power of two.
+    ///
+    /// Returns the (x, y) coords at the given index.
+    pub fn decode(i: u32, hilbert_size: u32) -> (i32, i32) {
+        assert!(hilbert_size > 0);
+        let hilbert_size = 1 << (31 - u32::leading_zeros(hilbert_size));
+
+        let hilbert_cells = hilbert_size * hilbert_size;
+
+        let hilbert_i = i % hilbert_cells;
+        let spiral_i = i / hilbert_cells;
+
+        let (mut sx, mut sy, section) = decode_spiral(spiral_i);
+        sx = (sx * hilbert_size as i32) - (hilbert_size / 2) as i32;
+        sy = (sy * hilbert_size as i32) - (hilbert_size / 2) as i32;
+
+        let (hx, hy) = {
+            let (hx, hy) = super::hilbert::decode(hilbert_i, hilbert_size);
+            let a = hilbert_size - 1;
+            match section {
+                0 => (hy, hx),
+                1 => (a - hx, a - hy),
+                2 => (a - hy, a - hx),
+                3 => (hx, hy),
+                _ => unreachable!(),
+            }
+        };
+
+        (sx + hx as i32, sy + hy as i32)
+    }
+
+    pub fn decode_spiral(i: u32) -> (i32, i32, u32) {
+        if i == 0 {
+            return (0, 0, 3);
+        }
+
+        // 0 = first ring outside of center, 1 = second, and so on.
+        let ring = (((i as f64).sqrt() - 1.0) / 2.0) as u32;
+
+        // The size of the ring along one side.
+        let size = 1 + ((ring + 1) * 2);
+
+        let n = i - ((size - 2) * (size - 2)); // The zero-indexed cell of the ring.
+        let arm = n / (size - 1); // The arm of the ring.
+        let arm_n = n % (size - 1); // The index within the arm of the ring.
+
+        // The two coordinates.  They just need to be flipped around depending on the arm.
+        let radius = ring as i32 + 1;
+        let d = -(size as i32 / 2) + 1 + arm_n as i32;
+
+        match arm {
+            0 => (radius, d, 0),
+            1 => (-d, radius, if arm_n == (size - 2) { 2 } else { 1 }),
+            2 => (-radius, -d, 2),
+            3 => (d, -radius, 3),
+            _ => unreachable!(),
+        }
+    }
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn hilbert_reversible() {
+        let i = 0x4c8587a2;
+        let (x, y) = hilbert::decode(i, 1 << 16);
+        let i2 = hilbert::encode(x, y, 1 << 16);
+
+        assert_eq!(i, i2);
+    }
+
+    #[test]
+    fn morton_reversible() {
+        let i = 0x4c8587a2;
+        let (x, y) = morton::decode(i);
+        let i2 = morton::encode(x, y);
+
+        assert_eq!(i, i2);
+    }
+}

src/surface/micropoly_batch.rs

@@ -9,7 +9,7 @@ use crate::{
     bbox::BBox,
     boundable::Boundable,
     lerp::lerp_slice,
-    math::{cross, dot, Normal, Point, Transform},
+    math::{cross, dot, Normal, Point, Xform},
     ray::{RayBatch, RayStack},
     shading::SurfaceClosure,
 };
@@ -150,13 +150,13 @@ impl<'a> MicropolyBatch<'a> {
         rays: &mut RayBatch,
         ray_stack: &mut RayStack,
         isects: &mut [SurfaceIntersection],
-        space: &[Transform],
+        space: &[Xform],
     ) {
         // Precalculate transform for non-motion blur cases
         let static_mat_space = if space.len() == 1 {
-            lerp_slice(space, 0.0).inverse()
+            space[0]
         } else {
-            Transform::new()
+            Xform::identity()
         };
 
         self.accel
@@ -182,11 +182,11 @@ impl<'a> MicropolyBatch<'a> {
                             );
                             if !space.is_empty() {
                                 (*tri_cache[i].as_mut_ptr()).0 =
-                                    (*tri_cache[i].as_mut_ptr()).0 * static_mat_space;
+                                    (*tri_cache[i].as_mut_ptr()).0.xform(&static_mat_space);
                                 (*tri_cache[i].as_mut_ptr()).1 =
-                                    (*tri_cache[i].as_mut_ptr()).1 * static_mat_space;
+                                    (*tri_cache[i].as_mut_ptr()).1.xform(&static_mat_space);
                                 (*tri_cache[i].as_mut_ptr()).2 =
-                                    (*tri_cache[i].as_mut_ptr()).2 * static_mat_space;
+                                    (*tri_cache[i].as_mut_ptr()).2.xform(&static_mat_space);
                             }
                         }
                     }
@@ -205,7 +205,7 @@ impl<'a> MicropolyBatch<'a> {
                     // Calculate the ray space, if necessary.
                     let mat_space = if space.len() > 1 {
                         // Per-ray transform, for motion blur
-                        lerp_slice(space, ray_time).inverse()
+                        lerp_slice(space, ray_time)
                     } else {
                         static_mat_space
                     };
@@ -251,9 +251,9 @@ impl<'a> MicropolyBatch<'a> {
                             };
 
                             if !space.is_empty() {
-                                tri.0 = tri.0 * mat_space;
-                                tri.1 = tri.1 * mat_space;
-                                tri.2 = tri.2 * mat_space;
+                                tri.0 = tri.0.xform(&mat_space);
+                                tri.1 = tri.1.xform(&mat_space);
+                                tri.2 = tri.2.xform(&mat_space);
                             }
 
                             tri
@@ -284,6 +284,13 @@ impl<'a> MicropolyBatch<'a> {
 
                     // Calculate intersection data if necessary.
                     if non_shadow_hit {
+                        // Get the full space data.
+                        let mat_space = if let Some(space) = mat_space.to_full() {
+                            space
+                        } else {
+                            return;
+                        };
+
                         let hit_tri = unsafe { hit_tri.assume_init() };
                         let hit_tri_indices = unsafe { hit_tri_indices.assume_init() };
                         let (t, b0, b1, b2) = unsafe { hit_tri_data.assume_init() };
@@ -311,7 +318,7 @@ impl<'a> MicropolyBatch<'a> {
                             let n1 = lerp_slice(n1_slice, ray_time).normalized();
                             let n2 = lerp_slice(n2_slice, ray_time).normalized();
 
-                            let s_nor = ((n0 * b0) + (n1 * b1) + (n2 * b2)) * mat_space;
+                            let s_nor = ((n0 * b0) + (n1 * b1) + (n2 * b2)).xform_fast(&mat_space);
                             if dot(s_nor, geo_normal) >= 0.0 {
                                 s_nor
                             } else {

src/surface/mod.rs

@@ -2,7 +2,6 @@
 
 // pub mod micropoly_batch;
 pub mod bilinear_patch;
-pub mod micropoly_batch;
 pub mod triangle;
 pub mod triangle_mesh;
 
@@ -10,8 +9,8 @@ use std::fmt::Debug;
 
 use crate::{
     boundable::Boundable,
-    math::{Normal, Point, Transform, Vector},
-    ray::{RayBatch, RayStack},
+    math::{Normal, Point, Vector, XformFull},
+    ray::{LocalRay, Ray},
     shading::surface_closure::SurfaceClosure,
     shading::SurfaceShader,
 };
@@ -19,13 +18,13 @@ use crate::{
 const MAX_EDGE_DICE: u32 = 128;
 
 pub trait Surface: Boundable + Debug + Sync {
-    fn intersect_rays(
+    fn intersect_ray(
         &self,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
-        isects: &mut [SurfaceIntersection],
-        shader: &dyn SurfaceShader,
-        space: &[Transform],
+        ray: &mut Ray,
+        local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
+        shaders: &[&dyn SurfaceShader],
     );
 }
 
@@ -80,13 +79,13 @@ pub enum SurfaceIntersection {
 
 #[derive(Debug, Copy, Clone)]
 pub struct SurfaceIntersectionData {
-    pub incoming: Vector, // Direction of the incoming ray
-    pub pos: Point,       // Position of the intersection
+    pub incoming: Vector, // Direction of the incoming ray.
+    pub pos: Point,       // Position of the intersection.
     pub pos_err: f32,     // Error magnitude of the intersection position.  Imagine
     // a cube centered around `pos` with dimensions of `2 * pos_err`.
-    pub nor: Normal,            // Shading normal
-    pub nor_g: Normal,          // True geometric normal
-    pub local_space: Transform, // Matrix from global space to local space
-    pub t: f32,                 // Ray t-value at the intersection point
-    pub sample_pdf: f32,        // The PDF of getting this point by explicitly sampling the surface
+    pub nor: Normal,            // Shading normal.
+    pub nor_g: Normal,          // True geometric normal.
+    pub local_space: XformFull, // Matrix from local to world space.
+    pub t: f32,                 // Ray t-value at the intersection point.
+    pub sample_pdf: f32,        // The PDF of getting this point by explicitly sampling the surface.
 }

src/surface/triangle.rs

@@ -162,7 +162,7 @@ pub fn surface_point(tri: (Point, Point, Point), bary: (f32, f32, f32)) -> (Poin
         + (tri.1.into_vector().abs() * bary.1)
         + (tri.2.into_vector().abs() * bary.2))
         * fp_gamma(7))
-    .co
+    .0
     .max_element();
 
     (pos, pos_err)

src/surface/triangle_mesh.rs

@@ -6,10 +6,11 @@ use crate::{
     accel::BVH4,
     bbox::BBox,
     boundable::Boundable,
+    color::Color,
     lerp::lerp_slice,
-    math::{cross, dot, Normal, Point, Transform},
-    ray::{RayBatch, RayStack},
-    shading::SurfaceShader,
+    math::{cross, dot, Normal, Point, XformFull},
+    ray::{LocalRay, Ray},
+    shading::{SimpleSurfaceShader, SurfaceShader},
 };
 
 use super::{triangle, Surface, SurfaceIntersection, SurfaceIntersectionData};
@@ -18,6 +19,7 @@ const MAX_LEAF_TRIANGLE_COUNT: usize = 3;
 
 #[derive(Copy, Clone, Debug)]
 pub struct TriangleMesh<'a> {
+    pub shader_idx: Option<usize>,
     time_sample_count: usize,
     vertices: &'a [Point], // Vertices, with the time samples for each vertex stored contiguously
     normals: Option<&'a [Normal]>, // Vertex normals, organized the same as `vertices`
@@ -28,6 +30,7 @@ pub struct TriangleMesh<'a> {
 impl<'a> TriangleMesh<'a> {
     pub fn from_verts_and_indices<'b>(
         arena: &'b Arena,
+        shader_idx: Option<usize>,
         verts: &[Vec<Point>],
         vert_normals: &Option<Vec<Vec<Normal>>>,
         tri_indices: &[(usize, usize, usize)],
@@ -106,6 +109,7 @@ impl<'a> TriangleMesh<'a> {
         });
 
         TriangleMesh {
+            shader_idx: shader_idx,
             time_sample_count: time_sample_count,
             vertices: vertices,
             normals: normals,
@@ -122,202 +126,135 @@ impl<'a> Boundable for TriangleMesh<'a> {
 }
 
 impl<'a> Surface for TriangleMesh<'a> {
-    fn intersect_rays(
+    fn intersect_ray(
         &self,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
-        isects: &mut [SurfaceIntersection],
-        shader: &dyn SurfaceShader,
-        space: &[Transform],
+        ray: &mut Ray,
+        local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
+        shaders: &[&dyn SurfaceShader],
     ) {
-        // Precalculate transform for non-motion blur cases
-        let static_mat_space = if space.len() == 1 {
-            lerp_slice(space, 0.0).inverse()
-        } else {
-            Transform::new()
+        let unassigned_shader = SimpleSurfaceShader::Emit {
+            color: Color::new_xyz(color::rec709_to_xyz((1.0, 0.0, 1.0))),
         };
 
-        self.accel
-            .traverse(rays, ray_stack, |idx_range, rays, ray_stack| {
-                let tri_count = idx_range.end - idx_range.start;
+        let shader = if let Some(idx) = self.shader_idx {
+            shaders[idx]
+        } else {
+            &unassigned_shader
+        };
 
-                // Build the triangle cache if we can!
-                let is_cached = ray_stack.ray_count_in_next_task() >= tri_count
-                    && self.time_sample_count == 1
-                    && space.len() <= 1;
-                let mut tri_cache = [std::mem::MaybeUninit::uninit(); MAX_LEAF_TRIANGLE_COUNT];
-                if is_cached {
-                    for tri_idx in idx_range.clone() {
-                        let i = tri_idx - idx_range.start;
-                        let tri_indices = self.indices[tri_idx];
+        self.accel.traverse(ray, local_ray, |idx_range, ray| {
+            // Iterate through the triangles and test the ray against them.
+            let mut non_shadow_hit = false;
+            let mut hit_tri = std::mem::MaybeUninit::uninit();
+            let mut hit_tri_indices = std::mem::MaybeUninit::uninit();
+            let mut hit_tri_data = std::mem::MaybeUninit::uninit();
+            let ray_pre = triangle::RayTriPrecompute::new(ray.dir);
+            for tri_idx in idx_range.clone() {
+                let tri_indices = self.indices[tri_idx];
 
-                        // For static triangles with static transforms, cache them.
+                // Get triangle.
+                let mut tri = if self.time_sample_count == 1 {
+                    // No deformation motion blur, so fast-path it.
+                    (
+                        self.vertices[tri_indices.0 as usize],
+                        self.vertices[tri_indices.1 as usize],
+                        self.vertices[tri_indices.2 as usize],
+                    )
+                } else {
+                    // Deformation motion blur, need to interpolate.
+                    let p0_slice = &self.vertices[(tri_indices.0 as usize * self.time_sample_count)
+                        ..((tri_indices.0 as usize + 1) * self.time_sample_count)];
+                    let p1_slice = &self.vertices[(tri_indices.1 as usize * self.time_sample_count)
+                        ..((tri_indices.1 as usize + 1) * self.time_sample_count)];
+                    let p2_slice = &self.vertices[(tri_indices.2 as usize * self.time_sample_count)
+                        ..((tri_indices.2 as usize + 1) * self.time_sample_count)];
+
+                    let p0 = lerp_slice(p0_slice, ray.time);
+                    let p1 = lerp_slice(p1_slice, ray.time);
+                    let p2 = lerp_slice(p2_slice, ray.time);
+
+                    (p0, p1, p2)
+                };
+
+                // Transform triangle into world space.
+                tri.0 = tri.0.xform(space);
+                tri.1 = tri.1.xform(space);
+                tri.2 = tri.2.xform(space);
+
+                // Test ray against triangle
+                if let Some((t, b0, b1, b2)) =
+                    triangle::intersect_ray(ray.orig, ray_pre, ray.max_t, tri)
+                {
+                    if ray.is_occlusion() {
+                        *isect = SurfaceIntersection::Occlude;
+                        ray.mark_done();
+                        break;
+                    } else {
+                        non_shadow_hit = true;
+                        ray.max_t = t;
                         unsafe {
-                            *tri_cache[i].as_mut_ptr() = (
-                                self.vertices[tri_indices.0 as usize],
-                                self.vertices[tri_indices.1 as usize],
-                                self.vertices[tri_indices.2 as usize],
-                            );
-                            if !space.is_empty() {
-                                (*tri_cache[i].as_mut_ptr()).0 =
-                                    (*tri_cache[i].as_mut_ptr()).0 * static_mat_space;
-                                (*tri_cache[i].as_mut_ptr()).1 =
-                                    (*tri_cache[i].as_mut_ptr()).1 * static_mat_space;
-                                (*tri_cache[i].as_mut_ptr()).2 =
-                                    (*tri_cache[i].as_mut_ptr()).2 * static_mat_space;
-                            }
+                            *hit_tri.as_mut_ptr() = tri;
+                            *hit_tri_indices.as_mut_ptr() = tri_indices;
+                            *hit_tri_data.as_mut_ptr() = (t, b0, b1, b2);
                         }
                     }
                 }
+            }
 
-                // Test each ray against the triangles.
-                ray_stack.do_next_task(|ray_idx| {
-                    let ray_idx = ray_idx as usize;
+            // Calculate intersection data if necessary.
+            if non_shadow_hit {
+                let hit_tri = unsafe { hit_tri.assume_init() };
+                let (t, b0, b1, b2) = unsafe { hit_tri_data.assume_init() };
 
-                    if rays.is_done(ray_idx) {
-                        return;
-                    }
+                // Calculate intersection point and error magnitudes
+                let (pos, pos_err) = triangle::surface_point(hit_tri, (b0, b1, b2));
 
-                    let ray_time = rays.time(ray_idx);
+                // Calculate geometric surface normal
+                let geo_normal = cross(hit_tri.0 - hit_tri.1, hit_tri.0 - hit_tri.2).into_normal();
 
-                    // Calculate the ray space, if necessary.
-                    let mat_space = if space.len() > 1 {
-                        // Per-ray transform, for motion blur
-                        lerp_slice(space, ray_time).inverse()
+                // Calculate interpolated surface normal, if any
+                let shading_normal = if let Some(normals) = self.normals {
+                    let hit_tri_indices = unsafe { hit_tri_indices.assume_init() };
+                    let n0_slice = &normals[(hit_tri_indices.0 as usize * self.time_sample_count)
+                        ..((hit_tri_indices.0 as usize + 1) * self.time_sample_count)];
+                    let n1_slice = &normals[(hit_tri_indices.1 as usize * self.time_sample_count)
+                        ..((hit_tri_indices.1 as usize + 1) * self.time_sample_count)];
+                    let n2_slice = &normals[(hit_tri_indices.2 as usize * self.time_sample_count)
+                        ..((hit_tri_indices.2 as usize + 1) * self.time_sample_count)];
+
+                    let n0 = lerp_slice(n0_slice, ray.time).normalized();
+                    let n1 = lerp_slice(n1_slice, ray.time).normalized();
+                    let n2 = lerp_slice(n2_slice, ray.time).normalized();
+
+                    let s_nor = ((n0 * b0) + (n1 * b1) + (n2 * b2)).xform_fast(&space);
+                    if dot(s_nor, geo_normal) >= 0.0 {
+                        s_nor
                     } else {
-                        static_mat_space
-                    };
-
-                    // Iterate through the triangles and test the ray against them.
-                    let mut non_shadow_hit = false;
-                    let mut hit_tri = std::mem::MaybeUninit::uninit();
-                    let mut hit_tri_indices = std::mem::MaybeUninit::uninit();
-                    let mut hit_tri_data = std::mem::MaybeUninit::uninit();
-                    let ray_pre = triangle::RayTriPrecompute::new(rays.dir(ray_idx));
-                    for tri_idx in idx_range.clone() {
-                        let tri_indices = self.indices[tri_idx];
-
-                        // Get triangle if necessary
-                        let tri = if is_cached {
-                            let i = tri_idx - idx_range.start;
-                            unsafe { tri_cache[i].assume_init() }
-                        } else {
-                            let mut tri = if self.time_sample_count == 1 {
-                                // No deformation motion blur, so fast-path it.
-                                (
-                                    self.vertices[tri_indices.0 as usize],
-                                    self.vertices[tri_indices.1 as usize],
-                                    self.vertices[tri_indices.2 as usize],
-                                )
-                            } else {
-                                // Deformation motion blur, need to interpolate.
-                                let p0_slice = &self.vertices[(tri_indices.0 as usize
-                                    * self.time_sample_count)
-                                    ..((tri_indices.0 as usize + 1) * self.time_sample_count)];
-                                let p1_slice = &self.vertices[(tri_indices.1 as usize
-                                    * self.time_sample_count)
-                                    ..((tri_indices.1 as usize + 1) * self.time_sample_count)];
-                                let p2_slice = &self.vertices[(tri_indices.2 as usize
-                                    * self.time_sample_count)
-                                    ..((tri_indices.2 as usize + 1) * self.time_sample_count)];
-
-                                let p0 = lerp_slice(p0_slice, ray_time);
-                                let p1 = lerp_slice(p1_slice, ray_time);
-                                let p2 = lerp_slice(p2_slice, ray_time);
-
-                                (p0, p1, p2)
-                            };
-
-                            if !space.is_empty() {
-                                tri.0 = tri.0 * mat_space;
-                                tri.1 = tri.1 * mat_space;
-                                tri.2 = tri.2 * mat_space;
-                            }
-
-                            tri
-                        };
-
-                        // Test ray against triangle
-                        if let Some((t, b0, b1, b2)) = triangle::intersect_ray(
-                            rays.orig(ray_idx),
-                            ray_pre,
-                            rays.max_t(ray_idx),
-                            tri,
-                        ) {
-                            if rays.is_occlusion(ray_idx) {
-                                isects[ray_idx] = SurfaceIntersection::Occlude;
-                                rays.mark_done(ray_idx);
-                                break;
-                            } else {
-                                non_shadow_hit = true;
-                                rays.set_max_t(ray_idx, t);
-                                unsafe {
-                                    *hit_tri.as_mut_ptr() = tri;
-                                    *hit_tri_indices.as_mut_ptr() = tri_indices;
-                                    *hit_tri_data.as_mut_ptr() = (t, b0, b1, b2);
-                                }
-                            }
-                        }
+                        -s_nor
                     }
+                } else {
+                    geo_normal
+                };
 
-                    // Calculate intersection data if necessary.
-                    if non_shadow_hit {
-                        let hit_tri = unsafe { hit_tri.assume_init() };
-                        let (t, b0, b1, b2) = unsafe { hit_tri_data.assume_init() };
+                let intersection_data = SurfaceIntersectionData {
+                    incoming: ray.dir,
+                    t: t,
+                    pos: pos,
+                    pos_err: pos_err,
+                    nor: shading_normal,
+                    nor_g: geo_normal,
+                    local_space: *space,
+                    sample_pdf: 0.0,
+                };
 
-                        // Calculate intersection point and error magnitudes
-                        let (pos, pos_err) = triangle::surface_point(hit_tri, (b0, b1, b2));
-
-                        // Calculate geometric surface normal
-                        let geo_normal =
-                            cross(hit_tri.0 - hit_tri.1, hit_tri.0 - hit_tri.2).into_normal();
-
-                        // Calculate interpolated surface normal, if any
-                        let shading_normal = if let Some(normals) = self.normals {
-                            let hit_tri_indices = unsafe { hit_tri_indices.assume_init() };
-                            let n0_slice = &normals[(hit_tri_indices.0 as usize
-                                * self.time_sample_count)
-                                ..((hit_tri_indices.0 as usize + 1) * self.time_sample_count)];
-                            let n1_slice = &normals[(hit_tri_indices.1 as usize
-                                * self.time_sample_count)
-                                ..((hit_tri_indices.1 as usize + 1) * self.time_sample_count)];
-                            let n2_slice = &normals[(hit_tri_indices.2 as usize
-                                * self.time_sample_count)
-                                ..((hit_tri_indices.2 as usize + 1) * self.time_sample_count)];
-
-                            let n0 = lerp_slice(n0_slice, ray_time).normalized();
-                            let n1 = lerp_slice(n1_slice, ray_time).normalized();
-                            let n2 = lerp_slice(n2_slice, ray_time).normalized();
-
-                            let s_nor = ((n0 * b0) + (n1 * b1) + (n2 * b2)) * mat_space;
-                            if dot(s_nor, geo_normal) >= 0.0 {
-                                s_nor
-                            } else {
-                                -s_nor
-                            }
-                        } else {
-                            geo_normal
-                        };
-
-                        let intersection_data = SurfaceIntersectionData {
-                            incoming: rays.dir(ray_idx),
-                            t: t,
-                            pos: pos,
-                            pos_err: pos_err,
-                            nor: shading_normal,
-                            nor_g: geo_normal,
-                            local_space: mat_space,
-                            sample_pdf: 0.0,
-                        };
-
-                        // Fill in intersection data
-                        isects[ray_idx] = SurfaceIntersection::Hit {
-                            intersection_data: intersection_data,
-                            closure: shader.shade(&intersection_data, ray_time),
-                        };
-                    }
-                });
-                ray_stack.pop_task();
-            });
+                // Fill in intersection data
+                *isect = SurfaceIntersection::Hit {
+                    intersection_data: intersection_data,
+                    closure: shader.shade(&intersection_data, ray.time),
+                };
+            }
+        });
     }
 }

src/tracer.rs

@@ -1,190 +1,119 @@
-use std::iter;
-
 use crate::{
-    accel::ray_code,
-    color::{rec709_to_xyz, Color},
     lerp::lerp_slice,
-    math::Transform,
-    ray::{RayBatch, RayStack},
+    math::XformFull,
+    ray::{LocalRay, Ray},
     scene::{Assembly, InstanceType, Object},
-    shading::{SimpleSurfaceShader, SurfaceShader},
+    shading::SurfaceShader,
     surface::SurfaceIntersection,
-    transform_stack::TransformStack,
 };
 
 pub struct Tracer<'a> {
+    root: &'a Assembly<'a>,
     ray_trace_count: u64,
-    ray_stack: RayStack,
-    inner: TracerInner<'a>,
 }
 
 impl<'a> Tracer<'a> {
     pub fn from_assembly(assembly: &'a Assembly) -> Tracer<'a> {
         Tracer {
+            root: assembly,
             ray_trace_count: 0,
-            ray_stack: RayStack::new(),
-            inner: TracerInner {
-                root: assembly,
-                xform_stack: TransformStack::new(),
-                isects: Vec::new(),
-            },
         }
     }
 
-    pub fn trace<'b>(&'b mut self, rays: &mut RayBatch) -> &'b [SurfaceIntersection] {
-        self.ray_trace_count += rays.len() as u64;
-        self.inner.trace(rays, &mut self.ray_stack)
-    }
-
     pub fn rays_traced(&self) -> u64 {
         self.ray_trace_count
     }
-}
 
-struct TracerInner<'a> {
-    root: &'a Assembly<'a>,
-    xform_stack: TransformStack,
-    isects: Vec<SurfaceIntersection>,
-}
+    pub fn trace(&mut self, mut ray: Ray) -> SurfaceIntersection {
+        self.ray_trace_count += 1;
 
-impl<'a> TracerInner<'a> {
-    fn trace<'b>(
-        &'b mut self,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
-    ) -> &'b [SurfaceIntersection] {
-        ray_stack.clear();
+        let local_ray = ray.to_local();
+        let space = XformFull::identity();
+        let mut isect = SurfaceIntersection::Miss;
 
-        // Ready the isects
-        self.isects.clear();
-        self.isects.reserve(rays.len());
-        self.isects
-            .extend(iter::repeat(SurfaceIntersection::Miss).take(rays.len()));
+        self.trace_assembly(self.root, &mut ray, &local_ray, &space, &mut isect);
 
-        // Prep the accel part of the rays.
-        {
-            let ident = Transform::new();
-            for i in 0..rays.len() {
-                rays.update_local(i, &ident);
-            }
-        }
-
-        // Divide the rays into 8 different lanes by direction.
-        ray_stack.ensure_lane_count(8);
-        for i in 0..rays.len() {
-            ray_stack.push_ray_index(i, ray_code(rays.dir(i)));
-        }
-        ray_stack.push_lanes_to_tasks(&[0, 1, 2, 3, 4, 5, 6, 7]);
-
-        // Trace each of the 8 lanes separately.
-        while !ray_stack.is_empty() {
-            self.trace_assembly(self.root, rays, ray_stack);
-        }
-
-        &self.isects
+        isect
     }
 
-    fn trace_assembly<'b>(
-        &'b mut self,
+    fn trace_assembly(
+        &mut self,
         assembly: &Assembly,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
+        ray: &mut Ray,
+        local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
     ) {
         assembly
             .object_accel
-            .traverse(rays, ray_stack, |idx_range, rays, ray_stack| {
-                let inst = &assembly.instances[idx_range.start];
+            .traverse(ray, local_ray, |idx_range, ray| {
+                for inst_idx in idx_range {
+                    let inst = &assembly.instances[inst_idx];
 
-                // Transform rays if needed
-                if let Some((xstart, xend)) = inst.transform_indices {
-                    // Push transforms to stack
-                    self.xform_stack.push(&assembly.xforms[xstart..xend]);
+                    // Handle transforms if needed.
+                    let (local_space, local_ray) = if let Some((xstart, xend)) =
+                        inst.transform_indices
+                    {
+                        let instance_xform = lerp_slice(&assembly.xforms[xstart..xend], ray.time);
+                        let combined_xform = instance_xform.compose(&space.fwd);
 
-                    // Do transforms
-                    // TODO: re-divide rays based on direction (maybe?).
-                    let xforms = self.xform_stack.top();
-                    ray_stack.do_next_task(|ray_idx| {
-                        let t = rays.time(ray_idx);
-                        rays.update_local(ray_idx, &lerp_slice(xforms, t));
-                    });
-                    ray_stack.duplicate_next_task();
-                }
-
-                // Trace rays
-                match inst.instance_type {
-                    InstanceType::Object => {
-                        self.trace_object(
-                            &assembly.objects[inst.data_index],
-                            inst.surface_shader_index
-                                .map(|i| assembly.surface_shaders[i]),
-                            rays,
-                            ray_stack,
-                        );
-                    }
-
-                    InstanceType::Assembly => {
-                        self.trace_assembly(&assembly.assemblies[inst.data_index], rays, ray_stack);
-                    }
-                }
-
-                // Un-transform rays if needed
-                if inst.transform_indices.is_some() {
-                    // Pop transforms off stack
-                    self.xform_stack.pop();
-
-                    // Undo transforms
-                    let xforms = self.xform_stack.top();
-                    if !xforms.is_empty() {
-                        ray_stack.pop_do_next_task(|ray_idx| {
-                            let t = rays.time(ray_idx);
-                            rays.update_local(ray_idx, &lerp_slice(xforms, t));
-                        });
+                        if let Some(xform) = combined_xform.to_full() {
+                            (xform, ray.to_local_xform(&xform))
+                        } else {
+                            // Invalid transform, so skip traversing into this instance.
+                            continue;
+                        }
                     } else {
-                        let ident = Transform::new();
-                        ray_stack.pop_do_next_task(|ray_idx| {
-                            rays.update_local(ray_idx, &ident);
-                        });
+                        (*space, *local_ray)
+                    };
+
+                    // Trace ray.
+                    match inst.instance_type {
+                        InstanceType::Object => {
+                            self.trace_object(
+                                &assembly.objects[inst.data_index],
+                                ray,
+                                &local_ray,
+                                &local_space,
+                                isect,
+                                assembly.surface_shaders,
+                            );
+                        }
+
+                        InstanceType::Assembly => {
+                            self.trace_assembly(
+                                &assembly.assemblies[inst.data_index],
+                                ray,
+                                &local_ray,
+                                &local_space,
+                                isect,
+                            );
+                        }
+                    }
+
+                    if ray.is_done() {
+                        return;
                     }
                 }
             });
     }
 
     fn trace_object<'b>(
-        &'b mut self,
+        &mut self,
         obj: &Object,
-        surface_shader: Option<&dyn SurfaceShader>,
-        rays: &mut RayBatch,
-        ray_stack: &mut RayStack,
+        ray: &mut Ray,
+        local_ray: &LocalRay,
+        space: &XformFull,
+        isect: &mut SurfaceIntersection,
+        shaders: &[&dyn SurfaceShader],
     ) {
         match *obj {
             Object::Surface(surface) => {
-                let unassigned_shader = SimpleSurfaceShader::Emit {
-                    color: Color::new_xyz(rec709_to_xyz((1.0, 0.0, 1.0))),
-                };
-                let shader = surface_shader.unwrap_or(&unassigned_shader);
-
-                surface.intersect_rays(
-                    rays,
-                    ray_stack,
-                    &mut self.isects,
-                    shader,
-                    self.xform_stack.top(),
-                );
+                surface.intersect_ray(ray, local_ray, space, isect, shaders);
             }
 
             Object::SurfaceLight(surface) => {
-                // Lights don't use shaders
-                let bogus_shader = SimpleSurfaceShader::Emit {
-                    color: Color::new_xyz(rec709_to_xyz((1.0, 0.0, 1.0))),
-                };
-
-                surface.intersect_rays(
-                    rays,
-                    ray_stack,
-                    &mut self.isects,
-                    &bogus_shader,
-                    self.xform_stack.top(),
-                );
+                surface.intersect_ray(ray, local_ray, space, isect, shaders);
             }
         }
     }

src/transform_stack.rs

@@ -1,83 +1,30 @@
-use std::{
-    cmp,
-    mem::{transmute, MaybeUninit},
-};
-
-use crate::{algorithm::merge_slices_to, math::Transform};
+use crate::math::Xform;
 
 pub struct TransformStack {
-    stack: Vec<MaybeUninit<Transform>>,
-    stack_indices: Vec<usize>,
+    stack: Vec<Xform>,
 }
 
 impl TransformStack {
     pub fn new() -> TransformStack {
-        let mut ts = TransformStack {
-            stack: Vec::new(),
-            stack_indices: Vec::new(),
-        };
-
-        ts.stack_indices.push(0);
-        ts.stack_indices.push(0);
-
-        ts
+        TransformStack { stack: Vec::new() }
     }
 
     pub fn clear(&mut self) {
         self.stack.clear();
-        self.stack_indices.clear();
-        self.stack_indices.push(0);
-        self.stack_indices.push(0);
     }
 
-    pub fn push(&mut self, xforms: &[Transform]) {
-        assert!(!xforms.is_empty());
-
-        if self.stack.is_empty() {
-            let xforms: &[MaybeUninit<Transform>] = unsafe { transmute(xforms) };
-            self.stack.extend(xforms);
-        } else {
-            let sil = self.stack_indices.len();
-            let i1 = self.stack_indices[sil - 2];
-            let i2 = self.stack_indices[sil - 1];
-            // Reserve stack space for the new transforms.
-            // Note this leaves exposed uninitialized memory.  The subsequent call to
-            // merge_slices_to() fills that memory in.
-            {
-                let maxlen = cmp::max(xforms.len(), i2 - i1);
-                self.stack.reserve(maxlen);
-                let l = self.stack.len();
-                unsafe { self.stack.set_len(l + maxlen) };
-            }
-            let (xfs1, xfs2) = self.stack.split_at_mut(i2);
-            merge_slices_to(
-                unsafe { transmute(&xfs1[i1..i2]) },
-                xforms,
-                xfs2,
-                |xf1, xf2| *xf1 * *xf2,
-            );
+    pub fn push(&mut self, xform: Xform) {
+        match self.stack.last() {
+            None => self.stack.push(xform),
+            Some(prev_xform) => self.stack.push(xform.compose(prev_xform)),
         }
-
-        self.stack_indices.push(self.stack.len());
     }
 
-    pub fn pop(&mut self) {
-        assert!(self.stack_indices.len() > 2);
-
-        let sl = self.stack.len();
-        let sil = self.stack_indices.len();
-        let i1 = self.stack_indices[sil - 2];
-        let i2 = self.stack_indices[sil - 1];
-
-        self.stack.truncate(sl - (i2 - i1));
-        self.stack_indices.pop();
+    pub fn pop(&mut self) -> Option<Xform> {
+        self.stack.pop()
     }
 
-    pub fn top(&self) -> &[Transform] {
-        let sil = self.stack_indices.len();
-        let i1 = self.stack_indices[sil - 2];
-        let i2 = self.stack_indices[sil - 1];
-
-        unsafe { transmute(&self.stack[i1..i2]) }
+    pub fn top(&self) -> Option<&Xform> {
+        self.stack.last()
     }
 }

sub_crates/color/Cargo.toml

@@ -6,6 +6,9 @@ edition = "2018"
 license = "MIT, Apache 2.0"
 build = "build.rs"
 
+[build-dependencies]
+colorbox = "0.3"
+
 [lib]
 name = "color"
 path = "src/lib.rs"

sub_crates/color/build.rs

@@ -1,76 +1,46 @@
 use std::{env, fs::File, io::Write, path::Path};
 
-#[derive(Copy, Clone)]
-struct Chromaticities {
-    r: (f64, f64),
-    g: (f64, f64),
-    b: (f64, f64),
-    w: (f64, f64),
-}
+use colorbox::{
+    chroma::{self, Chromaticities},
+    matrix::{invert, rgb_to_xyz_matrix, xyz_chromatic_adaptation_matrix, AdaptationMethod},
+    matrix_compose,
+};
 
 fn main() {
     let out_dir = env::var("OUT_DIR").unwrap();
 
     // Rec709
     {
-        let chroma = Chromaticities {
-            r: (0.640, 0.330),
-            g: (0.300, 0.600),
-            b: (0.150, 0.060),
-            w: (0.3127, 0.3290),
-        };
-
         let dest_path = Path::new(&out_dir).join("rec709_inc.rs");
         let mut f = File::create(&dest_path).unwrap();
-        write_conversion_functions("rec709", chroma, &mut f);
+        write_conversion_functions("rec709", chroma::REC709, &mut f);
     }
 
     // Rec2020
     {
-        let chroma = Chromaticities {
-            r: (0.708, 0.292),
-            g: (0.170, 0.797),
-            b: (0.131, 0.046),
-            w: (0.3127, 0.3290),
-        };
-
         let dest_path = Path::new(&out_dir).join("rec2020_inc.rs");
         let mut f = File::create(&dest_path).unwrap();
-        write_conversion_functions("rec2020", chroma, &mut f);
+        write_conversion_functions("rec2020", chroma::REC2020, &mut f);
     }
 
     // ACES AP0
     {
-        let chroma = Chromaticities {
-            r: (0.73470, 0.26530),
-            g: (0.00000, 1.00000),
-            b: (0.00010, -0.07700),
-            w: (0.32168, 0.33767),
-        };
-
         let dest_path = Path::new(&out_dir).join("aces_ap0_inc.rs");
         let mut f = File::create(&dest_path).unwrap();
-        write_conversion_functions("aces_ap0", chroma, &mut f);
+        write_conversion_functions("aces_ap0", chroma::ACES_AP0, &mut f);
     }
 
     // ACES AP1
     {
-        let chroma = Chromaticities {
-            r: (0.713, 0.293),
-            g: (0.165, 0.830),
-            b: (0.128, 0.044),
-            w: (0.32168, 0.33767),
-        };
-
         let dest_path = Path::new(&out_dir).join("aces_ap1_inc.rs");
         let mut f = File::create(&dest_path).unwrap();
-        write_conversion_functions("aces_ap1", chroma, &mut f);
+        write_conversion_functions("aces_ap1", chroma::ACES_AP1, &mut f);
     }
 }
 
 /// Generates conversion functions for the given rgb to xyz transform matrix.
 fn write_conversion_functions(space_name: &str, chroma: Chromaticities, f: &mut File) {
-    let to_xyz = rgb_to_xyz(chroma, 1.0);
+    let to_xyz = rgb_to_xyz_matrix(chroma);
 
     f.write_all(
         format!(
@@ -99,7 +69,7 @@ pub fn {}_to_xyz(rgb: (f32, f32, f32)) -> (f32, f32, f32) {{
     )
     .unwrap();
 
-    let inv = inverse(to_xyz);
+    let inv = invert(to_xyz).unwrap();
     f.write_all(
         format!(
             r#"
@@ -127,12 +97,14 @@ pub fn xyz_to_{}(xyz: (f32, f32, f32)) -> (f32, f32, f32) {{
     )
     .unwrap();
 
-    let e_chroma = {
-        let mut e_chroma = chroma;
-        e_chroma.w = (1.0 / 3.0, 1.0 / 3.0);
-        e_chroma
-    };
-    let e_to_xyz = rgb_to_xyz(e_chroma, 1.0);
+    let e_to_xyz = matrix_compose!(
+        rgb_to_xyz_matrix(chroma),
+        xyz_chromatic_adaptation_matrix(
+            chroma.w,
+            (1.0 / 3.0, 1.0 / 3.0),
+            AdaptationMethod::Bradford,
+        ),
+    );
     f.write_all(
         format!(
             r#"
@@ -160,7 +132,7 @@ pub fn {}_e_to_xyz(rgb: (f32, f32, f32)) -> (f32, f32, f32) {{
     )
     .unwrap();
 
-    let inv_e = inverse(e_to_xyz);
+    let inv_e = invert(e_to_xyz).unwrap();
     f.write_all(
         format!(
             r#"
@@ -188,135 +160,3 @@ pub fn xyz_to_{}_e(xyz: (f32, f32, f32)) -> (f32, f32, f32) {{
     )
     .unwrap();
 }
-
-/// Port of the RGBtoXYZ function from the ACES CTL reference implementation.
-/// See lib/IlmCtlMath/CtlColorSpace.cpp in the CTL reference implementation.
-///
-/// This takes the chromaticities of an RGB colorspace and generates a
-/// transform matrix from that space to XYZ.
-///
-/// * `chroma` is the chromaticities.
-/// * `y` is the XYZ "Y" value that should map to RGB (1,1,1)
-fn rgb_to_xyz(chroma: Chromaticities, y: f64) -> [[f64; 3]; 3] {
-    // X and Z values of RGB value (1, 1, 1), or "white"
-    let x = chroma.w.0 * y / chroma.w.1;
-    let z = (1.0 - chroma.w.0 - chroma.w.1) * y / chroma.w.1;
-
-    // Scale factors for matrix rows
-    let d = chroma.r.0 * (chroma.b.1 - chroma.g.1)
-        + chroma.b.0 * (chroma.g.1 - chroma.r.1)
-        + chroma.g.0 * (chroma.r.1 - chroma.b.1);
-
-    let sr = (x * (chroma.b.1 - chroma.g.1)
-        - chroma.g.0 * (y * (chroma.b.1 - 1.0) + chroma.b.1 * (x + z))
-        + chroma.b.0 * (y * (chroma.g.1 - 1.0) + chroma.g.1 * (x + z)))
-        / d;
-
-    let sg = (x * (chroma.r.1 - chroma.b.1)
-        + chroma.r.0 * (y * (chroma.b.1 - 1.0) + chroma.b.1 * (x + z))
-        - chroma.b.0 * (y * (chroma.r.1 - 1.0) + chroma.r.1 * (x + z)))
-        / d;
-
-    let sb = (x * (chroma.g.1 - chroma.r.1)
-        - chroma.r.0 * (y * (chroma.g.1 - 1.0) + chroma.g.1 * (x + z))
-        + chroma.g.0 * (y * (chroma.r.1 - 1.0) + chroma.r.1 * (x + z)))
-        / d;
-
-    // Assemble the matrix
-    let mut mat = [[0.0; 3]; 3];
-
-    mat[0][0] = sr * chroma.r.0;
-    mat[0][1] = sg * chroma.g.0;
-    mat[0][2] = sb * chroma.b.0;
-
-    mat[1][0] = sr * chroma.r.1;
-    mat[1][1] = sg * chroma.g.1;
-    mat[1][2] = sb * chroma.b.1;
-
-    mat[2][0] = sr * (1.0 - chroma.r.0 - chroma.r.1);
-    mat[2][1] = sg * (1.0 - chroma.g.0 - chroma.g.1);
-    mat[2][2] = sb * (1.0 - chroma.b.0 - chroma.b.1);
-
-    mat
-}
-
-/// Calculates the inverse of the given 3x3 matrix.
-///
-/// Ported to Rust from `gjInverse()` in IlmBase's Imath/ImathMatrix.h
-fn inverse(m: [[f64; 3]; 3]) -> [[f64; 3]; 3] {
-    let mut s = [[1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]];
-    let mut t = m;
-
-    // Forward elimination
-    for i in 0..2 {
-        let mut pivot = i;
-        let mut pivotsize = t[i][i];
-
-        if pivotsize < 0.0 {
-            pivotsize = -pivotsize;
-        }
-
-        for j in (i + 1)..3 {
-            let mut tmp = t[j][i];
-
-            if tmp < 0.0 {
-                tmp = -tmp;
-            }
-
-            if tmp > pivotsize {
-                pivot = j;
-                pivotsize = tmp;
-            }
-        }
-
-        if pivotsize == 0.0 {
-            panic!("Cannot invert singular matrix.");
-        }
-
-        if pivot != i {
-            for j in 0..3 {
-                let mut tmp = t[i][j];
-                t[i][j] = t[pivot][j];
-                t[pivot][j] = tmp;
-
-                tmp = s[i][j];
-                s[i][j] = s[pivot][j];
-                s[pivot][j] = tmp;
-            }
-        }
-
-        for j in (i + 1)..3 {
-            let f = t[j][i] / t[i][i];
-
-            for k in 0..3 {
-                t[j][k] -= f * t[i][k];
-                s[j][k] -= f * s[i][k];
-            }
-        }
-    }
-
-    // Backward substitution
-    for i in (0..3).rev() {
-        let f = t[i][i];
-
-        if t[i][i] == 0.0 {
-            panic!("Cannot invert singular matrix.");
-        }
-
-        for j in 0..3 {
-            t[i][j] /= f;
-            s[i][j] /= f;
-        }
-
-        for j in 0..i {
-            let f = t[j][i];
-
-            for k in 0..3 {
-                t[j][k] -= f * t[i][k];
-                s[j][k] -= f * s[i][k];
-            }
-        }
-    }
-
-    s
-}

sub_crates/data_tree/Cargo.toml

@@ -0,0 +1,10 @@
+[package]
+name = "data_tree"
+version = "0.1.0"
+authors = ["Nathan Vegdahl <cessen@cessen.com>"]
+edition = "2018"
+license = "MIT"
+
+[lib]
+name = "data_tree"
+path = "src/lib.rs"

sub_crates/data_tree/src/lib.rs

@@ -0,0 +1,199 @@
+#![allow(clippy::redundant_field_names)]
+#![allow(clippy::needless_lifetimes)]
+
+mod parse;
+
+use parse::{ParseError, ParseEvent, Parser};
+
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+pub enum Event<'a> {
+    InnerOpen {
+        type_name: &'a str,
+        byte_offset: usize,
+    },
+    InnerClose {
+        byte_offset: usize,
+    },
+    Leaf {
+        type_name: &'a str,
+        contents: &'a str,
+        byte_offset: usize,
+    },
+    EOF,
+}
+
+//----------------------------------------------------------------------------
+
+#[derive(Debug)]
+pub enum Error {
+    ExpectedNameOrClose(usize),
+    ExpectedOpen(usize),
+    UnexpectedClose(usize),
+    UnexpectedEOF,
+    IO(std::io::Error),
+}
+
+impl std::error::Error for Error {}
+
+impl std::fmt::Display for Error {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
+        write!(f, "{:?}", self)
+    }
+}
+
+impl From<ParseError> for Error {
+    fn from(e: ParseError) -> Self {
+        match e {
+            ParseError::ExpectedNameOrClose(byte_offset) => Error::ExpectedNameOrClose(byte_offset),
+            ParseError::ExpectedOpen(byte_offset) => Error::ExpectedOpen(byte_offset),
+            ParseError::UnexpectedClose(byte_offset) => Error::UnexpectedClose(byte_offset),
+        }
+    }
+}
+
+impl From<std::io::Error> for Error {
+    fn from(e: std::io::Error) -> Self {
+        Error::IO(e)
+    }
+}
+
+//-------------------------------------------------------------
+
+#[derive(Debug)]
+pub struct DataTreeReader<R: std::io::BufRead> {
+    parser: Parser,
+    reader: R,
+    buf: String,
+    eof: bool,
+}
+
+impl<R: std::io::BufRead> DataTreeReader<R> {
+    pub fn new(reader: R) -> Self {
+        Self {
+            parser: Parser::new(),
+            reader: reader,
+            buf: String::new(),
+            eof: false,
+        }
+    }
+
+    pub fn next_event<'a>(&'a mut self) -> Result<Event<'a>, Error> {
+        loop {
+            let valid_end = match self.parser.next_event()? {
+                ParseEvent::ValidEnd => true,
+                ParseEvent::NeedMoreInput => false,
+
+                // The transmutes below are because the borrow checker is
+                // over-conservative about this.  It thinks
+                // the liftime isn't valid, but since we aren't
+                // mutating self after returning (and in fact
+                // can't because of the borrow) there's no way for
+                // the references in this to become invalid.
+                ParseEvent::InnerOpen {
+                    type_name,
+                    byte_offset,
+                } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::InnerOpen {
+                            type_name,
+                            byte_offset,
+                        })
+                    });
+                }
+                ParseEvent::InnerClose { byte_offset } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::InnerClose { byte_offset })
+                    });
+                }
+                ParseEvent::Leaf {
+                    type_name,
+                    contents,
+                    byte_offset,
+                } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::Leaf {
+                            type_name,
+                            contents,
+                            byte_offset,
+                        })
+                    });
+                }
+            };
+
+            if !self.eof {
+                self.buf.clear();
+                let read = self.reader.read_line(&mut self.buf)?;
+                self.parser.push_data(&self.buf);
+                if read == 0 {
+                    self.eof = true;
+                }
+            } else if !valid_end {
+                return Err(Error::UnexpectedEOF);
+            } else {
+                return Ok(Event::EOF);
+            }
+        }
+    }
+
+    pub fn peek_event<'a>(&'a mut self) -> Result<Event<'a>, Error> {
+        loop {
+            let valid_end = match self.parser.peek_event()? {
+                ParseEvent::ValidEnd => true,
+                ParseEvent::NeedMoreInput => false,
+
+                // The transmutes below are because the borrow checker is
+                // over-conservative about this.  It thinks
+                // the liftime isn't valid, but since we aren't
+                // mutating self after returning (and in fact
+                // can't because of the borrow) there's no way for
+                // the references in this to become invalid.
+                ParseEvent::InnerOpen {
+                    type_name,
+                    byte_offset,
+                } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::InnerOpen {
+                            type_name,
+                            byte_offset,
+                        })
+                    });
+                }
+                ParseEvent::InnerClose { byte_offset } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::InnerClose { byte_offset })
+                    });
+                }
+                ParseEvent::Leaf {
+                    type_name,
+                    contents,
+                    byte_offset,
+                } => {
+                    return Ok(unsafe {
+                        std::mem::transmute::<Event, Event>(Event::Leaf {
+                            type_name,
+                            contents,
+                            byte_offset,
+                        })
+                    });
+                }
+            };
+
+            if !self.eof {
+                self.buf.clear();
+                let read = self.reader.read_line(&mut self.buf)?;
+                self.parser.push_data(&self.buf);
+                if read == 0 {
+                    self.eof = true;
+                }
+            } else if !valid_end {
+                return Err(Error::UnexpectedEOF);
+            } else {
+                return Ok(Event::EOF);
+            }
+        }
+    }
+
+    pub fn byte_offset(&self) -> usize {
+        self.parser.byte_offset()
+    }
+}

sub_crates/data_tree/src/parse.rs

@@ -0,0 +1,762 @@
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+pub enum ParseError {
+    ExpectedNameOrClose(usize),
+    ExpectedOpen(usize),
+    UnexpectedClose(usize),
+}
+
+impl std::error::Error for ParseError {}
+
+impl std::fmt::Display for ParseError {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
+        write!(f, "{:?}", self)
+    }
+}
+
+//---------------------------------------------------------------------
+
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+pub enum ParseEvent<'a> {
+    InnerOpen {
+        type_name: &'a str,
+        byte_offset: usize,
+    },
+    InnerClose {
+        byte_offset: usize,
+    },
+    Leaf {
+        type_name: &'a str,
+        contents: &'a str,
+        byte_offset: usize,
+    },
+    NeedMoreInput,
+    ValidEnd, // All data so far is consumed, and this is a
+              // valid place to finish the parse.
+}
+
+impl<'a> ParseEvent<'a> {
+    fn add_to_byte_offset(&self, offset: usize) -> ParseEvent<'a> {
+        match *self {
+            ParseEvent::InnerOpen {
+                type_name,
+                byte_offset,
+            } => ParseEvent::InnerOpen {
+                type_name: type_name,
+                byte_offset: byte_offset + offset,
+            },
+            ParseEvent::InnerClose { byte_offset } => ParseEvent::InnerClose {
+                byte_offset: byte_offset + offset,
+            },
+            ParseEvent::Leaf {
+                type_name,
+                contents,
+                byte_offset,
+            } => ParseEvent::Leaf {
+                type_name: type_name,
+                contents: contents,
+                byte_offset: byte_offset + offset,
+            },
+            ParseEvent::NeedMoreInput => *self,
+            ParseEvent::ValidEnd => *self,
+        }
+    }
+}
+
+//---------------------------------------------------------------------
+
+#[derive(Debug)]
+pub struct Parser {
+    buffer: String,
+    buf_consumed_idx: usize,
+    total_bytes_processed: usize,
+    inner_opens: usize,
+}
+
+impl Parser {
+    pub fn new() -> Parser {
+        Parser {
+            buffer: String::with_capacity(1024),
+            buf_consumed_idx: 0,
+            total_bytes_processed: 0,
+            inner_opens: 0,
+        }
+    }
+
+    pub fn push_data(&mut self, text: &str) {
+        // Remove any consumed data.
+        if self.buf_consumed_idx > 0 {
+            self.buffer.replace_range(..self.buf_consumed_idx, "");
+            self.buf_consumed_idx = 0;
+        }
+
+        // Add the new data.
+        self.buffer.push_str(text);
+    }
+
+    pub fn next_event<'a>(&'a mut self) -> Result<ParseEvent<'a>, ParseError> {
+        // Remove any consumed data.
+        if self.buf_consumed_idx > 0 {
+            self.buffer.replace_range(..self.buf_consumed_idx, "");
+            self.buf_consumed_idx = 0;
+        }
+
+        // Try to parse an event from the valid prefix.
+        match try_parse_event(&self.buffer) {
+            ParseEventParse::Ok(event, bytes_consumed) => {
+                // Update internal state.
+                if let ParseEvent::InnerOpen { .. } = event {
+                    self.inner_opens += 1;
+                } else if let ParseEvent::InnerClose { byte_offset, .. } = event {
+                    if self.inner_opens == 0 {
+                        return Err(ParseError::UnexpectedClose(
+                            byte_offset + self.total_bytes_processed,
+                        ));
+                    } else {
+                        self.inner_opens -= 1;
+                    }
+                }
+                self.buf_consumed_idx += bytes_consumed;
+                self.total_bytes_processed += bytes_consumed;
+
+                Ok(event.add_to_byte_offset(self.total_bytes_processed - self.buf_consumed_idx))
+            }
+            ParseEventParse::ReachedEnd => {
+                // If we consumed all data, then if all nodes are properly
+                // closed we're done.  Otherwise we need more input.
+                if self.inner_opens == 0 {
+                    Ok(ParseEvent::ValidEnd)
+                } else {
+                    Ok(ParseEvent::NeedMoreInput)
+                }
+            }
+            ParseEventParse::IncompleteData => Ok(ParseEvent::NeedMoreInput),
+
+            // Hard errors.
+            ParseEventParse::ExpectedNameOrInnerClose(byte_offset) => Err(
+                ParseError::ExpectedNameOrClose(byte_offset + self.total_bytes_processed),
+            ),
+            ParseEventParse::ExpectedOpen(byte_offset) => Err(ParseError::ExpectedOpen(
+                byte_offset + self.total_bytes_processed,
+            )),
+        }
+    }
+
+    pub fn peek_event<'a>(&'a mut self) -> Result<ParseEvent<'a>, ParseError> {
+        // Remove any consumed data.
+        if self.buf_consumed_idx > 0 {
+            self.buffer.replace_range(..self.buf_consumed_idx, "");
+            self.buf_consumed_idx = 0;
+        }
+
+        // Try to parse an event from the valid prefix.
+        match try_parse_event(&self.buffer) {
+            ParseEventParse::Ok(event, _bytes_consumed) => {
+                if let ParseEvent::InnerClose { byte_offset, .. } = event {
+                    if self.inner_opens == 0 {
+                        return Err(ParseError::UnexpectedClose(
+                            byte_offset + self.total_bytes_processed,
+                        ));
+                    }
+                }
+                Ok(event.add_to_byte_offset(self.total_bytes_processed))
+            }
+            ParseEventParse::ReachedEnd => {
+                // If we consumed all data, then if all nodes are properly
+                // closed we're done.  Otherwise we need more input.
+                if self.inner_opens == 0 {
+                    Ok(ParseEvent::ValidEnd)
+                } else {
+                    Ok(ParseEvent::NeedMoreInput)
+                }
+            }
+            ParseEventParse::IncompleteData => Ok(ParseEvent::NeedMoreInput),
+
+            // Hard errors.
+            ParseEventParse::ExpectedNameOrInnerClose(byte_offset) => Err(
+                ParseError::ExpectedNameOrClose(byte_offset + self.total_bytes_processed),
+            ),
+            ParseEventParse::ExpectedOpen(byte_offset) => Err(ParseError::ExpectedOpen(
+                byte_offset + self.total_bytes_processed,
+            )),
+        }
+    }
+
+    pub fn byte_offset(&self) -> usize {
+        self.total_bytes_processed + self.buf_consumed_idx
+    }
+}
+
+//--------------------------------------------------------------------------
+
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+enum ParseEventParse<'a> {
+    Ok(ParseEvent<'a>, usize), // (event, bytes consumed)
+    ReachedEnd,                // Reached the end of the buffer in a valid state, with no event.
+    IncompleteData,            // Need more data to parse.
+
+    // ParseErrors.
+    ExpectedNameOrInnerClose(usize),
+    ExpectedOpen(usize),
+}
+
+fn try_parse_event<'a>(text: &'a str) -> ParseEventParse<'a> {
+    // Remove leading whitespace and comments.
+    let mut source_text = skip_ws_and_comments((0, text));
+    let start_idx = source_text.0;
+
+    // First token.
+    let type_name = match next_token(source_text) {
+        // Type name, record and continue.
+        (Token::Name(tn), tail) => {
+            source_text = tail;
+            tn
+        }
+
+        // Closing tag for inner node.  Return.
+        (Token::CloseInner, tail) => {
+            return ParseEventParse::Ok(
+                ParseEvent::InnerClose {
+                    byte_offset: start_idx,
+                },
+                tail.0,
+            );
+        }
+
+        // We consumed everything as whitespace and/or
+        // comments.  Return.
+        (Token::End, _) => {
+            return ParseEventParse::ReachedEnd;
+        }
+
+        // Invalid.
+        _ => return ParseEventParse::ExpectedNameOrInnerClose(start_idx),
+    };
+
+    // Skip whitespace and comments to get the start of
+    // where there should be an open tag, for use later in error.
+    source_text = skip_ws_and_comments(source_text);
+    let open_start_idx = source_text.0;
+
+    // Last part of the event.
+    match next_token(source_text) {
+        // Begining of an inner node.
+        (Token::OpenInner, tail) => ParseEventParse::Ok(
+            ParseEvent::InnerOpen {
+                type_name: type_name,
+                byte_offset: start_idx,
+            },
+            tail.0,
+        ),
+
+        // Try to parse entire leaf node.
+        (Token::OpenLeaf, tail) => {
+            // Get contents.
+            let (contents, tail2) = parse_leaf_content(tail);
+            source_text = tail2;
+
+            // Try to get closing tag.
+            match next_token(source_text) {
+                // If it's a leaf closing tag, we're done!
+                // Return the leaf event.
+                (Token::CloseLeaf, tail) => ParseEventParse::Ok(
+                    ParseEvent::Leaf {
+                        type_name: type_name,
+                        contents: contents,
+                        byte_offset: start_idx,
+                    },
+                    tail.0,
+                ),
+
+                // Otherwise...
+                _ => {
+                    if source_text.1.is_empty() {
+                        // If there's no text left, we're just incomplete.
+                        ParseEventParse::IncompleteData
+                    } else {
+                        // Otherwise, this would be a parse error...
+                        // except that this shouldn't be reachable,
+                        // since everything should be consumable for
+                        // leaf content up until a close tag.
+                        unreachable!("Expected leaf close tag.")
+                    }
+                }
+            }
+        }
+
+        // We consumed everything else as whitespace
+        // and/or comments, so we're incomplete.  Return.
+        (Token::End, _) => ParseEventParse::IncompleteData,
+
+        // Invalid.
+        _ => ParseEventParse::ExpectedOpen(open_start_idx),
+    }
+}
+
+fn parse_leaf_content(source_text: (usize, &str)) -> (&str, (usize, &str)) {
+    let mut si = 1;
+    let mut escaped = false;
+    let mut reached_end = true;
+    for (i, c) in source_text.1.char_indices() {
+        si = i;
+        if escaped {
+            escaped = false;
+        } else if c == '\\' {
+            escaped = true;
+        } else if c == ']' {
+            reached_end = false;
+            break;
+        }
+    }
+
+    if reached_end {
+        si = source_text.1.len();
+    }
+
+    (
+        &source_text.1[0..si],
+        (source_text.0 + si, &source_text.1[si..]),
+    )
+}
+
+//--------------------------------------------------------------------------
+
+#[derive(Debug, Copy, Clone, PartialEq, Eq)]
+enum Token<'a> {
+    OpenInner,
+    CloseInner,
+    OpenLeaf,
+    CloseLeaf,
+    Name(&'a str),
+    End,
+    Unknown,
+}
+
+fn next_token<'a>(source_text: (usize, &'a str)) -> (Token<'a>, (usize, &'a str)) {
+    let text1 = skip_ws_and_comments(source_text);
+
+    if let Some(c) = text1.1.chars().next() {
+        let text2 = (text1.0 + c.len_utf8(), &text1.1[c.len_utf8()..]);
+        match c {
+            '{' => (Token::OpenInner, text2),
+
+            '}' => (Token::CloseInner, text2),
+
+            '[' => (Token::OpenLeaf, text2),
+
+            ']' => (Token::CloseLeaf, text2),
+
+            _ => {
+                if is_ident_char(c) {
+                    // Parse type
+                    let mut si = 0;
+                    let mut reached_end = true;
+                    for (i, c) in text1.1.char_indices() {
+                        si = i;
+                        if !is_ident_char(c) {
+                            reached_end = false;
+                            break;
+                        }
+                    }
+
+                    if reached_end {
+                        si = text1.1.len();
+                    }
+
+                    (Token::Name(&text1.1[0..si]), (text1.0 + si, &text1.1[si..]))
+                } else {
+                    (Token::Unknown, text1)
+                }
+            }
+        }
+    } else {
+        (Token::End, text1)
+    }
+}
+
+fn is_ws(c: char) -> bool {
+    matches!(c, '\n' | '\r' | '\t' | ' ')
+}
+
+fn is_nl(c: char) -> bool {
+    c == '\n'
+}
+
+fn is_ident_char(c: char) -> bool {
+    c.is_alphanumeric() || c == '-' || c == '_'
+}
+
+fn skip_ws(text: &str) -> &str {
+    let mut si = 0;
+    let mut reached_end = true;
+    for (i, c) in text.char_indices() {
+        si = i;
+        if !is_ws(c) {
+            reached_end = false;
+            break;
+        }
+    }
+
+    if reached_end {
+        si = text.len();
+    }
+
+    &text[si..]
+}
+
+fn skip_comment(text: &str) -> &str {
+    let mut si = 0;
+    if text.starts_with('#') {
+        let mut reached_end = true;
+        for (i, c) in text.char_indices() {
+            si = i;
+            if is_nl(c) {
+                reached_end = false;
+                break;
+            }
+        }
+
+        if reached_end {
+            si = text.len();
+        }
+    }
+
+    &text[si..]
+}
+
+fn skip_ws_and_comments(text: (usize, &str)) -> (usize, &str) {
+    let mut remaining_text = text.1;
+
+    loop {
+        let tmp = skip_comment(skip_ws(remaining_text));
+
+        if tmp.len() == remaining_text.len() {
+            break;
+        } else {
+            remaining_text = tmp;
+        }
+    }
+
+    let offset = text.0 + text.1.len() - remaining_text.len();
+    (offset, remaining_text)
+}
+
+//--------------------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use super::{next_token, Token};
+
+    #[test]
+    fn tokenize_01() {
+        let input = (0, "Thing");
+
+        assert_eq!(next_token(input), (Token::Name("Thing"), (5, "")));
+    }
+
+    #[test]
+    fn tokenize_02() {
+        let input = (0, "  \n# gdfgdf gfdg dggdf\\sg dfgsd \n   Thing");
+
+        assert_eq!(next_token(input), (Token::Name("Thing"), (41, "")));
+    }
+
+    #[test]
+    fn tokenize_03() {
+        let input1 = (0, " Thing { }");
+        let (token1, input2) = next_token(input1);
+        let (token2, input3) = next_token(input2);
+        let (token3, input4) = next_token(input3);
+
+        assert_eq!((token1, input2.1), (Token::Name("Thing"), " { }"));
+        assert_eq!((token2, input3.1), (Token::OpenInner, " }"));
+        assert_eq!((token3, input4.1), (Token::CloseInner, ""));
+    }
+
+    #[test]
+    fn tokenize_04() {
+        let input1 = (0, " hi the[re");
+        let (token1, input2) = next_token(input1);
+        let (token2, input3) = next_token(input2);
+        let (token3, input4) = next_token(input3);
+        let (token4, input5) = next_token(input4);
+        let (token5, input6) = next_token(input5);
+
+        assert_eq!((token1, input2), (Token::Name("hi"), (3, " the[re")));
+        assert_eq!((token2, input3), (Token::Name("the"), (7, "[re")));
+        assert_eq!((token3, input4), (Token::OpenLeaf, (8, "re")));
+        assert_eq!((token4, input5), (Token::Name("re"), (10, "")));
+        assert_eq!((token5, input6), (Token::End, (10, "")));
+    }
+
+    #[test]
+    fn tokenize_05() {
+        let input1 = (0, "Thing { # A comment\n\tThing2 []\n}");
+        let (token1, input2) = next_token(input1);
+        let (token2, input3) = next_token(input2);
+        let (token3, input4) = next_token(input3);
+        let (token4, input5) = next_token(input4);
+        let (token5, input6) = next_token(input5);
+        let (token6, input7) = next_token(input6);
+        let (token7, input8) = next_token(input7);
+
+        assert_eq!(
+            (token1, input2),
+            (Token::Name("Thing"), (5, " { # A comment\n\tThing2 []\n}",))
+        );
+        assert_eq!(
+            (token2, input3),
+            (Token::OpenInner, (7, " # A comment\n\tThing2 []\n}",))
+        );
+        assert_eq!((token3, input4), (Token::Name("Thing2"), (27, " []\n}")));
+        assert_eq!((token4, input5), (Token::OpenLeaf, (29, "]\n}")));
+        assert_eq!((token5, input6), (Token::CloseLeaf, (30, "\n}")));
+        assert_eq!((token6, input7), (Token::CloseInner, (32, "")));
+        assert_eq!((token7, input8), (Token::End, (32, "")));
+    }
+
+    #[test]
+    fn try_parse_event_01() {
+        assert_eq!(try_parse_event("H"), ParseEventParse::IncompleteData,);
+    }
+
+    #[test]
+    fn try_parse_event_02() {
+        assert_eq!(try_parse_event("Hello "), ParseEventParse::IncompleteData,);
+    }
+
+    #[test]
+    fn try_parse_event_03() {
+        assert_eq!(
+            try_parse_event("Hello {"),
+            ParseEventParse::Ok(
+                ParseEvent::InnerOpen {
+                    type_name: "Hello",
+                    byte_offset: 0,
+                },
+                7
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_04() {
+        assert_eq!(
+            try_parse_event("  Hello {"),
+            ParseEventParse::Ok(
+                ParseEvent::InnerOpen {
+                    type_name: "Hello",
+                    byte_offset: 2,
+                },
+                9
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_05() {
+        assert_eq!(
+            try_parse_event("Hello {  "),
+            ParseEventParse::Ok(
+                ParseEvent::InnerOpen {
+                    type_name: "Hello",
+                    byte_offset: 0,
+                },
+                7
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_06() {
+        assert_eq!(try_parse_event("Hello ["), ParseEventParse::IncompleteData,);
+    }
+
+    #[test]
+    fn try_parse_event_07() {
+        assert_eq!(
+            try_parse_event("Hello [some contents"),
+            ParseEventParse::IncompleteData,
+        );
+    }
+
+    #[test]
+    fn try_parse_event_08() {
+        assert_eq!(
+            try_parse_event("Hello [some contents]"),
+            ParseEventParse::Ok(
+                ParseEvent::Leaf {
+                    type_name: "Hello",
+                    contents: "some contents",
+                    byte_offset: 0,
+                },
+                21
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_09() {
+        assert_eq!(
+            try_parse_event("Hello [some contents]  "),
+            ParseEventParse::Ok(
+                ParseEvent::Leaf {
+                    type_name: "Hello",
+                    contents: "some contents",
+                    byte_offset: 0,
+                },
+                21
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_10() {
+        assert_eq!(
+            try_parse_event(r#"Hello [some \co\]ntents]"#),
+            ParseEventParse::Ok(
+                ParseEvent::Leaf {
+                    type_name: "Hello",
+                    contents: r#"some \co\]ntents"#,
+                    byte_offset: 0,
+                },
+                24
+            ),
+        );
+    }
+
+    #[test]
+    fn try_parse_event_11() {
+        assert_eq!(
+            try_parse_event("  # A comment\n\n     "),
+            ParseEventParse::ReachedEnd,
+        );
+    }
+
+    #[test]
+    fn parser_01() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello");
+        assert_eq!(parser.next_event(), Ok(ParseEvent::NeedMoreInput));
+
+        parser.push_data("{");
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::InnerOpen {
+                type_name: "Hello",
+                byte_offset: 0,
+            })
+        );
+
+        assert_eq!(parser.next_event(), Ok(ParseEvent::NeedMoreInput));
+
+        parser.push_data("}");
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::InnerClose { byte_offset: 6 })
+        );
+
+        assert_eq!(parser.next_event(), Ok(ParseEvent::ValidEnd));
+    }
+
+    #[test]
+    fn parser_02() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello");
+        assert_eq!(parser.next_event(), Ok(ParseEvent::NeedMoreInput));
+
+        parser.push_data("[");
+        assert_eq!(parser.next_event(), Ok(ParseEvent::NeedMoreInput));
+
+        parser.push_data("1.0 2.0 3.");
+        assert_eq!(parser.next_event(), Ok(ParseEvent::NeedMoreInput));
+
+        parser.push_data("0]");
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::Leaf {
+                type_name: "Hello",
+                contents: "1.0 2.0 3.0",
+                byte_offset: 0,
+            })
+        );
+
+        assert_eq!(parser.next_event(), Ok(ParseEvent::ValidEnd));
+    }
+
+    #[test]
+    fn parser_03() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello { World [1.0 2.0 3.0] }");
+
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::InnerOpen {
+                type_name: "Hello",
+                byte_offset: 0,
+            })
+        );
+
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::Leaf {
+                type_name: "World",
+                contents: "1.0 2.0 3.0",
+                byte_offset: 8,
+            })
+        );
+
+        assert_eq!(
+            parser.next_event(),
+            Ok(ParseEvent::InnerClose { byte_offset: 28 })
+        );
+
+        // Make sure repeated calls are stable.
+        assert_eq!(parser.next_event(), Ok(ParseEvent::ValidEnd));
+        assert_eq!(parser.next_event(), Ok(ParseEvent::ValidEnd));
+        assert_eq!(parser.next_event(), Ok(ParseEvent::ValidEnd));
+    }
+
+    #[test]
+    fn parser_04() {
+        let mut parser = Parser::new();
+
+        parser.push_data("$%^&");
+        assert_eq!(parser.next_event(), Err(ParseError::ExpectedNameOrClose(0)));
+    }
+
+    #[test]
+    fn parser_05() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello]");
+        assert_eq!(parser.next_event(), Err(ParseError::ExpectedOpen(5)));
+    }
+
+    #[test]
+    fn parser_06() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello}");
+        assert_eq!(parser.next_event(), Err(ParseError::ExpectedOpen(5)));
+    }
+
+    #[test]
+    fn parser_07() {
+        let mut parser = Parser::new();
+
+        parser.push_data("Hello $*@^ [");
+        assert_eq!(parser.next_event(), Err(ParseError::ExpectedOpen(6)));
+    }
+
+    #[test]
+    fn parser_08() {
+        let mut parser = Parser::new();
+
+        parser.push_data("}");
+        assert_eq!(parser.next_event(), Err(ParseError::UnexpectedClose(0)));
+    }
+}

sub_crates/math3d/src/lib.rs

@@ -1,28 +0,0 @@
-#![allow(dead_code)]
-
-mod normal;
-mod point;
-mod transform;
-mod vector;
-
-pub use self::{normal::Normal, point::Point, transform::Transform, vector::Vector};
-
-/// Trait for calculating dot products.
-pub trait DotProduct {
-    fn dot(self, other: Self) -> f32;
-}
-
-#[inline]
-pub fn dot<T: DotProduct>(a: T, b: T) -> f32 {
-    a.dot(b)
-}
-
-/// Trait for calculating cross products.
-pub trait CrossProduct {
-    fn cross(self, other: Self) -> Self;
-}
-
-#[inline]
-pub fn cross<T: CrossProduct>(a: T, b: T) -> T {
-    a.cross(b)
-}

sub_crates/math3d/src/normal.rs

@@ -1,270 +0,0 @@
-#![allow(dead_code)]
-
-use std::{
-    cmp::PartialEq,
-    ops::{Add, Div, Mul, Neg, Sub},
-};
-
-use glam::Vec3A;
-
-use super::{CrossProduct, DotProduct, Transform, Vector};
-
-/// A surface normal in 3d homogeneous space.
-#[derive(Debug, Copy, Clone)]
-pub struct Normal {
-    pub co: Vec3A,
-}
-
-impl Normal {
-    #[inline(always)]
-    pub fn new(x: f32, y: f32, z: f32) -> Normal {
-        Normal {
-            co: Vec3A::new(x, y, z),
-        }
-    }
-
-    #[inline(always)]
-    pub fn length(&self) -> f32 {
-        self.co.length()
-    }
-
-    #[inline(always)]
-    pub fn length2(&self) -> f32 {
-        self.co.length_squared()
-    }
-
-    #[inline(always)]
-    pub fn normalized(&self) -> Normal {
-        Normal {
-            co: self.co.normalize(),
-        }
-    }
-
-    #[inline(always)]
-    pub fn into_vector(self) -> Vector {
-        Vector { co: self.co }
-    }
-
-    #[inline(always)]
-    pub fn get_n(&self, n: usize) -> f32 {
-        match n {
-            0 => self.x(),
-            1 => self.y(),
-            2 => self.z(),
-            _ => panic!("Attempt to access dimension beyond z."),
-        }
-    }
-
-    #[inline(always)]
-    pub fn x(&self) -> f32 {
-        self.co[0]
-    }
-
-    #[inline(always)]
-    pub fn y(&self) -> f32 {
-        self.co[1]
-    }
-
-    #[inline(always)]
-    pub fn z(&self) -> f32 {
-        self.co[2]
-    }
-
-    #[inline(always)]
-    pub fn set_x(&mut self, x: f32) {
-        self.co[0] = x;
-    }
-
-    #[inline(always)]
-    pub fn set_y(&mut self, y: f32) {
-        self.co[1] = y;
-    }
-
-    #[inline(always)]
-    pub fn set_z(&mut self, z: f32) {
-        self.co[2] = z;
-    }
-}
-
-impl PartialEq for Normal {
-    #[inline(always)]
-    fn eq(&self, other: &Normal) -> bool {
-        self.co == other.co
-    }
-}
-
-impl Add for Normal {
-    type Output = Normal;
-
-    #[inline(always)]
-    fn add(self, other: Normal) -> Normal {
-        Normal {
-            co: self.co + other.co,
-        }
-    }
-}
-
-impl Sub for Normal {
-    type Output = Normal;
-
-    #[inline(always)]
-    fn sub(self, other: Normal) -> Normal {
-        Normal {
-            co: self.co - other.co,
-        }
-    }
-}
-
-impl Mul<f32> for Normal {
-    type Output = Normal;
-
-    #[inline(always)]
-    fn mul(self, other: f32) -> Normal {
-        Normal {
-            co: self.co * other,
-        }
-    }
-}
-
-impl Mul<Transform> for Normal {
-    type Output = Normal;
-
-    #[inline]
-    fn mul(self, other: Transform) -> Normal {
-        Normal {
-            co: other.0.matrix3.inverse().transpose().mul_vec3a(self.co),
-        }
-    }
-}
-
-impl Div<f32> for Normal {
-    type Output = Normal;
-
-    #[inline(always)]
-    fn div(self, other: f32) -> Normal {
-        Normal {
-            co: self.co / other,
-        }
-    }
-}
-
-impl Neg for Normal {
-    type Output = Normal;
-
-    #[inline(always)]
-    fn neg(self) -> Normal {
-        Normal { co: self.co * -1.0 }
-    }
-}
-
-impl DotProduct for Normal {
-    #[inline(always)]
-    fn dot(self, other: Normal) -> f32 {
-        self.co.dot(other.co)
-    }
-}
-
-impl CrossProduct for Normal {
-    #[inline]
-    fn cross(self, other: Normal) -> Normal {
-        Normal {
-            co: self.co.cross(other.co),
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::super::{CrossProduct, DotProduct, Transform};
-    use super::*;
-    use approx::assert_ulps_eq;
-
-    #[test]
-    fn add() {
-        let v1 = Normal::new(1.0, 2.0, 3.0);
-        let v2 = Normal::new(1.5, 4.5, 2.5);
-        let v3 = Normal::new(2.5, 6.5, 5.5);
-
-        assert_eq!(v3, v1 + v2);
-    }
-
-    #[test]
-    fn sub() {
-        let v1 = Normal::new(1.0, 2.0, 3.0);
-        let v2 = Normal::new(1.5, 4.5, 2.5);
-        let v3 = Normal::new(-0.5, -2.5, 0.5);
-
-        assert_eq!(v3, v1 - v2);
-    }
-
-    #[test]
-    fn mul_scalar() {
-        let v1 = Normal::new(1.0, 2.0, 3.0);
-        let v2 = 2.0;
-        let v3 = Normal::new(2.0, 4.0, 6.0);
-
-        assert_eq!(v3, v1 * v2);
-    }
-
-    #[test]
-    fn mul_matrix_1() {
-        let n = Normal::new(1.0, 2.5, 4.0);
-        let m = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        let nm = n * m;
-        let nm2 = Normal::new(-4.0625, 1.78125, -0.03125);
-        for i in 0..3 {
-            assert_ulps_eq!(nm.co[i], nm2.co[i], max_ulps = 4);
-        }
-    }
-
-    #[test]
-    fn div() {
-        let v1 = Normal::new(1.0, 2.0, 3.0);
-        let v2 = 2.0;
-        let v3 = Normal::new(0.5, 1.0, 1.5);
-
-        assert_eq!(v3, v1 / v2);
-    }
-
-    #[test]
-    fn length() {
-        let n = Normal::new(1.0, 2.0, 3.0);
-        assert!((n.length() - 3.7416573867739413).abs() < 0.000001);
-    }
-
-    #[test]
-    fn length2() {
-        let n = Normal::new(1.0, 2.0, 3.0);
-        assert_eq!(n.length2(), 14.0);
-    }
-
-    #[test]
-    fn normalized() {
-        let n1 = Normal::new(1.0, 2.0, 3.0);
-        let n2 = Normal::new(0.2672612419124244, 0.5345224838248488, 0.8017837257372732);
-        let n3 = n1.normalized();
-        assert!((n3.x() - n2.x()).abs() < 0.000001);
-        assert!((n3.y() - n2.y()).abs() < 0.000001);
-        assert!((n3.z() - n2.z()).abs() < 0.000001);
-    }
-
-    #[test]
-    fn dot_test() {
-        let v1 = Normal::new(1.0, 2.0, 3.0);
-        let v2 = Normal::new(1.5, 4.5, 2.5);
-        let v3 = 18.0f32;
-
-        assert_eq!(v3, v1.dot(v2));
-    }
-
-    #[test]
-    fn cross_test() {
-        let v1 = Normal::new(1.0, 0.0, 0.0);
-        let v2 = Normal::new(0.0, 1.0, 0.0);
-        let v3 = Normal::new(0.0, 0.0, 1.0);
-
-        assert_eq!(v3, v1.cross(v2));
-    }
-}

sub_crates/math3d/src/point.rs

@@ -1,202 +0,0 @@
-#![allow(dead_code)]
-
-use std::{
-    cmp::PartialEq,
-    ops::{Add, Mul, Sub},
-};
-
-use glam::Vec3A;
-
-use super::{Transform, Vector};
-
-/// A position in 3d homogeneous space.
-#[derive(Debug, Copy, Clone)]
-pub struct Point {
-    pub co: Vec3A,
-}
-
-impl Point {
-    #[inline(always)]
-    pub fn new(x: f32, y: f32, z: f32) -> Point {
-        Point {
-            co: Vec3A::new(x, y, z),
-        }
-    }
-
-    #[inline(always)]
-    pub fn min(&self, other: Point) -> Point {
-        let n1 = self;
-        let n2 = other;
-
-        Point {
-            co: n1.co.min(n2.co),
-        }
-    }
-
-    #[inline(always)]
-    pub fn max(&self, other: Point) -> Point {
-        let n1 = self;
-        let n2 = other;
-
-        Point {
-            co: n1.co.max(n2.co),
-        }
-    }
-
-    #[inline(always)]
-    pub fn into_vector(self) -> Vector {
-        Vector { co: self.co }
-    }
-
-    #[inline(always)]
-    pub fn get_n(&self, n: usize) -> f32 {
-        match n {
-            0 => self.x(),
-            1 => self.y(),
-            2 => self.z(),
-            _ => panic!("Attempt to access dimension beyond z."),
-        }
-    }
-
-    #[inline(always)]
-    pub fn x(&self) -> f32 {
-        self.co[0]
-    }
-
-    #[inline(always)]
-    pub fn y(&self) -> f32 {
-        self.co[1]
-    }
-
-    #[inline(always)]
-    pub fn z(&self) -> f32 {
-        self.co[2]
-    }
-
-    #[inline(always)]
-    pub fn set_x(&mut self, x: f32) {
-        self.co[0] = x;
-    }
-
-    #[inline(always)]
-    pub fn set_y(&mut self, y: f32) {
-        self.co[1] = y;
-    }
-
-    #[inline(always)]
-    pub fn set_z(&mut self, z: f32) {
-        self.co[2] = z;
-    }
-}
-
-impl PartialEq for Point {
-    #[inline(always)]
-    fn eq(&self, other: &Point) -> bool {
-        self.co == other.co
-    }
-}
-
-impl Add<Vector> for Point {
-    type Output = Point;
-
-    #[inline(always)]
-    fn add(self, other: Vector) -> Point {
-        Point {
-            co: self.co + other.co,
-        }
-    }
-}
-
-impl Sub for Point {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn sub(self, other: Point) -> Vector {
-        Vector {
-            co: self.co - other.co,
-        }
-    }
-}
-
-impl Sub<Vector> for Point {
-    type Output = Point;
-
-    #[inline(always)]
-    fn sub(self, other: Vector) -> Point {
-        Point {
-            co: self.co - other.co,
-        }
-    }
-}
-
-impl Mul<Transform> for Point {
-    type Output = Point;
-
-    #[inline]
-    fn mul(self, other: Transform) -> Point {
-        Point {
-            co: other.0.transform_point3a(self.co),
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::super::{Transform, Vector};
-    use super::*;
-
-    #[test]
-    fn add() {
-        let p1 = Point::new(1.0, 2.0, 3.0);
-        let v1 = Vector::new(1.5, 4.5, 2.5);
-        let p2 = Point::new(2.5, 6.5, 5.5);
-
-        assert_eq!(p2, p1 + v1);
-    }
-
-    #[test]
-    fn sub() {
-        let p1 = Point::new(1.0, 2.0, 3.0);
-        let p2 = Point::new(1.5, 4.5, 2.5);
-        let v1 = Vector::new(-0.5, -2.5, 0.5);
-
-        assert_eq!(v1, p1 - p2);
-    }
-
-    #[test]
-    fn mul_matrix_1() {
-        let p = Point::new(1.0, 2.5, 4.0);
-        let m = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        let pm = Point::new(15.5, 54.0, 70.0);
-        assert_eq!(p * m, pm);
-    }
-
-    #[test]
-    fn mul_matrix_2() {
-        let p = Point::new(1.0, 2.5, 4.0);
-        let m = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        let pm = Point::new(15.5, 54.0, 70.0);
-        assert_eq!(p * m, pm);
-    }
-
-    #[test]
-    fn mul_matrix_3() {
-        // Make sure matrix multiplication composes the way one would expect
-        let p = Point::new(1.0, 2.5, 4.0);
-        let m1 = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        let m2 =
-            Transform::new_from_values(4.0, 1.0, 2.0, 3.5, 3.0, 6.0, 5.0, 2.0, 2.0, 2.0, 4.0, 12.0);
-        println!("{:?}", m1 * m2);
-
-        let pmm1 = p * (m1 * m2);
-        let pmm2 = (p * m1) * m2;
-
-        assert!((pmm1 - pmm2).length2() <= 0.00001); // Assert pmm1 and pmm2 are roughly equal
-    }
-}

sub_crates/math3d/src/transform.rs

@@ -1,178 +0,0 @@
-#![allow(dead_code)]
-
-use std::ops::{Add, Mul};
-
-use approx::relative_eq;
-use glam::{Affine3A, Mat3, Mat4, Vec3};
-
-use super::Point;
-
-/// A 4x3 affine transform matrix, used for transforms.
-#[derive(Debug, Copy, Clone, PartialEq)]
-pub struct Transform(pub Affine3A);
-
-impl Transform {
-    /// Creates a new identity matrix
-    #[inline]
-    pub fn new() -> Transform {
-        Transform(Affine3A::IDENTITY)
-    }
-
-    /// Creates a new matrix with the specified values:
-    /// a b c d
-    /// e f g h
-    /// i j k l
-    /// m n o p
-    #[inline]
-    #[allow(clippy::many_single_char_names)]
-    #[allow(clippy::too_many_arguments)]
-    pub fn new_from_values(
-        a: f32,
-        b: f32,
-        c: f32,
-        d: f32,
-        e: f32,
-        f: f32,
-        g: f32,
-        h: f32,
-        i: f32,
-        j: f32,
-        k: f32,
-        l: f32,
-    ) -> Transform {
-        Transform(Affine3A::from_mat3_translation(
-            Mat3::from_cols(Vec3::new(a, e, i), Vec3::new(b, f, j), Vec3::new(c, g, k)),
-            Vec3::new(d, h, l),
-        ))
-    }
-
-    #[inline]
-    pub fn from_location(loc: Point) -> Transform {
-        Transform(Affine3A::from_translation(loc.co.into()))
-    }
-
-    /// Returns whether the matrices are approximately equal to each other.
-    /// Each corresponding element in the matrices cannot have a relative
-    /// error exceeding epsilon.
-    #[inline]
-    pub fn aprx_eq(&self, other: Transform, epsilon: f32) -> bool {
-        let mut eq = true;
-        for c in 0..3 {
-            for r in 0..3 {
-                let a = self.0.matrix3.col(c)[r];
-                let b = other.0.matrix3.col(c)[r];
-                eq &= relative_eq!(a, b, epsilon = epsilon);
-            }
-        }
-        for i in 0..3 {
-            let a = self.0.translation[i];
-            let b = other.0.translation[i];
-            eq &= relative_eq!(a, b, epsilon = epsilon);
-        }
-        eq
-    }
-
-    /// Returns the inverse of the Matrix
-    #[inline]
-    pub fn inverse(&self) -> Transform {
-        Transform(self.0.inverse())
-    }
-}
-
-impl Default for Transform {
-    fn default() -> Self {
-        Self::new()
-    }
-}
-
-/// Multiply two matrices together
-impl Mul for Transform {
-    type Output = Self;
-
-    #[inline]
-    fn mul(self, other: Self) -> Self {
-        Self(other.0 * self.0)
-    }
-}
-
-/// Multiply a matrix by a f32
-impl Mul<f32> for Transform {
-    type Output = Self;
-
-    #[inline]
-    fn mul(self, other: f32) -> Self {
-        Self(Affine3A::from_mat4(Mat4::from(self.0) * other))
-    }
-}
-
-/// Add two matrices together
-impl Add for Transform {
-    type Output = Self;
-
-    #[inline]
-    fn add(self, other: Self) -> Self {
-        Self(Affine3A::from_mat4(
-            Mat4::from(self.0) + Mat4::from(other.0),
-        ))
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn equality_test() {
-        let a = Transform::new();
-        let b = Transform::new();
-        let c =
-            Transform::new_from_values(1.1, 0.0, 0.0, 0.0, 0.0, 1.1, 0.0, 0.0, 0.0, 0.0, 1.1, 0.0);
-
-        assert_eq!(a, b);
-        assert!(a != c);
-    }
-
-    #[test]
-    fn approximate_equality_test() {
-        let a = Transform::new();
-        let b = Transform::new_from_values(
-            1.000001, 0.0, 0.0, 0.0, 0.0, 1.000001, 0.0, 0.0, 0.0, 0.0, 1.000001, 0.0,
-        );
-        let c = Transform::new_from_values(
-            1.000003, 0.0, 0.0, 0.0, 0.0, 1.000003, 0.0, 0.0, 0.0, 0.0, 1.000003, 0.0,
-        );
-        let d = Transform::new_from_values(
-            -1.000001, 0.0, 0.0, 0.0, 0.0, -1.000001, 0.0, 0.0, 0.0, 0.0, -1.000001, 0.0,
-        );
-
-        assert!(a.aprx_eq(b, 0.000001));
-        assert!(!a.aprx_eq(c, 0.000001));
-        assert!(!a.aprx_eq(d, 0.000001));
-    }
-
-    #[test]
-    fn multiply_test() {
-        let a = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        let b = Transform::new_from_values(
-            1.0, 5.0, 9.0, 13.0, 2.0, 6.0, 10.0, 14.0, 3.0, 7.0, 11.0, 15.0,
-        );
-        let c = Transform::new_from_values(
-            97.0, 50.0, 136.0, 162.5, 110.0, 60.0, 156.0, 185.0, 123.0, 70.0, 176.0, 207.5,
-        );
-
-        assert_eq!(a * b, c);
-    }
-
-    #[test]
-    fn inverse_test() {
-        let a = Transform::new_from_values(
-            1.0, 0.33, 0.0, -2.0, 0.0, 1.0, 0.0, 0.0, 2.1, 0.7, 1.3, 0.0,
-        );
-        let b = a.inverse();
-        let c = Transform::new();
-
-        assert!((dbg!(a * b)).aprx_eq(dbg!(c), 0.0000001));
-    }
-}

sub_crates/math3d/src/vector.rs

@@ -1,286 +0,0 @@
-#![allow(dead_code)]
-
-use std::{
-    cmp::PartialEq,
-    ops::{Add, Div, Mul, Neg, Sub},
-};
-
-use glam::Vec3A;
-
-use super::{CrossProduct, DotProduct, Normal, Point, Transform};
-
-/// A direction vector in 3d homogeneous space.
-#[derive(Debug, Copy, Clone)]
-pub struct Vector {
-    pub co: Vec3A,
-}
-
-impl Vector {
-    #[inline(always)]
-    pub fn new(x: f32, y: f32, z: f32) -> Vector {
-        Vector {
-            co: Vec3A::new(x, y, z),
-        }
-    }
-
-    #[inline(always)]
-    pub fn length(&self) -> f32 {
-        self.co.length()
-    }
-
-    #[inline(always)]
-    pub fn length2(&self) -> f32 {
-        self.co.length_squared()
-    }
-
-    #[inline(always)]
-    pub fn normalized(&self) -> Vector {
-        Vector {
-            co: self.co.normalize(),
-        }
-    }
-
-    #[inline(always)]
-    pub fn abs(&self) -> Vector {
-        Vector {
-            co: self.co * self.co.signum(),
-        }
-    }
-
-    #[inline(always)]
-    pub fn into_point(self) -> Point {
-        Point { co: self.co }
-    }
-
-    #[inline(always)]
-    pub fn into_normal(self) -> Normal {
-        Normal { co: self.co }
-    }
-
-    #[inline(always)]
-    pub fn get_n(&self, n: usize) -> f32 {
-        match n {
-            0 => self.x(),
-            1 => self.y(),
-            2 => self.z(),
-            _ => panic!("Attempt to access dimension beyond z."),
-        }
-    }
-
-    #[inline(always)]
-    pub fn x(&self) -> f32 {
-        self.co[0]
-    }
-
-    #[inline(always)]
-    pub fn y(&self) -> f32 {
-        self.co[1]
-    }
-
-    #[inline(always)]
-    pub fn z(&self) -> f32 {
-        self.co[2]
-    }
-
-    #[inline(always)]
-    pub fn set_x(&mut self, x: f32) {
-        self.co[0] = x;
-    }
-
-    #[inline(always)]
-    pub fn set_y(&mut self, y: f32) {
-        self.co[1] = y;
-    }
-
-    #[inline(always)]
-    pub fn set_z(&mut self, z: f32) {
-        self.co[2] = z;
-    }
-}
-
-impl PartialEq for Vector {
-    #[inline(always)]
-    fn eq(&self, other: &Vector) -> bool {
-        self.co == other.co
-    }
-}
-
-impl Add for Vector {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn add(self, other: Vector) -> Vector {
-        Vector {
-            co: self.co + other.co,
-        }
-    }
-}
-
-impl Sub for Vector {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn sub(self, other: Vector) -> Vector {
-        Vector {
-            co: self.co - other.co,
-        }
-    }
-}
-
-impl Mul<f32> for Vector {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn mul(self, other: f32) -> Vector {
-        Vector {
-            co: self.co * other,
-        }
-    }
-}
-
-impl Mul<Transform> for Vector {
-    type Output = Vector;
-
-    #[inline]
-    fn mul(self, other: Transform) -> Vector {
-        Vector {
-            co: other.0.transform_vector3a(self.co),
-        }
-    }
-}
-
-impl Div<f32> for Vector {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn div(self, other: f32) -> Vector {
-        Vector {
-            co: self.co / other,
-        }
-    }
-}
-
-impl Neg for Vector {
-    type Output = Vector;
-
-    #[inline(always)]
-    fn neg(self) -> Vector {
-        Vector { co: self.co * -1.0 }
-    }
-}
-
-impl DotProduct for Vector {
-    #[inline(always)]
-    fn dot(self, other: Vector) -> f32 {
-        self.co.dot(other.co)
-    }
-}
-
-impl CrossProduct for Vector {
-    #[inline]
-    fn cross(self, other: Vector) -> Vector {
-        Vector {
-            co: self.co.cross(other.co),
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::super::{CrossProduct, DotProduct, Transform};
-    use super::*;
-
-    #[test]
-    fn add() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = Vector::new(1.5, 4.5, 2.5);
-        let v3 = Vector::new(2.5, 6.5, 5.5);
-
-        assert_eq!(v3, v1 + v2);
-    }
-
-    #[test]
-    fn sub() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = Vector::new(1.5, 4.5, 2.5);
-        let v3 = Vector::new(-0.5, -2.5, 0.5);
-
-        assert_eq!(v3, v1 - v2);
-    }
-
-    #[test]
-    fn mul_scalar() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = 2.0;
-        let v3 = Vector::new(2.0, 4.0, 6.0);
-
-        assert_eq!(v3, v1 * v2);
-    }
-
-    #[test]
-    fn mul_matrix_1() {
-        let v = Vector::new(1.0, 2.5, 4.0);
-        let m = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        assert_eq!(v * m, Vector::new(14.0, 46.0, 58.0));
-    }
-
-    #[test]
-    fn mul_matrix_2() {
-        let v = Vector::new(1.0, 2.5, 4.0);
-        let m = Transform::new_from_values(
-            1.0, 2.0, 2.0, 1.5, 3.0, 6.0, 7.0, 8.0, 9.0, 2.0, 11.0, 12.0,
-        );
-        assert_eq!(v * m, Vector::new(14.0, 46.0, 58.0));
-    }
-
-    #[test]
-    fn div() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = 2.0;
-        let v3 = Vector::new(0.5, 1.0, 1.5);
-
-        assert_eq!(v3, v1 / v2);
-    }
-
-    #[test]
-    fn length() {
-        let v = Vector::new(1.0, 2.0, 3.0);
-        assert!((v.length() - 3.7416573867739413).abs() < 0.000001);
-    }
-
-    #[test]
-    fn length2() {
-        let v = Vector::new(1.0, 2.0, 3.0);
-        assert_eq!(v.length2(), 14.0);
-    }
-
-    #[test]
-    fn normalized() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = Vector::new(0.2672612419124244, 0.5345224838248488, 0.8017837257372732);
-        let v3 = v1.normalized();
-        assert!((v3.x() - v2.x()).abs() < 0.000001);
-        assert!((v3.y() - v2.y()).abs() < 0.000001);
-        assert!((v3.z() - v2.z()).abs() < 0.000001);
-    }
-
-    #[test]
-    fn dot_test() {
-        let v1 = Vector::new(1.0, 2.0, 3.0);
-        let v2 = Vector::new(1.5, 4.5, 2.5);
-        let v3 = 18.0f32;
-
-        assert_eq!(v3, v1.dot(v2));
-    }
-
-    #[test]
-    fn cross_test() {
-        let v1 = Vector::new(1.0, 0.0, 0.0);
-        let v2 = Vector::new(0.0, 1.0, 0.0);
-        let v3 = Vector::new(0.0, 0.0, 1.0);
-
-        assert_eq!(v3, v1.cross(v2));
-    }
-}

sub_crates/rmath/Cargo.toml

@@ -1,15 +1,18 @@
 [package]
-name = "math3d"
+name = "rmath"
 version = "0.1.0"
 authors = ["Nathan Vegdahl <cessen@cessen.com>"]
 edition = "2018"
 license = "MIT, Apache 2.0"
 
 [lib]
-name = "math3d"
+name = "rmath"
 path = "src/lib.rs"
 
-# Local crate dependencies
-[dependencies]
-glam = "0.15"
-approx = "0.4"
+[dev-dependencies]
+bencher = "0.1.5"
+rand = "0.6"
+
+[[bench]]
+name = "bench"
+harness = false

sub_crates/rmath/benches/bench.rs

@@ -0,0 +1,202 @@
+use bencher::{benchmark_group, benchmark_main, black_box, Bencher};
+use rand::{rngs::SmallRng, FromEntropy, Rng};
+use rmath::{CrossProduct, DotProduct, Normal, Point, Vector, Xform, XformFull};
+
+//----
+
+fn vector_cross_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let v1 = Vector::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let v2 = Vector::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        for _ in 0..10000 {
+            black_box(black_box(v1).cross(black_box(v2)));
+        }
+    });
+}
+
+fn vector_dot_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let v1 = Vector::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let v2 = Vector::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        for _ in 0..10000 {
+            black_box(black_box(v1).dot(black_box(v2)));
+        }
+    });
+}
+
+fn xform_vector_mul_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let v = Vector::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let x = Xform::new(
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        );
+        for _ in 0..10000 {
+            black_box(black_box(v).xform(black_box(&x)));
+        }
+    });
+}
+
+fn xform_point_mul_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let p = Point::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let x = Xform::new(
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        );
+        for _ in 0..10000 {
+            black_box(black_box(p).xform(black_box(&x)));
+        }
+    });
+}
+
+fn xform_point_mul_inv_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let p = Point::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let x = Xform::new(
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        )
+        .to_full()
+        .unwrap();
+        for _ in 0..10000 {
+            black_box(black_box(p).xform_inv(black_box(&x)));
+        }
+    });
+}
+
+fn xform_normal_mul_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let n = Normal::new(rng.gen::<f32>(), rng.gen::<f32>(), rng.gen::<f32>());
+        let x = Xform::new(
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            1.0,
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        )
+        .to_full()
+        .unwrap();
+        for _ in 0..10000 {
+            black_box(black_box(n).xform(black_box(&x)));
+        }
+    });
+}
+
+fn xform_xform_mul_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let x1 = Xform::new(
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        );
+        let x2 = Xform::new(
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        );
+        for _ in 0..10000 {
+            black_box(black_box(x1).compose(black_box(&x2)));
+        }
+    });
+}
+
+fn xform_to_xformfull_10000(bench: &mut Bencher) {
+    let mut rng = SmallRng::from_entropy();
+    bench.iter(|| {
+        let x = Xform::new(
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+            rng.gen::<f32>(),
+        );
+        for _ in 0..10000 {
+            black_box(black_box(x).to_full());
+        }
+    });
+}
+
+//----
+
+benchmark_group!(
+    benches,
+    vector_cross_10000,
+    vector_dot_10000,
+    xform_vector_mul_10000,
+    xform_point_mul_10000,
+    xform_point_mul_inv_10000,
+    xform_normal_mul_10000,
+    xform_xform_mul_10000,
+    xform_to_xformfull_10000,
+);
+benchmark_main!(benches);

sub_crates/rmath/examples/precision.rs

@@ -0,0 +1,268 @@
+use rand::{rngs::SmallRng, FromEntropy, Rng};
+
+use rmath::{utils::ulp_diff, wide4::Float4};
+
+type D4 = [f64; 4];
+
+fn main() {
+    let mut rng = SmallRng::from_entropy();
+
+    // Convenience functions for generating random Float4's.
+    let mut rf4 = || {
+        let mut rf = || {
+            let range = 268435456.0;
+            let n = rng.gen::<f64>();
+            ((n * range * 2.0) - range) as f32
+        };
+        Float4::new(rf(), rf(), rf(), rf())
+    };
+
+    // Dot product test.
+    println!("Dot product:");
+    {
+        let mut max_ulp_diff = 0u32;
+        for _ in 0..10000000 {
+            let v1 = rf4();
+            let v2 = rf4();
+
+            let dpa = Float4::dot_3(v1, v2);
+            let dpb = dot_3(f4_to_d4(v1), f4_to_d4(v2));
+
+            let ud = ulp_diff(dpa, dpb as f32);
+            max_ulp_diff = max_ulp_diff.max(ud);
+        }
+
+        println!("  Max error (ulps):\n    {:?}\n", max_ulp_diff);
+    }
+
+    // Cross product test.
+    println!("Cross product:");
+    {
+        let mut max_ulp_diff = [0u32; 4];
+        for _ in 0..10000000 {
+            let v1 = rf4();
+            let v2 = rf4();
+
+            let v3a = Float4::cross_3(v1, v2);
+            let v3b = cross_3(f4_to_d4(v1), f4_to_d4(v2));
+
+            let ud = ulp_diff_f4d4(v3a, v3b);
+            for i in 0..4 {
+                max_ulp_diff[i] = max_ulp_diff[i].max(ud[i]);
+            }
+        }
+
+        println!("  Max error (ulps):\n    {:?}\n", max_ulp_diff);
+    }
+
+    // Matrix inversion test.
+    println!("Matrix inversion:");
+    {
+        let mut max_ulp_diff = [[0u32; 4]; 3];
+        let mut det_ulp_hist = [0u32; 9];
+        for _ in 0..2000000 {
+            let m = [rf4(), rf4(), rf4()];
+            let ima = Float4::invert_3x3_w_det(&m);
+            let imb = invert_3x3([f4_to_d4(m[0]), f4_to_d4(m[1]), f4_to_d4(m[2])]);
+
+            if let (Some((ima, deta)), Some((imb, detb))) = (ima, imb) {
+                let det_ulp_diff = ulp_diff(deta, detb as f32);
+                let mut hist_upper = 0;
+                for i in 0..det_ulp_hist.len() {
+                    if det_ulp_diff <= hist_upper {
+                        det_ulp_hist[i] += 1;
+                        break;
+                    }
+                    if hist_upper == 0 {
+                        hist_upper += 1;
+                    } else {
+                        hist_upper *= 10;
+                    }
+                }
+
+                if det_ulp_diff == 0 {
+                    for i in 0..3 {
+                        let ud = ulp_diff_f4d4(ima[i], imb[i]);
+                        for j in 0..4 {
+                            max_ulp_diff[i][j] = max_ulp_diff[i][j].max(ud[j]);
+                        }
+                    }
+                }
+            }
+        }
+
+        println!(
+            "  Max error when determinant has 0-ulp error (ulps):\n    {:?}",
+            max_ulp_diff
+        );
+
+        let total: u32 = det_ulp_hist.iter().sum();
+        let mut ulp = 0;
+        let mut sum = 0;
+        println!("  Determinant error distribution:");
+        for h in det_ulp_hist.iter() {
+            sum += *h;
+            println!(
+                "    {:.8}% <= {} ulps",
+                sum as f64 / total as f64 * 100.0,
+                ulp
+            );
+            if ulp == 0 {
+                ulp += 1;
+            } else {
+                ulp *= 10;
+            }
+        }
+        println!();
+    }
+}
+
+//-------------------------------------------------------------
+
+fn f4_to_d4(v: Float4) -> D4 {
+    [v.a() as f64, v.b() as f64, v.c() as f64, v.d() as f64]
+}
+
+fn ulp_diff_f4d4(a: Float4, b: D4) -> [u32; 4] {
+    [
+        ulp_diff(a.a(), b[0] as f32),
+        ulp_diff(a.b(), b[1] as f32),
+        ulp_diff(a.c(), b[2] as f32),
+        ulp_diff(a.d(), b[3] as f32),
+    ]
+}
+
+//-------------------------------------------------------------
+
+fn dot_3(a: D4, b: D4) -> f64 {
+    // Products.
+    let (x, x_err) = two_prod(a[0], b[0]);
+    let (y, y_err) = two_prod(a[1], b[1]);
+    let (z, z_err) = two_prod(a[2], b[2]);
+
+    // Sums.
+    let (s1, s1_err) = two_sum(x, y);
+    let err1 = x_err + (y_err + s1_err);
+
+    let (s2, s2_err) = two_sum(s1, z);
+    let err2 = z_err + (err1 + s2_err);
+
+    // Final result with rounding error compensation.
+    s2 + err2
+}
+
+fn cross_3(a: D4, b: D4) -> D4 {
+    [
+        difference_of_products(a[1], b[2], a[2], b[1]),
+        difference_of_products(a[2], b[0], a[0], b[2]),
+        difference_of_products(a[0], b[1], a[1], b[0]),
+        difference_of_products(a[3], b[3], a[3], b[3]),
+    ]
+}
+
+fn invert_3x3(m: [D4; 3]) -> Option<([D4; 3], f64)> {
+    let m0_bca = [m[0][1], m[0][2], m[0][0], m[0][3]];
+    let m1_bca = [m[1][1], m[1][2], m[1][0], m[1][3]];
+    let m2_bca = [m[2][1], m[2][2], m[2][0], m[2][3]];
+    let m0_cab = [m[0][2], m[0][0], m[0][1], m[0][3]];
+    let m1_cab = [m[1][2], m[1][0], m[1][1], m[1][3]];
+    let m2_cab = [m[2][2], m[2][0], m[2][1], m[2][3]];
+
+    let abc = [
+        difference_of_products(m1_bca[0], m2_cab[0], m1_cab[0], m2_bca[0]),
+        difference_of_products(m1_bca[1], m2_cab[1], m1_cab[1], m2_bca[1]),
+        difference_of_products(m1_bca[2], m2_cab[2], m1_cab[2], m2_bca[2]),
+        difference_of_products(m1_bca[3], m2_cab[3], m1_cab[3], m2_bca[3]),
+    ];
+    let def = [
+        difference_of_products(m2_bca[0], m0_cab[0], m2_cab[0], m0_bca[0]),
+        difference_of_products(m2_bca[1], m0_cab[1], m2_cab[1], m0_bca[1]),
+        difference_of_products(m2_bca[2], m0_cab[2], m2_cab[2], m0_bca[2]),
+        difference_of_products(m2_bca[3], m0_cab[3], m2_cab[3], m0_bca[3]),
+    ];
+    let ghi = [
+        difference_of_products(m0_bca[0], m1_cab[0], m0_cab[0], m1_bca[0]),
+        difference_of_products(m0_bca[1], m1_cab[1], m0_cab[1], m1_bca[1]),
+        difference_of_products(m0_bca[2], m1_cab[2], m0_cab[2], m1_bca[2]),
+        difference_of_products(m0_bca[3], m1_cab[3], m0_cab[3], m1_bca[3]),
+    ];
+
+    let det = dot_3(
+        [abc[0], def[0], ghi[0], 0.0],
+        [m[0][0], m[1][0], m[2][0], 0.0],
+    );
+
+    if det == 0.0 {
+        None
+    } else {
+        Some((
+            [
+                [abc[0] / det, def[0] / det, ghi[0] / det, 0.0],
+                [abc[1] / det, def[1] / det, ghi[1] / det, 0.0],
+                [abc[2] / det, def[2] / det, ghi[2] / det, 0.0],
+            ],
+            // [
+            //     [abc[0], def[0], ghi[0], 0.0],
+            //     [abc[1], def[1], ghi[1], 0.0],
+            //     [abc[2], def[2], ghi[2], 0.0],
+            // ],
+            det,
+        ))
+    }
+}
+
+fn rel_diff(a: f64, b: f64) -> f64 {
+    (a - b).abs() / a.abs().max(b.abs())
+}
+
+//-------------------------------------------------------------
+
+/// `(a * b) - (c * d)` but done with high precision via floating point tricks.
+///
+/// See https://pharr.org/matt/blog/2019/11/03/difference-of-floats
+#[inline(always)]
+fn difference_of_products(a: f64, b: f64, c: f64, d: f64) -> f64 {
+    let cd = c * d;
+    let dop = a.mul_add(b, -cd);
+    let err = (-c).mul_add(d, cd);
+    dop + err
+}
+
+/// `(a * b) + (c * d)` but done with high precision via floating point tricks.
+#[inline(always)]
+fn sum_of_products(a: f64, b: f64, c: f64, d: f64) -> f64 {
+    let cd = c * d;
+    let sop = a.mul_add(b, cd);
+    let err = c.mul_add(d, -cd);
+    sop + err
+}
+
+/// `a * b` but also returns a rounding error for precise composition
+/// with other operations.
+#[inline(always)]
+fn two_prod(a: f64, b: f64) -> (f64, f64)
+// (product, rounding_err)
+{
+    let ab = a * b;
+    (ab, a.mul_add(b, -ab))
+}
+
+/// `a + b` but also returns a rounding error for precise composition
+/// with other operations.
+#[inline(always)]
+fn two_sum(a: f64, b: f64) -> (f64, f64)
+// (sum, rounding_err)
+{
+    let sum = a + b;
+    let delta = sum - a;
+    (sum, (a - (sum - delta)) + (b - delta))
+}
+
+#[inline(always)]
+fn two_diff(a: f64, b: f64) -> (f64, f64)
+// (diff, rounding_err)
+{
+    let diff = a - b;
+    let delta = diff - a;
+    (diff, (a - (diff - delta)) - (b + delta))
+}

sub_crates/rmath/src/lib.rs

@@ -0,0 +1,129 @@
+//! RMath: a math library for building CPU-based renderers.
+
+#![allow(dead_code)]
+
+mod normal;
+mod point;
+mod sealed;
+pub mod utils;
+mod vector;
+pub mod wide4;
+mod xform;
+
+use std::ops::{Add, Mul, Neg, Sub};
+
+pub use self::{
+    normal::Normal, point::Point, vector::Vector, xform::AsXform, xform::Xform, xform::XformFull,
+};
+
+/// Trait for calculating dot products.
+pub trait DotProduct {
+    fn dot(self, other: Self) -> f32;
+
+    fn dot_fast(self, other: Self) -> f32;
+}
+
+#[inline(always)]
+pub fn dot<T: DotProduct>(a: T, b: T) -> f32 {
+    a.dot(b)
+}
+
+#[inline(always)]
+pub fn dot_fast<T: DotProduct>(a: T, b: T) -> f32 {
+    a.dot_fast(b)
+}
+
+/// Trait for calculating cross products.
+pub trait CrossProduct {
+    fn cross(self, other: Self) -> Self;
+
+    fn cross_fast(self, other: Self) -> Self;
+}
+
+#[inline(always)]
+pub fn cross<T: CrossProduct>(a: T, b: T) -> T {
+    a.cross(b)
+}
+
+#[inline(always)]
+pub fn cross_fast<T: CrossProduct>(a: T, b: T) -> T {
+    a.cross_fast(b)
+}
+
+//-------------------------------------------------------------
+
+/// Trait representing types that can do fused multiply-add.
+trait FMulAdd {
+    /// `(self * b) + c` with only one floating point rounding error.
+    fn fma(self, b: Self, c: Self) -> Self;
+}
+
+impl FMulAdd for f32 {
+    fn fma(self, b: Self, c: Self) -> Self {
+        self.mul_add(b, c)
+    }
+}
+
+/// `(a * b) - (c * d)` but done with high precision via floating point tricks.
+///
+/// See https://pharr.org/matt/blog/2019/11/03/difference-of-floats
+#[inline(always)]
+fn difference_of_products<T>(a: T, b: T, c: T, d: T) -> T
+where
+    T: Copy + FMulAdd + Add<Output = T> + Mul<Output = T> + Neg<Output = T>,
+{
+    let cd = c * d;
+    let dop = a.fma(b, -cd);
+    let err = (-c).fma(d, cd);
+    dop + err
+}
+
+/// `(a * b) + (c * d)` but done with high precision via floating point tricks.
+#[inline(always)]
+fn sum_of_products<T>(a: T, b: T, c: T, d: T) -> T
+where
+    T: Copy + FMulAdd + Add<Output = T> + Mul<Output = T> + Neg<Output = T>,
+{
+    let cd = c * d;
+    let sop = a.fma(b, cd);
+    let err = c.fma(d, -cd);
+    sop + err
+}
+
+/// `a * b` but also returns a rounding error for precise composition
+/// with other operations.
+#[inline(always)]
+fn two_prod<T>(a: T, b: T) -> (T, T)
+// (product, rounding_err)
+where
+    T: Copy + FMulAdd + Mul<Output = T> + Neg<Output = T>,
+{
+    let ab = a * b;
+    (ab, a.fma(b, -ab))
+}
+
+/// `a + b` but also returns a rounding error for precise composition
+/// with other operations.
+#[inline(always)]
+fn two_sum<T>(a: T, b: T) -> (T, T)
+// (sum, rounding_err)
+where
+    T: Copy + Add<Output = T> + Sub<Output = T>,
+{
+    let sum = a + b;
+    let delta = sum - a;
+    (sum, (a - (sum - delta)) + (b - delta))
+}
+
+/// `a - b` but also returns a rounding error for precise composition
+/// with other operations.
+#[inline(always)]
+fn two_diff<T>(a: T, b: T) -> (T, T)
+// (diff, rounding_err)
+where
+    T: Copy + Add<Output = T> + Sub<Output = T>,
+{
+    let diff = a - b;
+    let delta = diff - a;
+    (diff, (a - (diff - delta)) - (b + delta))
+}

sub_crates/rmath/src/normal.rs

@@ -0,0 +1,318 @@
+#![allow(dead_code)]
+
+use std::cmp::PartialEq;
+use std::ops::{Add, Div, Mul, Neg, Sub};
+
+use crate::wide4::Float4;
+use crate::xform::{AsXform, XformFull};
+use crate::Vector;
+use crate::{CrossProduct, DotProduct};
+
+/// A surface normal in 3D space.
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Normal(pub Float4);
+
+impl Normal {
+    #[inline(always)]
+    pub fn new(x: f32, y: f32, z: f32) -> Self {
+        Self(Float4::new(x, y, z, 0.0))
+    }
+
+    #[inline(always)]
+    pub fn length(self) -> f32 {
+        self.length2().sqrt()
+    }
+
+    #[inline(always)]
+    pub fn length2(self) -> f32 {
+        let sqr = self.0 * self.0;
+        sqr.a() + sqr.b() + sqr.c()
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn normalized(self) -> Self {
+        Self(self.0 / self.length())
+    }
+
+    #[inline(always)]
+    pub fn into_vector(self) -> Vector {
+        Vector(self.0)
+    }
+
+    #[inline(always)]
+    pub fn x(self) -> f32 {
+        self.0.a()
+    }
+
+    #[inline(always)]
+    pub fn y(self) -> f32 {
+        self.0.b()
+    }
+
+    #[inline(always)]
+    pub fn z(self) -> f32 {
+        self.0.c()
+    }
+
+    #[inline(always)]
+    pub fn get_n(self, i: usize) -> f32 {
+        match i {
+            0 => self.x(),
+            1 => self.y(),
+            2 => self.z(),
+            _ => panic!("Out of bounds index into 3D vector."),
+        }
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_x(self, x: f32) -> Self {
+        Self(self.0.set_a(x))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_y(self, y: f32) -> Self {
+        Self(self.0.set_b(y))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_z(self, z: f32) -> Self {
+        Self(self.0.set_c(z))
+    }
+
+    //-------------
+    // Transforms.
+
+    /// Forward-transform the normal.
+    #[inline(always)]
+    pub fn xform(self, xform: &XformFull) -> Self {
+        Self(self.0.vec_mul_3x3(&Float4::transpose_3x3(&xform.inv_m)))
+    }
+
+    /// Inverse-transform the normal.
+    #[inline(always)]
+    pub fn xform_inv<T: AsXform>(self, xform: &T) -> Self {
+        Self(
+            self.0
+                .vec_mul_3x3(&Float4::transpose_3x3(&xform.as_xform().m)),
+        )
+    }
+
+    /// Faster but less precise version of `xform()`.
+    #[inline(always)]
+    pub fn xform_fast(self, xform: &XformFull) -> Self {
+        Self(
+            self.0
+                .vec_mul_3x3_fast(&Float4::transpose_3x3(&xform.inv_m)),
+        )
+    }
+
+    /// Faster but less precise version of `xform_inv()`.
+    #[inline(always)]
+    pub fn xform_inv_fast<T: AsXform>(self, xform: &T) -> Self {
+        Self(
+            self.0
+                .vec_mul_3x3_fast(&Float4::transpose_3x3(&xform.as_xform().m)),
+        )
+    }
+}
+
+impl Add for Normal {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, other: Self) -> Self {
+        Self(self.0 + other.0)
+    }
+}
+
+impl Sub for Normal {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, other: Self) -> Self {
+        Self(self.0 - other.0)
+    }
+}
+
+impl Mul<f32> for Normal {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, other: f32) -> Self {
+        Self(self.0 * other)
+    }
+}
+
+impl Div<f32> for Normal {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, other: f32) -> Self {
+        Self(self.0 / other)
+    }
+}
+
+impl Neg for Normal {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self(-self.0)
+    }
+}
+
+impl PartialEq for Normal {
+    #[inline(always)]
+    fn eq(&self, rhs: &Self) -> bool {
+        self.0.a() == rhs.0.a() && self.0.b() == rhs.0.b() && self.0.c() == rhs.0.c()
+    }
+}
+
+impl DotProduct for Normal {
+    #[inline(always)]
+    fn dot(self, other: Self) -> f32 {
+        Float4::dot_3(self.0, other.0)
+    }
+
+    #[inline(always)]
+    fn dot_fast(self, other: Self) -> f32 {
+        Float4::dot_3_fast(self.0, other.0)
+    }
+}
+
+impl CrossProduct for Normal {
+    #[inline(always)]
+    fn cross(self, other: Self) -> Self {
+        Self(Float4::cross_3(self.0, other.0))
+    }
+
+    #[inline(always)]
+    fn cross_fast(self, other: Self) -> Self {
+        Self(Float4::cross_3_fast(self.0, other.0))
+    }
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{CrossProduct, DotProduct, Xform};
+
+    #[test]
+    fn add() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = Normal::new(1.5, 4.5, 2.5);
+        let v3 = Normal::new(2.5, 6.5, 5.5);
+
+        assert_eq!(v3, v1 + v2);
+    }
+
+    #[test]
+    fn sub() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = Normal::new(1.5, 4.5, 2.5);
+        let v3 = Normal::new(-0.5, -2.5, 0.5);
+
+        assert_eq!(v3, v1 - v2);
+    }
+
+    #[test]
+    fn mul_scalar() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = 2.0;
+        let v3 = Normal::new(2.0, 4.0, 6.0);
+
+        assert_eq!(v3, v1 * v2);
+    }
+
+    #[test]
+    fn xform() {
+        let n = Normal::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+
+        assert_eq!(n.xform(&m), Normal::new(-4.0625, 1.78125, -0.03125));
+        assert_eq!(n.xform(&m).xform_inv(&m), n);
+    }
+
+    #[test]
+    fn xform_fast() {
+        let n = Normal::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+
+        assert_eq!(n.xform_fast(&m), Normal::new(-4.0625, 1.78125, -0.03125));
+        assert_eq!(n.xform_fast(&m).xform_inv_fast(&m), n);
+    }
+
+    #[test]
+    fn div() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = 2.0;
+        let v3 = Normal::new(0.5, 1.0, 1.5);
+
+        assert_eq!(v3, v1 / v2);
+    }
+
+    #[test]
+    fn length() {
+        let n = Normal::new(1.0, 2.0, 3.0);
+        assert!((n.length() - 3.7416573867739413).abs() < 0.000001);
+    }
+
+    #[test]
+    fn length2() {
+        let n = Normal::new(1.0, 2.0, 3.0);
+        assert_eq!(n.length2(), 14.0);
+    }
+
+    #[test]
+    fn normalized() {
+        let n1 = Normal::new(1.0, 2.0, 3.0);
+        let n2 = Normal::new(0.2672612419124244, 0.5345224838248488, 0.8017837257372732);
+        let n3 = n1.normalized();
+        assert!((n3.x() - n2.x()).abs() < 0.000001);
+        assert!((n3.y() - n2.y()).abs() < 0.000001);
+        assert!((n3.z() - n2.z()).abs() < 0.000001);
+    }
+
+    #[test]
+    fn dot() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = Normal::new(1.5, 4.5, 2.5);
+
+        assert_eq!(v1.dot(v2), 18.0);
+    }
+
+    #[test]
+    fn dot_fast() {
+        let v1 = Normal::new(1.0, 2.0, 3.0);
+        let v2 = Normal::new(1.5, 4.5, 2.5);
+
+        assert_eq!(v1.dot_fast(v2), 18.0);
+    }
+
+    #[test]
+    fn cross() {
+        let v1 = Normal::new(1.0, 0.0, 0.0);
+        let v2 = Normal::new(0.0, 1.0, 0.0);
+
+        assert_eq!(v1.cross(v2), Normal::new(0.0, 0.0, 1.0));
+    }
+
+    #[test]
+    fn cross_fast() {
+        let v1 = Normal::new(1.0, 0.0, 0.0);
+        let v2 = Normal::new(0.0, 1.0, 0.0);
+
+        assert_eq!(v1.cross_fast(v2), Normal::new(0.0, 0.0, 1.0));
+    }
+}

sub_crates/rmath/src/point.rs

@@ -0,0 +1,186 @@
+#![allow(dead_code)]
+use std::cmp::PartialEq;
+use std::ops::{Add, Sub};
+
+use crate::vector::Vector;
+use crate::wide4::Float4;
+use crate::xform::{AsXform, XformFull};
+
+/// A position in 3D space.
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Point(pub Float4);
+
+impl Point {
+    #[inline(always)]
+    pub fn new(x: f32, y: f32, z: f32) -> Self {
+        Self(Float4::new(x, y, z, 0.0))
+    }
+
+    #[inline(always)]
+    pub fn min(self, other: Self) -> Self {
+        Self(self.0.min(other.0))
+    }
+
+    #[inline(always)]
+    pub fn max(self, other: Self) -> Self {
+        Self(self.0.max(other.0))
+    }
+
+    #[inline(always)]
+    pub fn into_vector(self) -> Vector {
+        Vector(self.0)
+    }
+
+    #[inline(always)]
+    pub fn x(self) -> f32 {
+        self.0.a()
+    }
+
+    #[inline(always)]
+    pub fn y(self) -> f32 {
+        self.0.b()
+    }
+
+    #[inline(always)]
+    pub fn z(self) -> f32 {
+        self.0.c()
+    }
+
+    #[inline(always)]
+    pub fn get_n(self, i: usize) -> f32 {
+        match i {
+            0 => self.x(),
+            1 => self.y(),
+            2 => self.z(),
+            _ => panic!("Out of bounds index into 3D vector."),
+        }
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_x(self, x: f32) -> Self {
+        Self(self.0.set_a(x))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_y(self, y: f32) -> Self {
+        Self(self.0.set_b(y))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_z(self, z: f32) -> Self {
+        Self(self.0.set_c(z))
+    }
+
+    //-------------
+    // Transforms.
+
+    /// Forward-transform the point.
+    #[inline(always)]
+    pub fn xform<T: AsXform>(self, xform: &T) -> Self {
+        let xform = xform.as_xform();
+        Self(self.0.vec_mul_affine(&xform.m, xform.t))
+    }
+
+    /// Inverse-transform the point.
+    #[inline(always)]
+    pub fn xform_inv(self, xform: &XformFull) -> Self {
+        Self(self.0.vec_mul_affine_rev(&xform.inv_m, xform.fwd.t))
+    }
+
+    /// Faster but less precise version of `xform()`.
+    #[inline(always)]
+    pub fn xform_fast<T: AsXform>(self, xform: &T) -> Self {
+        let xform = xform.as_xform();
+        Self(self.0.vec_mul_affine_fast(&xform.m, xform.t))
+    }
+
+    /// Faster but less precise version of `xform_inv()`.
+    #[inline(always)]
+    pub fn xform_inv_fast(self, xform: &XformFull) -> Self {
+        Self(self.0.vec_mul_affine_rev_fast(&xform.inv_m, xform.fwd.t))
+    }
+}
+
+impl Add<Vector> for Point {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, other: Vector) -> Self {
+        Self(self.0 + other.0)
+    }
+}
+
+impl Sub for Point {
+    type Output = Vector;
+
+    #[inline(always)]
+    fn sub(self, other: Self) -> Vector {
+        Vector(self.0 - other.0)
+    }
+}
+
+impl Sub<Vector> for Point {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, other: Vector) -> Self {
+        Self(self.0 - other.0)
+    }
+}
+
+impl PartialEq for Point {
+    #[inline(always)]
+    fn eq(&self, rhs: &Self) -> bool {
+        self.0.a() == rhs.0.a() && self.0.b() == rhs.0.b() && self.0.c() == rhs.0.c()
+    }
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{Vector, Xform};
+
+    #[test]
+    fn add() {
+        let p1 = Point::new(1.0, 2.0, 3.0);
+        let v1 = Vector::new(1.5, 4.5, 2.5);
+        let p2 = Point::new(2.5, 6.5, 5.5);
+
+        assert_eq!(p2, p1 + v1);
+    }
+
+    #[test]
+    fn sub() {
+        let p1 = Point::new(1.0, 2.0, 3.0);
+        let p2 = Point::new(1.5, 4.5, 2.5);
+        let v1 = Vector::new(-0.5, -2.5, 0.5);
+
+        assert_eq!(v1, p1 - p2);
+    }
+
+    #[test]
+    fn xform() {
+        let p = Point::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+        assert_eq!(p.xform(&m), Point::new(15.5, 54.0, 70.0));
+        assert_eq!(p.xform(&m).xform_inv(&m), p);
+    }
+
+    #[test]
+    fn xform_fast() {
+        let p = Point::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+        assert_eq!(p.xform_fast(&m), Point::new(15.5, 54.0, 70.0));
+        assert_eq!(p.xform_fast(&m).xform_inv_fast(&m), p);
+    }
+}

sub_crates/rmath/src/sealed.rs

@@ -0,0 +1,5 @@
+/// For sealing other traits.
+///
+/// Even though this is marked as public, this module isn't, and
+/// therefore this trait is not available outside the crate.
+pub trait Sealed {}

sub_crates/rmath/src/utils.rs

@@ -0,0 +1,196 @@
+/// Compute how different two floats are in ulps.
+///
+/// Notes:
+/// - Treats 0.0 and -0.0 as zero ulps apart, and extends the
+///   implications of that to the rest of the numbers.  E.g. the numbers
+///   just before and after 0.0/-0.0 are only two ulps apart, not three.
+/// - Infinity is one ulp past float max, and converse for -infinity.
+/// - If either number is NaN, returns `u32::MAX`.
+#[inline(always)]
+pub fn ulp_diff(a: f32, b: f32) -> u32 {
+    const SIGN_BIT: u32 = 1 << 31;
+    const INFINITY: u32 = 0x7f800000;
+
+    let a = a.to_bits();
+    let b = b.to_bits();
+
+    let a_sign = a & SIGN_BIT;
+    let b_sign = b & SIGN_BIT;
+    let a_abs = a & !SIGN_BIT;
+    let b_abs = b & !SIGN_BIT;
+
+    if a_abs > INFINITY || b_abs > INFINITY {
+        // NaNs always return maximum ulps apart.
+        u32::MAX
+    } else if a_sign == b_sign {
+        a_abs.max(b_abs) - a_abs.min(b_abs)
+    } else {
+        a_abs + b_abs
+    }
+}
+
+/// Checks if two floats are approximately equal, within `max_ulps`.
+#[inline(always)]
+pub fn ulps_eq(a: f32, b: f32, max_ulps: u32) -> bool {
+    // The minimum ensures that NaNs never return true.
+    ulp_diff(a, b) <= max_ulps.min(u32::MAX - 1)
+}
+
+/// Increments to the next representable floating point number.
+///
+/// Notes:
+/// - 0.0 and -0.0 are treated as the same value.  E.g. starting from the
+///   number just before -0.0, it only takes two increments to get to the
+///   number just after 0.0.
+/// - Infinity, NaN, and their negative counterparts are returned
+///   unchanged.
+/// - Incrementing `f32::MAX` results in infinity.
+#[inline(always)]
+pub fn increment_ulp(v: f32) -> f32 {
+    if v.is_finite() {
+        if v > 0.0 {
+            f32::from_bits(v.to_bits() + 1)
+        } else if v < -0.0 {
+            f32::from_bits(v.to_bits() - 1)
+        } else {
+            f32::from_bits(1)
+        }
+    } else {
+        // Infinity or NaN.
+        v
+    }
+}
+
+/// Decrements to the previous representable floating point number.
+///
+/// Notes:
+/// - 0.0 and -0.0 are treated as the same value.  E.g. starting from the
+///   number just after 0.0, it only takes two decrements to get to the
+///   number just before -0.0.
+/// - Infinity, NaN, and their negative counterparts are returned
+///   unchanged.
+/// - Decrementing `-f32::MAX` results in -infinity.
+#[inline(always)]
+pub fn decrement_ulp(v: f32) -> f32 {
+    if v.is_finite() {
+        if v > 0.0 {
+            f32::from_bits(v.to_bits() - 1)
+        } else if v < -0.0 {
+            f32::from_bits(v.to_bits() + 1)
+        } else {
+            f32::from_bits(0x80000001)
+        }
+    } else {
+        // Infinity or NaN.
+        v
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn ulp_diff_test() {
+        assert_eq!(ulp_diff(1.0, 1.0), 0);
+        assert_eq!(ulp_diff(0.0, 0.0), 0);
+        assert_eq!(ulp_diff(0.0, -0.0), 0);
+        assert_eq!(ulp_diff(1.0, 2.0), 1 << 23);
+        assert_eq!(ulp_diff(2.0, 4.0), 1 << 23);
+        assert_eq!(ulp_diff(-1.0, -2.0), 1 << 23);
+        assert_eq!(ulp_diff(-2.0, -4.0), 1 << 23);
+        assert_eq!(ulp_diff(-1.0, 1.0), 0x7f000000);
+        assert_eq!(ulp_diff(0.0, 1.0), 0x3f800000);
+        assert_eq!(ulp_diff(-0.0, 1.0), 0x3f800000);
+        assert_eq!(ulp_diff(0.0, -1.0), 0x3f800000);
+        assert_eq!(ulp_diff(-0.0, -1.0), 0x3f800000);
+        assert_eq!(ulp_diff(f32::INFINITY, -f32::INFINITY), 0xff000000);
+        assert_eq!(ulp_diff(f32::NAN, f32::NAN), 0xffffffff);
+        assert_eq!(ulp_diff(f32::NAN, 1.0), 0xffffffff);
+        assert_eq!(ulp_diff(1.0, f32::NAN), 0xffffffff);
+        assert_eq!(ulp_diff(-f32::NAN, 1.0), 0xffffffff);
+        assert_eq!(ulp_diff(1.0, -f32::NAN), 0xffffffff);
+        assert_eq!(ulp_diff(0.0, f32::from_bits(0.0f32.to_bits() + 1)), 1);
+        assert_eq!(ulp_diff(-0.0, f32::from_bits(0.0f32.to_bits() + 1)), 1);
+    }
+
+    #[test]
+    fn ulps_eq_test() {
+        assert!(ulps_eq(1.0, 1.0, 0));
+        assert!(ulps_eq(1.0, 1.0, 1));
+        assert!(ulps_eq(0.0, 0.0, 0));
+        assert!(ulps_eq(0.0, -0.0, 0));
+
+        assert!(ulps_eq(1.0, 2.0, 1 << 23));
+        assert!(!ulps_eq(1.0, 2.0, (1 << 23) - 1));
+
+        assert!(ulps_eq(0.0, f32::from_bits(0.0f32.to_bits() + 1), 1));
+        assert!(!ulps_eq(0.0, f32::from_bits(0.0f32.to_bits() + 1), 0));
+
+        assert!(ulps_eq(-0.0, f32::from_bits(0.0f32.to_bits() + 1), 1));
+        assert!(!ulps_eq(-0.0, f32::from_bits(0.0f32.to_bits() + 1), 0));
+
+        assert!(ulps_eq(std::f32::INFINITY, -std::f32::INFINITY, 0xff000000));
+        assert!(!ulps_eq(
+            std::f32::INFINITY,
+            -std::f32::INFINITY,
+            0xff000000 - 1
+        ));
+
+        assert!(!ulps_eq(std::f32::NAN, std::f32::NAN, 0));
+        assert!(!ulps_eq(-std::f32::NAN, -std::f32::NAN, 0));
+        assert!(!ulps_eq(std::f32::NAN, std::f32::NAN, u32::MAX));
+        assert!(!ulps_eq(std::f32::NAN, std::f32::INFINITY, 1 << 31));
+        assert!(!ulps_eq(std::f32::INFINITY, std::f32::NAN, 1 << 31));
+    }
+
+    #[test]
+    fn inc_ulp() {
+        assert!(increment_ulp(1.0) > 1.0);
+        assert!(increment_ulp(-1.0) > -1.0);
+        assert!(increment_ulp(0.0) > 0.0);
+        assert!(increment_ulp(0.0) > -0.0);
+        assert!(increment_ulp(-0.0) > 0.0);
+        assert!(increment_ulp(-0.0) > -0.0);
+        assert!(increment_ulp(f32::MAX) == f32::INFINITY);
+        assert!(increment_ulp(f32::INFINITY) == f32::INFINITY);
+        assert!(increment_ulp(-f32::INFINITY) == -f32::INFINITY);
+        assert!(increment_ulp(f32::NAN).is_nan());
+        assert!(increment_ulp(-f32::NAN).is_nan());
+    }
+
+    #[test]
+    fn dec_ulp() {
+        assert!(decrement_ulp(1.0) < 1.0);
+        assert!(decrement_ulp(-1.0) < -1.0);
+        assert!(decrement_ulp(0.0) < 0.0);
+        assert!(decrement_ulp(0.0) < -0.0);
+        assert!(decrement_ulp(-0.0) < 0.0);
+        assert!(decrement_ulp(-0.0) < -0.0);
+        assert!(decrement_ulp(f32::MIN) == -f32::INFINITY);
+        assert!(decrement_ulp(f32::INFINITY) == f32::INFINITY);
+        assert!(decrement_ulp(-f32::INFINITY) == -f32::INFINITY);
+        assert!(decrement_ulp(f32::NAN).is_nan());
+        assert!(decrement_ulp(-f32::NAN).is_nan());
+    }
+
+    #[test]
+    fn inc_dec_ulp() {
+        assert_eq!(decrement_ulp(increment_ulp(0.0)), 0.0);
+        assert_eq!(decrement_ulp(increment_ulp(-0.0)), 0.0);
+        assert_eq!(decrement_ulp(increment_ulp(1.0)), 1.0);
+        assert_eq!(decrement_ulp(increment_ulp(-1.0)), -1.0);
+        assert_eq!(decrement_ulp(increment_ulp(1.2)), 1.2);
+        assert_eq!(decrement_ulp(increment_ulp(-1.2)), -1.2);
+    }
+
+    #[test]
+    fn dec_inc_ulp() {
+        assert_eq!(increment_ulp(decrement_ulp(0.0)), 0.0);
+        assert_eq!(increment_ulp(decrement_ulp(-0.0)), 0.0);
+        assert_eq!(increment_ulp(decrement_ulp(1.0)), 1.0);
+        assert_eq!(increment_ulp(decrement_ulp(-1.0)), -1.0);
+        assert_eq!(increment_ulp(decrement_ulp(1.2)), 1.2);
+        assert_eq!(increment_ulp(decrement_ulp(-1.2)), -1.2);
+    }
+}

sub_crates/rmath/src/vector.rs

@@ -0,0 +1,325 @@
+#![allow(dead_code)]
+
+use std::cmp::PartialEq;
+use std::ops::{Add, Div, Mul, Neg, Sub};
+
+use crate::normal::Normal;
+use crate::point::Point;
+use crate::wide4::Float4;
+use crate::xform::{AsXform, XformFull};
+use crate::{CrossProduct, DotProduct};
+
+/// A direction vector in 3D space.
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Vector(pub Float4);
+
+impl Vector {
+    #[inline(always)]
+    pub fn new(x: f32, y: f32, z: f32) -> Self {
+        Self(Float4::new(x, y, z, 0.0))
+    }
+
+    #[inline(always)]
+    pub fn length(self) -> f32 {
+        self.length2().sqrt()
+    }
+
+    #[inline(always)]
+    pub fn length2(self) -> f32 {
+        let sqr = self.0 * self.0;
+        sqr.a() + sqr.b() + sqr.c()
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn normalized(self) -> Self {
+        Self(self.0 / self.length())
+    }
+
+    #[inline(always)]
+    pub fn abs(self) -> Self {
+        Self(self.0.abs())
+    }
+
+    #[inline(always)]
+    pub fn recip(self) -> Self {
+        Self(self.0.recip())
+    }
+
+    #[inline(always)]
+    pub fn into_point(self) -> Point {
+        Point(self.0)
+    }
+
+    #[inline(always)]
+    pub fn into_normal(self) -> Normal {
+        Normal(self.0)
+    }
+
+    #[inline(always)]
+    pub fn x(self) -> f32 {
+        self.0.a()
+    }
+
+    #[inline(always)]
+    pub fn y(self) -> f32 {
+        self.0.b()
+    }
+
+    #[inline(always)]
+    pub fn z(self) -> f32 {
+        self.0.c()
+    }
+
+    #[inline(always)]
+    pub fn get_n(self, i: usize) -> f32 {
+        match i {
+            0 => self.x(),
+            1 => self.y(),
+            2 => self.z(),
+            _ => panic!("Out of bounds index into 3D vector."),
+        }
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_x(self, x: f32) -> Self {
+        Self(self.0.set_a(x))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_y(self, y: f32) -> Self {
+        Self(self.0.set_b(y))
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_z(self, z: f32) -> Self {
+        Self(self.0.set_c(z))
+    }
+
+    //-------------
+    // Transforms.
+
+    /// Forward-transform the vector.
+    #[inline(always)]
+    pub fn xform<T: AsXform>(self, xform: &T) -> Self {
+        Self(self.0.vec_mul_3x3(&xform.as_xform().m))
+    }
+
+    /// Inverse-transform the vector.
+    #[inline(always)]
+    pub fn xform_inv(self, xform: &XformFull) -> Self {
+        Self(self.0.vec_mul_3x3(&xform.inv_m))
+    }
+
+    /// Faster but less precise version of `xform()`.
+    #[inline(always)]
+    pub fn xform_fast<T: AsXform>(self, xform: &T) -> Self {
+        Self(self.0.vec_mul_3x3_fast(&xform.as_xform().m))
+    }
+
+    /// Faster but less precise version of `xform_inv()`.
+    #[inline(always)]
+    pub fn xform_inv_fast(self, xform: &XformFull) -> Self {
+        Self(self.0.vec_mul_3x3_fast(&xform.inv_m))
+    }
+}
+
+impl Add for Vector {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, other: Self) -> Self {
+        Self(self.0 + other.0)
+    }
+}
+
+impl Sub for Vector {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, other: Self) -> Self {
+        Self(self.0 - other.0)
+    }
+}
+
+impl Mul<f32> for Vector {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, other: f32) -> Self {
+        Self(self.0 * other)
+    }
+}
+
+impl Div<f32> for Vector {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, other: f32) -> Self {
+        Self(self.0 / other)
+    }
+}
+
+impl Neg for Vector {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self(-self.0)
+    }
+}
+
+impl PartialEq for Vector {
+    #[inline(always)]
+    fn eq(&self, rhs: &Self) -> bool {
+        self.0.a() == rhs.0.a() && self.0.b() == rhs.0.b() && self.0.c() == rhs.0.c()
+    }
+}
+
+impl DotProduct for Vector {
+    #[inline(always)]
+    fn dot(self, other: Self) -> f32 {
+        Float4::dot_3(self.0, other.0)
+    }
+
+    #[inline(always)]
+    fn dot_fast(self, other: Self) -> f32 {
+        Float4::dot_3_fast(self.0, other.0)
+    }
+}
+
+impl CrossProduct for Vector {
+    #[inline(always)]
+    fn cross(self, other: Self) -> Self {
+        Self(Float4::cross_3(self.0, other.0))
+    }
+
+    #[inline(always)]
+    fn cross_fast(self, other: Self) -> Self {
+        Self(Float4::cross_3_fast(self.0, other.0))
+    }
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{CrossProduct, DotProduct, Xform};
+
+    #[test]
+    fn add() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = Vector::new(1.5, 4.5, 2.5);
+        let v3 = Vector::new(2.5, 6.5, 5.5);
+
+        assert_eq!(v3, v1 + v2);
+    }
+
+    #[test]
+    fn sub() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = Vector::new(1.5, 4.5, 2.5);
+        let v3 = Vector::new(-0.5, -2.5, 0.5);
+
+        assert_eq!(v3, v1 - v2);
+    }
+
+    #[test]
+    fn mul_scalar() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = 2.0;
+        let v3 = Vector::new(2.0, 4.0, 6.0);
+
+        assert_eq!(v3, v1 * v2);
+    }
+
+    #[test]
+    fn xform() {
+        let v = Vector::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+
+        assert_eq!(v.xform(&m), Vector::new(14.0, 46.0, 58.0));
+        assert_eq!(v.xform(&m).xform_inv(&m), v);
+    }
+
+    #[test]
+    fn xform_fast() {
+        let v = Vector::new(1.0, 2.5, 4.0);
+        let m = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0)
+            .to_full()
+            .unwrap();
+
+        assert_eq!(v.xform_fast(&m), Vector::new(14.0, 46.0, 58.0));
+        assert_eq!(v.xform_fast(&m).xform_inv_fast(&m), v);
+    }
+
+    #[test]
+    fn div() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = 2.0;
+        let v3 = Vector::new(0.5, 1.0, 1.5);
+
+        assert_eq!(v3, v1 / v2);
+    }
+
+    #[test]
+    fn length() {
+        let v = Vector::new(1.0, 2.0, 3.0);
+        assert!((v.length() - 3.7416573867739413).abs() < 0.000001);
+    }
+
+    #[test]
+    fn length2() {
+        let v = Vector::new(1.0, 2.0, 3.0);
+        assert_eq!(v.length2(), 14.0);
+    }
+
+    #[test]
+    fn normalized() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = Vector::new(0.2672612419124244, 0.5345224838248488, 0.8017837257372732);
+        let v3 = v1.normalized();
+        assert!((v3.x() - v2.x()).abs() < 0.000001);
+        assert!((v3.y() - v2.y()).abs() < 0.000001);
+        assert!((v3.z() - v2.z()).abs() < 0.000001);
+    }
+
+    #[test]
+    fn dot() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = Vector::new(1.5, 4.5, 2.5);
+
+        assert_eq!(v1.dot(v2), 18.0);
+    }
+
+    #[test]
+    fn dot_fast() {
+        let v1 = Vector::new(1.0, 2.0, 3.0);
+        let v2 = Vector::new(1.5, 4.5, 2.5);
+
+        assert_eq!(v1.dot_fast(v2), 18.0);
+    }
+
+    #[test]
+    fn cross() {
+        let v1 = Vector::new(1.0, 0.0, 0.0);
+        let v2 = Vector::new(0.0, 1.0, 0.0);
+
+        assert_eq!(v1.cross(v2), Vector::new(0.0, 0.0, 1.0));
+    }
+
+    #[test]
+    fn cross_fast() {
+        let v1 = Vector::new(1.0, 0.0, 0.0);
+        let v2 = Vector::new(0.0, 1.0, 0.0);
+
+        assert_eq!(v1.cross_fast(v2), Vector::new(0.0, 0.0, 1.0));
+    }
+}

sub_crates/rmath/src/wide4/fallback.rs

@@ -0,0 +1,462 @@
+use std::ops::{Add, BitAnd, BitOr, BitXor, Div, Index, Mul, Neg, Not, Sub};
+
+use crate::FMulAdd;
+
+//=============================================================
+// Float4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(C, align(16))]
+pub struct Float4([f32; 4]);
+
+impl Float4 {
+    /// Create a new `Float4` with the given components.
+    #[inline(always)]
+    pub fn new(a: f32, b: f32, c: f32, d: f32) -> Self {
+        Self([a, b, c, d])
+    }
+
+    /// Create a new `Float4` with all elements set to `n`.
+    #[inline(always)]
+    pub fn splat(n: f32) -> Self {
+        Self([n, n, n, n])
+    }
+
+    /// Component-wise fused multiply-add.
+    ///
+    /// `(self * a) + b` with only one rounding error.
+    #[inline(always)]
+    pub fn mul_add(self, a: Self, b: Self) -> Self {
+        Self([
+            self.0[0].mul_add(a.0[0], b.0[0]),
+            self.0[1].mul_add(a.0[1], b.0[1]),
+            self.0[2].mul_add(a.0[2], b.0[2]),
+            self.0[3].mul_add(a.0[3], b.0[3]),
+        ])
+    }
+
+    /// Vertical minimum.
+    #[inline(always)]
+    pub fn min(self, a: Self) -> Self {
+        // Custom min to match behavior of SSE.
+        #[inline(always)]
+        pub fn minf(a: f32, b: f32) -> f32 {
+            if a < b {
+                a
+            } else {
+                b
+            }
+        }
+        Self([
+            minf(self.0[0], a.0[0]),
+            minf(self.0[1], a.0[1]),
+            minf(self.0[2], a.0[2]),
+            minf(self.0[3], a.0[3]),
+        ])
+    }
+
+    /// Vertical maximum.
+    #[inline(always)]
+    pub fn max(self, a: Self) -> Self {
+        // Custom max to match behavior of SSE.
+        #[inline(always)]
+        pub fn maxf(a: f32, b: f32) -> f32 {
+            if a > b {
+                a
+            } else {
+                b
+            }
+        }
+        Self([
+            maxf(self.0[0], a.0[0]),
+            maxf(self.0[1], a.0[1]),
+            maxf(self.0[2], a.0[2]),
+            maxf(self.0[3], a.0[3]),
+        ])
+    }
+
+    /// Horizontal minimum.
+    #[inline(always)]
+    pub fn min_element(self) -> f32 {
+        let a = self.0[0].min(self.0[1]);
+        let b = self.0[2].min(self.0[3]);
+        a.min(b)
+    }
+
+    /// Horizontal maximum.
+    #[inline(always)]
+    pub fn max_element(self) -> f32 {
+        let a = self.0[0].max(self.0[1]);
+        let b = self.0[2].max(self.0[3]);
+        a.max(b)
+    }
+
+    /// 1.0 / self
+    #[inline(always)]
+    pub fn recip(self) -> Self {
+        Float4::splat(1.0) / self
+    }
+
+    #[inline(always)]
+    pub fn abs(self) -> Self {
+        Float4::new(
+            self.a().abs(),
+            self.b().abs(),
+            self.c().abs(),
+            self.d().abs(),
+        )
+    }
+
+    //-----------------------------------------------------
+    // Comparisons.
+
+    /// Less than.
+    #[inline(always)]
+    pub fn cmplt(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] < rhs.0[0],
+            self.0[1] < rhs.0[1],
+            self.0[2] < rhs.0[2],
+            self.0[3] < rhs.0[3],
+        ])
+    }
+
+    /// Less than or equal.
+    #[inline(always)]
+    pub fn cmplte(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] <= rhs.0[0],
+            self.0[1] <= rhs.0[1],
+            self.0[2] <= rhs.0[2],
+            self.0[3] <= rhs.0[3],
+        ])
+    }
+
+    /// Greater than.
+    #[inline(always)]
+    pub fn cmpgt(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] > rhs.0[0],
+            self.0[1] > rhs.0[1],
+            self.0[2] > rhs.0[2],
+            self.0[3] > rhs.0[3],
+        ])
+    }
+
+    /// Greater than or equal.
+    #[inline(always)]
+    pub fn cmpgte(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] >= rhs.0[0],
+            self.0[1] >= rhs.0[1],
+            self.0[2] >= rhs.0[2],
+            self.0[3] >= rhs.0[3],
+        ])
+    }
+
+    /// Equal.
+    #[inline(always)]
+    pub fn cmpeq(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] == rhs.0[0],
+            self.0[1] == rhs.0[1],
+            self.0[2] == rhs.0[2],
+            self.0[3] == rhs.0[3],
+        ])
+    }
+
+    //-----------------------------------------------------
+    // Individual components.
+
+    #[inline(always)]
+    pub fn a(self) -> f32 {
+        self.0[0]
+    }
+
+    #[inline(always)]
+    pub fn b(self) -> f32 {
+        self.0[1]
+    }
+
+    #[inline(always)]
+    pub fn c(self) -> f32 {
+        self.0[2]
+    }
+
+    #[inline(always)]
+    pub fn d(self) -> f32 {
+        self.0[3]
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_a(self, n: f32) -> Self {
+        Self([n, self.0[1], self.0[2], self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_b(self, n: f32) -> Self {
+        Self([self.0[0], n, self.0[2], self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_c(self, n: f32) -> Self {
+        Self([self.0[0], self.0[1], n, self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_d(self, n: f32) -> Self {
+        Self([self.0[0], self.0[1], self.0[2], n])
+    }
+
+    //-----------------------------------------------------
+    // Shuffles.
+
+    #[inline(always)]
+    pub fn aaaa(self) -> Self {
+        let a = self.0[0];
+        Self([a, a, a, a])
+    }
+
+    #[inline(always)]
+    pub fn bbbb(self) -> Self {
+        let b = self.0[1];
+        Self([b, b, b, b])
+    }
+
+    #[inline(always)]
+    pub fn cccc(self) -> Self {
+        let c = self.0[2];
+        Self([c, c, c, c])
+    }
+
+    #[inline(always)]
+    pub fn dddd(self) -> Self {
+        let d = self.0[3];
+        Self([d, d, d, d])
+    }
+
+    #[inline(always)]
+    pub fn bcad(self) -> Self {
+        let a = self.0[0];
+        let b = self.0[1];
+        let c = self.0[2];
+        let d = self.0[3];
+        Self([b, c, a, d])
+    }
+
+    #[inline(always)]
+    pub fn cabd(self) -> Self {
+        let a = self.0[0];
+        let b = self.0[1];
+        let c = self.0[2];
+        let d = self.0[3];
+        Self([c, a, b, d])
+    }
+}
+
+impl Index<usize> for Float4 {
+    type Output = f32;
+
+    #[inline(always)]
+    fn index(&self, idx: usize) -> &f32 {
+        &self.0[idx]
+    }
+}
+
+impl Add for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] + rhs.0[0],
+            self.0[1] + rhs.0[1],
+            self.0[2] + rhs.0[2],
+            self.0[3] + rhs.0[3],
+        ])
+    }
+}
+
+impl Sub for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] - rhs.0[0],
+            self.0[1] - rhs.0[1],
+            self.0[2] - rhs.0[2],
+            self.0[3] - rhs.0[3],
+        ])
+    }
+}
+
+impl Mul for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] * rhs.0[0],
+            self.0[1] * rhs.0[1],
+            self.0[2] * rhs.0[2],
+            self.0[3] * rhs.0[3],
+        ])
+    }
+}
+
+impl Mul<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: f32) -> Self {
+        Self([
+            self.0[0] * rhs,
+            self.0[1] * rhs,
+            self.0[2] * rhs,
+            self.0[3] * rhs,
+        ])
+    }
+}
+
+impl Div for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] / rhs.0[0],
+            self.0[1] / rhs.0[1],
+            self.0[2] / rhs.0[2],
+            self.0[3] / rhs.0[3],
+        ])
+    }
+}
+
+impl Div<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: f32) -> Self {
+        Self([
+            self.0[0] / rhs,
+            self.0[1] / rhs,
+            self.0[2] / rhs,
+            self.0[3] / rhs,
+        ])
+    }
+}
+
+impl Neg for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self([-self.0[0], -self.0[1], -self.0[2], -self.0[3]])
+    }
+}
+
+impl FMulAdd for Float4 {
+    #[inline(always)]
+    fn fma(self, b: Self, c: Self) -> Self {
+        self.mul_add(b, c)
+    }
+}
+
+//=============================================================
+// Bool4
+
+#[derive(Copy, Clone)]
+#[repr(transparent)]
+pub struct Bool4([bool; 4]);
+
+impl Bool4 {
+    #[inline(always)]
+    pub fn new(a: bool, b: bool, c: bool, d: bool) -> Self {
+        Bool4([a, b, c, d])
+    }
+
+    #[inline(always)]
+    pub fn new_false() -> Self {
+        Self([false, false, false, false])
+    }
+
+    #[inline(always)]
+    pub fn to_bools(self) -> [bool; 4] {
+        self.0
+    }
+
+    /// Note: `a` goes to the least significant bit.
+    #[inline(always)]
+    pub fn bitmask(self) -> u8 {
+        self.0[0] as u8
+            | ((self.0[1] as u8) << 1)
+            | ((self.0[2] as u8) << 2)
+            | ((self.0[3] as u8) << 3)
+    }
+
+    #[inline(always)]
+    pub fn any(self) -> bool {
+        self.0[0] | &self.0[1] | self.0[2] | self.0[3]
+    }
+
+    #[inline(always)]
+    pub fn all(self) -> bool {
+        self.0[0] & &self.0[1] & self.0[2] & self.0[3]
+    }
+}
+
+impl BitAnd for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitand(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] & rhs.0[0],
+            self.0[1] & rhs.0[1],
+            self.0[2] & rhs.0[2],
+            self.0[3] & rhs.0[3],
+        ])
+    }
+}
+
+impl BitOr for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitor(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] | rhs.0[0],
+            self.0[1] | rhs.0[1],
+            self.0[2] | rhs.0[2],
+            self.0[3] | rhs.0[3],
+        ])
+    }
+}
+
+impl BitXor for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitxor(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] ^ rhs.0[0],
+            self.0[1] ^ rhs.0[1],
+            self.0[2] ^ rhs.0[2],
+            self.0[3] ^ rhs.0[3],
+        ])
+    }
+}
+
+impl Not for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn not(self) -> Self {
+        Self([!self.0[0], !self.0[1], !self.0[2], !self.0[3]])
+    }
+}

sub_crates/rmath/src/wide4/mod.rs

@@ -0,0 +1,812 @@
+use std::{
+    cmp::{Eq, PartialEq},
+    ops::{AddAssign, BitAndAssign, BitOrAssign, BitXorAssign, DivAssign, MulAssign, SubAssign},
+};
+
+use crate::utils::ulps_eq;
+use crate::{difference_of_products, two_diff, two_prod, two_sum};
+
+//-------------------------------------------------------------
+// Which implementation to use.
+
+mod fallback;
+#[cfg(not(any(target_arch = "x86_64")))]
+pub use fallback::{Bool4, Float4};
+
+#[cfg(target_arch = "x86_64")]
+mod sse;
+#[cfg(target_arch = "x86_64")]
+pub use sse::{Bool4, Float4};
+
+//-------------------------------------------------------------
+
+impl Float4 {
+    /// 3D dot product (only uses the first 3 components).
+    #[inline(always)]
+    pub fn dot_3(a: Self, b: Self) -> f32 {
+        let (p, p_err) = two_prod(a, b);
+
+        // Products.
+        let (x, x_err) = (p.a(), p_err.a());
+        let (y, y_err) = (p.b(), p_err.b());
+        let (z, z_err) = (p.c(), p_err.c());
+
+        // Sums.
+        let (s1, s1_err) = two_sum(x, y);
+        let err1 = x_err + (y_err + s1_err);
+
+        let (s2, s2_err) = two_sum(s1, z);
+        let err2 = z_err + (err1 + s2_err);
+
+        // Final result with rounding error compensation.
+        s2 + err2
+    }
+
+    /// Faster but less precise version of `dot_3()`.
+    #[inline(always)]
+    pub fn dot_3_fast(a: Self, b: Self) -> f32 {
+        let c = a * b;
+        c.a() + c.b() + c.c()
+    }
+
+    /// 3D cross product (only uses the first 3 components).
+    #[inline(always)]
+    pub fn cross_3(a: Self, b: Self) -> Self {
+        difference_of_products(a.bcad(), b.cabd(), a.cabd(), b.bcad())
+    }
+
+    /// Faster but less precise version `cross_3()`.
+    #[inline(always)]
+    pub fn cross_3_fast(a: Self, b: Self) -> Self {
+        (a.bcad() * b.cabd()) - (a.cabd() * b.bcad())
+    }
+
+    #[inline(always)]
+    pub fn transpose_3x3(m: &[Self; 3]) -> [Self; 3] {
+        [
+            // The fourth component in each row below is arbitrary,
+            // but in this case chosen so that it matches the
+            // behavior of the SSE version of transpose_3x3.
+            Self::new(m[0].a(), m[1].a(), m[2].a(), m[2].d()),
+            Self::new(m[0].b(), m[1].b(), m[2].b(), m[2].d()),
+            Self::new(m[0].c(), m[1].c(), m[2].c(), m[2].d()),
+        ]
+    }
+
+    /// Invert a 3x3 matrix.
+    ///
+    /// Returns `None` if not invertible.
+    #[inline]
+    pub fn invert_3x3(m: &[Self; 3]) -> Option<[Self; 3]> {
+        let m0_bca = m[0].bcad();
+        let m1_bca = m[1].bcad();
+        let m2_bca = m[2].bcad();
+        let m0_cab = m[0].cabd();
+        let m1_cab = m[1].cabd();
+        let m2_cab = m[2].cabd();
+        let abc = difference_of_products(m1_bca, m2_cab, m1_cab, m2_bca);
+        let def = difference_of_products(m2_bca, m0_cab, m2_cab, m0_bca);
+        let ghi = difference_of_products(m0_bca, m1_cab, m0_cab, m1_bca);
+
+        let det = Self::dot_3(
+            Self::new(abc.a(), def.a(), ghi.a(), 0.0),
+            Self::new(m[0].a(), m[1].a(), m[2].a(), 0.0),
+        );
+
+        if det == 0.0 {
+            None
+        } else {
+            Some(Self::transpose_3x3(&[abc / det, def / det, ghi / det]))
+        }
+    }
+
+    /// Invert a 3x3 matrix, and also return the computed determinant.
+    ///
+    /// Returns `None` if not invertible.
+    #[inline]
+    pub fn invert_3x3_w_det(m: &[Self; 3]) -> Option<([Self; 3], f32)> {
+        let m0_bca = m[0].bcad();
+        let m1_bca = m[1].bcad();
+        let m2_bca = m[2].bcad();
+        let m0_cab = m[0].cabd();
+        let m1_cab = m[1].cabd();
+        let m2_cab = m[2].cabd();
+        let abc = difference_of_products(m1_bca, m2_cab, m1_cab, m2_bca);
+        let def = difference_of_products(m2_bca, m0_cab, m2_cab, m0_bca);
+        let ghi = difference_of_products(m0_bca, m1_cab, m0_cab, m1_bca);
+
+        let det = Self::dot_3(
+            Self::new(abc.a(), def.a(), ghi.a(), 0.0),
+            Self::new(m[0].a(), m[1].a(), m[2].a(), 0.0),
+        );
+
+        if det == 0.0 {
+            None
+        } else {
+            Some((Self::transpose_3x3(&[abc / det, def / det, ghi / det]), det))
+        }
+    }
+
+    /// Faster but less precise version of `invert_3x3()`.
+    #[inline]
+    pub fn invert_3x3_fast(m: &[Self; 3]) -> Option<[Self; 3]> {
+        let m0_bca = m[0].bcad();
+        let m1_bca = m[1].bcad();
+        let m2_bca = m[2].bcad();
+        let m0_cab = m[0].cabd();
+        let m1_cab = m[1].cabd();
+        let m2_cab = m[2].cabd();
+        let abc = (m1_bca * m2_cab) - (m1_cab * m2_bca);
+        let def = (m2_bca * m0_cab) - (m2_cab * m0_bca);
+        let ghi = (m0_bca * m1_cab) - (m0_cab * m1_bca);
+
+        let det = Self::dot_3_fast(
+            Self::new(abc.a(), def.a(), ghi.a(), 0.0),
+            Self::new(m[0].a(), m[1].a(), m[2].a(), 0.0),
+        );
+
+        if det == 0.0 {
+            None
+        } else {
+            Some(Self::transpose_3x3(&[abc / det, def / det, ghi / det]))
+        }
+    }
+
+    /// Multiplies a 3D vector with a 3x3 matrix.
+    #[inline]
+    pub fn vec_mul_3x3(self, m: &[Self; 3]) -> Self {
+        let x = self.aaaa();
+        let y = self.bbbb();
+        let z = self.cccc();
+
+        // Products.
+        let (a, a_err) = two_prod(x, m[0]);
+        let (b, b_err) = two_prod(y, m[1]);
+        let (c, c_err) = two_prod(z, m[2]);
+
+        // Sums.
+        let (s1, s1_err) = two_sum(a, b);
+        let err1 = a_err + (b_err + s1_err);
+
+        let (s2, s2_err) = two_sum(c, s1);
+        let err2 = c_err + (err1 + s2_err);
+
+        s2 + err2
+    }
+
+    /// Faster but less precise version of `vec_mul_3x3()`.
+    #[inline]
+    pub fn vec_mul_3x3_fast(self, m: &[Self; 3]) -> Self {
+        let x = self.aaaa();
+        let y = self.bbbb();
+        let z = self.cccc();
+
+        (x * m[0]) + (y * m[1]) + (z * m[2])
+    }
+
+    /// Transforms a 3d point by an affine transform.
+    ///
+    /// `m` is the 3x3 part of the affine transform, `t` is the translation part.
+    #[inline]
+    pub fn vec_mul_affine(self, m: &[Self; 3], t: Self) -> Self {
+        let x = self.aaaa();
+        let y = self.bbbb();
+        let z = self.cccc();
+
+        // Products.
+        let (a, a_err) = two_prod(x, m[0]);
+        let (b, b_err) = two_prod(y, m[1]);
+        let (c, c_err) = two_prod(z, m[2]);
+
+        // Sums.
+        let (s1, s1_err) = two_sum(a, b);
+        let err1 = a_err + (b_err + s1_err);
+
+        let (s2, s2_err) = two_sum(c, s1);
+        let err2 = c_err + (err1 + s2_err);
+
+        let (s3, s3_err) = two_sum(t, s2);
+        let err3 = err2 + s3_err;
+
+        s3 + err3
+    }
+
+    /// Faster but less precise version of `vec_mul_affine()`.
+    #[inline]
+    pub fn vec_mul_affine_fast(self, m: &[Self; 3], t: Self) -> Self {
+        let x = self.aaaa();
+        let y = self.bbbb();
+        let z = self.cccc();
+
+        (x * m[0]) + (y * m[1]) + (z * m[2]) + t
+    }
+
+    /// Transforms a 3d point by an affine transform, except it does
+    /// `(vec - t) * inv_m` instead of `vec * m + t`.
+    ///
+    /// This is useful for performing efficient inverse transforms while
+    /// only having to invert the 3x3 part of the transform itself.
+    ///
+    /// `inv_m` is the inverse 3x3 part of the affine transform, `t` is
+    /// the forward translation part.
+    #[inline]
+    pub fn vec_mul_affine_rev(self, inv_m: &[Self; 3], t: Self) -> Self {
+        let (v, v_err) = two_diff(self, t);
+
+        let (x, x_err) = (v.aaaa(), v_err.aaaa());
+        let (y, y_err) = (v.bbbb(), v_err.bbbb());
+        let (z, z_err) = (v.cccc(), v_err.cccc());
+
+        // Products.
+        let ((a, a_err1), a_err2) = (two_prod(x, inv_m[0]), x_err * inv_m[0]);
+        let ((b, b_err1), b_err2) = (two_prod(y, inv_m[1]), y_err * inv_m[1]);
+        let ((c, c_err1), c_err2) = (two_prod(z, inv_m[2]), z_err * inv_m[2]);
+        let a_err = a_err1 + a_err2;
+        let b_err = b_err1 + b_err2;
+        let c_err = c_err1 + c_err2;
+
+        // Sums.
+        let (s1, s1_err) = two_sum(a, b);
+        let err1 = a_err + (b_err + s1_err);
+
+        let (s2, s2_err) = two_sum(c, s1);
+        let err2 = c_err + (err1 + s2_err);
+
+        s2 + err2
+    }
+
+    /// Faster but less precise version of `vec_mul_affine_rev()`.
+    #[inline]
+    pub fn vec_mul_affine_rev_fast(self, inv_m: &[Self; 3], t: Self) -> Self {
+        let v = self - t;
+
+        let x = v.aaaa();
+        let y = v.bbbb();
+        let z = v.cccc();
+
+        (x * inv_m[0]) + (y * inv_m[1]) + (z * inv_m[2])
+    }
+
+    /// Returns whether the `Float4`s are approximately equal to each
+    /// other.
+    ///
+    /// Each corresponding element cannot have a relative error exceeding
+    /// `epsilon`.
+    pub(crate) fn aprx_eq(a: Self, b: Self, max_ulps: u32) -> bool {
+        let mut eq = true;
+        eq &= ulps_eq(a.a(), b.a(), max_ulps);
+        eq &= ulps_eq(a.b(), b.b(), max_ulps);
+        eq &= ulps_eq(a.c(), b.c(), max_ulps);
+        eq &= ulps_eq(a.d(), b.d(), max_ulps);
+        eq
+    }
+
+    /// Transforms one affine transform by another.
+    ///
+    /// The result is an affine transform that acts as a sequence of the
+    /// first followed by the second.
+    ///
+    /// `m#` is the 3x3 part of the affine transform, `t#` is the translation part.
+    #[inline]
+    pub fn affine_mul_affine(
+        m1: &[Self; 3],
+        t1: Self,
+        m2: &[Self; 3],
+        t2: Self,
+    ) -> ([Self; 3], Self) {
+        (
+            [
+                m1[0].vec_mul_3x3(&m2),
+                m1[1].vec_mul_3x3(&m2),
+                m1[2].vec_mul_3x3(&m2),
+            ],
+            t1.vec_mul_affine(&m2, t2),
+        )
+    }
+
+    /// Faster but less precise version of `affine_mul_affine()`.
+    #[inline]
+    pub fn affine_mul_affine_fast(
+        m1: &[Self; 3],
+        t1: Self,
+        m2: &[Self; 3],
+        t2: Self,
+    ) -> ([Self; 3], Self) {
+        (
+            [
+                m1[0].vec_mul_3x3_fast(&m2),
+                m1[1].vec_mul_3x3_fast(&m2),
+                m1[2].vec_mul_3x3_fast(&m2),
+            ],
+            t1.vec_mul_affine_fast(&m2, t2),
+        )
+    }
+}
+
+impl From<Float4> for (f32, f32, f32, f32) {
+    fn from(v: Float4) -> (f32, f32, f32, f32) {
+        (v.a(), v.b(), v.c(), v.d())
+    }
+}
+
+impl AddAssign for Float4 {
+    #[inline(always)]
+    fn add_assign(&mut self, rhs: Self) {
+        *self = *self + rhs;
+    }
+}
+
+impl SubAssign for Float4 {
+    #[inline(always)]
+    fn sub_assign(&mut self, rhs: Self) {
+        *self = *self - rhs;
+    }
+}
+
+impl MulAssign for Float4 {
+    #[inline(always)]
+    fn mul_assign(&mut self, rhs: Self) {
+        *self = *self * rhs;
+    }
+}
+
+impl MulAssign<f32> for Float4 {
+    #[inline(always)]
+    fn mul_assign(&mut self, rhs: f32) {
+        *self = *self * rhs;
+    }
+}
+
+impl DivAssign for Float4 {
+    #[inline(always)]
+    fn div_assign(&mut self, rhs: Self) {
+        *self = *self / rhs;
+    }
+}
+
+impl DivAssign<f32> for Float4 {
+    #[inline(always)]
+    fn div_assign(&mut self, rhs: f32) {
+        *self = *self / rhs;
+    }
+}
+
+impl PartialEq for Float4 {
+    #[inline(always)]
+    fn eq(&self, rhs: &Self) -> bool {
+        self.cmpeq(*rhs).bitmask() == 0b1111
+    }
+}
+
+//--------
+
+impl BitAndAssign for Bool4 {
+    #[inline(always)]
+    fn bitand_assign(&mut self, rhs: Self) {
+        *self = *self & rhs;
+    }
+}
+
+impl BitOrAssign for Bool4 {
+    #[inline(always)]
+    fn bitor_assign(&mut self, rhs: Self) {
+        *self = *self | rhs;
+    }
+}
+
+impl BitXorAssign for Bool4 {
+    #[inline(always)]
+    fn bitxor_assign(&mut self, rhs: Self) {
+        *self = *self ^ rhs;
+    }
+}
+
+impl PartialEq for Bool4 {
+    #[inline(always)]
+    fn eq(&self, rhs: &Self) -> bool {
+        self.bitmask() == rhs.bitmask()
+    }
+}
+
+impl Eq for Bool4 {}
+
+impl std::fmt::Debug for Bool4 {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
+        f.write_str("Bool4(")?;
+        f.debug_list().entries(self.to_bools().iter()).finish()?;
+        f.write_str(")")?;
+
+        Ok(())
+    }
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    //------------
+    // Float4
+
+    #[test]
+    fn approximate_equality_test() {
+        let a = Float4::new(1.0, 2.0, 3.0, 4.0);
+        let b = Float4::new(1.00001, 2.00002, 3.00003, 4.00004);
+        let c = Float4::new(1.0e-43, 2.0e-43, 3.0e-43, 4.0e-43);
+        let d = Float4::new(-1.0e-43, -2.0e-43, -3.0e-43, -4.0e-43);
+
+        assert!(Float4::aprx_eq(a, a, 0));
+
+        assert!(Float4::aprx_eq(a, b, 130));
+        assert!(!Float4::aprx_eq(a, b, 120));
+
+        assert!(Float4::aprx_eq(c, d, 575));
+        assert!(!Float4::aprx_eq(c, d, 565));
+    }
+
+    #[test]
+    fn index() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+        assert_eq!(v[0], 0.0);
+        assert_eq!(v[1], 1.0);
+        assert_eq!(v[2], 2.0);
+        assert_eq!(v[3], 3.0);
+    }
+
+    #[test]
+    fn get() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+
+        assert_eq!(v.a(), 0.0);
+        assert_eq!(v.b(), 1.0);
+        assert_eq!(v.c(), 2.0);
+        assert_eq!(v.d(), 3.0);
+    }
+
+    #[test]
+    fn set() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+
+        assert_eq!(v.set_a(9.0), Float4::new(9.0, 1.0, 2.0, 3.0));
+        assert_eq!(v.set_b(9.0), Float4::new(0.0, 9.0, 2.0, 3.0));
+        assert_eq!(v.set_c(9.0), Float4::new(0.0, 1.0, 9.0, 3.0));
+        assert_eq!(v.set_d(9.0), Float4::new(0.0, 1.0, 2.0, 9.0));
+    }
+
+    #[test]
+    fn shuffle() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+
+        assert_eq!(v.aaaa(), Float4::splat(0.0));
+        assert_eq!(v.bbbb(), Float4::splat(1.0));
+        assert_eq!(v.cccc(), Float4::splat(2.0));
+        assert_eq!(v.dddd(), Float4::splat(3.0));
+
+        assert_eq!(v.bcad(), Float4::new(1.0, 2.0, 0.0, 3.0));
+        assert_eq!(v.cabd(), Float4::new(2.0, 0.0, 1.0, 3.0));
+    }
+
+    #[test]
+    fn abs() {
+        let v1 = Float4::new(-1.0, 2.0, -3.0, 4.0);
+        let v2 = Float4::new(1.0, -2.0, 3.0, -4.0);
+
+        let r = Float4::new(1.0, 2.0, 3.0, 4.0);
+
+        assert_eq!(v1.abs(), r);
+        assert_eq!(v2.abs(), r);
+    }
+
+    #[test]
+    fn neg() {
+        let v1 = Float4::new(-1.0, 2.0, -3.0, 4.0);
+        let v2 = Float4::new(1.0, -2.0, 3.0, -4.0);
+
+        assert_eq!(-v1, v2);
+        assert_eq!(-v2, v1);
+    }
+
+    #[test]
+    fn cmp_ops() {
+        let a = Float4::new(1.0, 2.0, -2.0, 0.0);
+        let b = Float4::new(1.0, -2.0, 2.0, -0.0);
+
+        assert_eq!(a.cmplt(b), Bool4::new(false, false, true, false));
+        assert_eq!(a.cmplte(b), Bool4::new(true, false, true, true));
+        assert_eq!(a.cmpgt(b), Bool4::new(false, true, false, false));
+        assert_eq!(a.cmpgte(b), Bool4::new(true, true, false, true));
+        assert_eq!(a.cmpeq(b), Bool4::new(true, false, false, true));
+    }
+
+    #[test]
+    fn min_max() {
+        let a = Float4::new(1.0, 2.0, -2.0, 4.0);
+        let b = Float4::new(1.0, -2.0, 2.0, 5.0);
+
+        assert_eq!(a.min(b), Float4::new(1.0, -2.0, -2.0, 4.0));
+        assert_eq!(a.max(b), Float4::new(1.0, 2.0, 2.0, 5.0));
+
+        let c = Float4::new(std::f32::INFINITY, 2.0, std::f32::NAN, 4.0);
+        let d = Float4::new(1.0, -std::f32::INFINITY, 2.0, std::f32::NAN);
+        let r_min = c.min(d);
+        let r_max = c.max(d);
+
+        assert_eq!(r_min.a(), 1.0);
+        assert_eq!(r_min.b(), -std::f32::INFINITY);
+        assert_eq!(r_min.c(), 2.0);
+        assert!(r_min.d().is_nan());
+
+        assert_eq!(r_max.a(), std::f32::INFINITY);
+        assert_eq!(r_max.b(), 2.0);
+        assert_eq!(r_max.c(), 2.0);
+        assert!(r_max.d().is_nan());
+    }
+
+    #[test]
+    fn dot_3() {
+        let v1 = Float4::new(1.0, 2.0, -3.0, 0.0);
+        let v2 = Float4::new(4.0, -5.0, 6.0, 0.0);
+
+        assert_eq!(Float4::dot_3(v1, v2), -24.0);
+        assert_eq!(Float4::dot_3_fast(v1, v2), -24.0);
+    }
+
+    #[test]
+    fn cross_3() {
+        let v1 = Float4::new(1.0, 2.0, -3.0, 0.0);
+        let v2 = Float4::new(4.0, -5.0, 6.0, 0.0);
+
+        let r = Float4::new(-3.0, -18.0, -13.0, 0.0);
+
+        assert_eq!(Float4::cross_3(v1, v2), r);
+        assert_eq!(Float4::cross_3(v2, v1), -r);
+        assert_eq!(Float4::cross_3_fast(v1, v2), r);
+        assert_eq!(Float4::cross_3_fast(v2, v1), -r);
+    }
+
+    #[test]
+    fn transpose_3x3() {
+        let m1 = [
+            Float4::new(1.0, 4.0, 7.0, 0.0),
+            Float4::new(2.0, 5.0, 8.0, 0.0),
+            Float4::new(3.0, 6.0, 9.0, 0.0),
+        ];
+        let m2 = [
+            Float4::new(1.0, 2.0, 3.0, 0.0),
+            Float4::new(4.0, 5.0, 6.0, 0.0),
+            Float4::new(7.0, 8.0, 9.0, 0.0),
+        ];
+
+        assert_eq!(Float4::transpose_3x3(&m1), m2);
+        assert_eq!(Float4::transpose_3x3(&m2), m1);
+    }
+
+    #[test]
+    fn invert_3x3() {
+        let m = [
+            Float4::new(1.0, 3.0, 9.0, 0.0),
+            Float4::new(2.0, 6.0, 2.0, 0.0),
+            Float4::new(2.0, 7.0, 11.0, 0.0),
+        ];
+        let inv_m = [
+            Float4::new(3.25, 1.875, -3.0, 0.0),
+            Float4::new(-1.125, -0.4375, 1.0, 0.0),
+            Float4::new(0.125, -0.0625, 0.0, 0.0),
+        ];
+
+        assert_eq!(Float4::invert_3x3(&m).unwrap(), inv_m);
+        assert_eq!(Float4::invert_3x3(&inv_m).unwrap(), m);
+        assert_eq!(Float4::invert_3x3_fast(&m).unwrap(), inv_m);
+        assert_eq!(Float4::invert_3x3_fast(&inv_m).unwrap(), m);
+    }
+
+    #[test]
+    fn vec_mul_3x3() {
+        let v = Float4::new(1.0, 2.5, 4.0, 0.0);
+        let m = [
+            Float4::new(1.0, 3.0, 9.0, 0.0),
+            Float4::new(2.0, 6.0, 2.0, 0.0),
+            Float4::new(2.0, 7.0, 11.0, 0.0),
+        ];
+
+        let r = Float4::new(14.0, 46.0, 58.0, 0.0);
+
+        assert_eq!(v.vec_mul_3x3(&m), r);
+        assert_eq!(v.vec_mul_3x3_fast(&m), r);
+    }
+
+    #[test]
+    fn vec_mul_affine() {
+        let p = Float4::new(1.0, 2.5, 4.0, 0.0);
+        let xform = (
+            [
+                Float4::new(1.0, 3.0, 9.0, 0.0),
+                Float4::new(2.0, 6.0, 2.0, 0.0),
+                Float4::new(2.0, 7.0, 11.0, 0.0),
+            ],
+            Float4::new(1.5, 8.0, 12.0, 0.0),
+        );
+
+        let r = Float4::new(15.5, 54.0, 70.0, 0.0);
+
+        assert_eq!(p.vec_mul_affine(&xform.0, xform.1), r);
+    }
+
+    #[test]
+    fn vec_mul_affine_rev() {
+        let p = Float4::new(15.5, 54.0, 70.0, 0.0);
+        let inv_m = [
+            Float4::new(3.25, 1.875, -3.0, 0.0),
+            Float4::new(-1.125, -0.4375, 1.0, 0.0),
+            Float4::new(0.125, -0.0625, 0.0, 0.0),
+        ];
+        let t = Float4::new(1.5, 8.0, 12.0, 0.0);
+
+        let r = Float4::new(1.0, 2.5, 4.0, 0.0);
+
+        assert_eq!(p.vec_mul_affine_rev(&inv_m, t), r);
+        assert_eq!(p.vec_mul_affine_rev_fast(&inv_m, t), r);
+    }
+
+    #[test]
+    fn affine_mul_affine() {
+        let a = (
+            [
+                Float4::new(1.0, 3.0, 9.0, 0.0),
+                Float4::new(2.0, 6.0, 2.0, 0.0),
+                Float4::new(2.0, 7.0, 11.0, 0.0),
+            ],
+            Float4::new(1.5, 8.0, 12.0, 0.0),
+        );
+        let b = (
+            [
+                Float4::new(1.0, 2.0, 3.0, 0.0),
+                Float4::new(5.0, 6.0, 7.0, 0.0),
+                Float4::new(9.0, 10.0, 11.0, 0.0),
+            ],
+            Float4::new(13.0, 14.0, 15.0, 0.0),
+        );
+
+        let r = (
+            [
+                Float4::new(97.0, 110.0, 123.0, 0.0),
+                Float4::new(50.0, 60.0, 70.0, 0.0),
+                Float4::new(136.0, 156.0, 176.0, 0.0),
+            ],
+            Float4::new(162.5, 185.0, 207.5, 0.0),
+        );
+
+        assert_eq!(Float4::affine_mul_affine(&a.0, a.1, &b.0, b.1), r);
+        assert_eq!(Float4::affine_mul_affine_fast(&a.0, a.1, &b.0, b.1), r);
+    }
+
+    //------------
+    // Bool4
+
+    #[test]
+    fn bitmask() {
+        assert_eq!(Bool4::new(true, false, false, false).bitmask(), 0b0001);
+        assert_eq!(Bool4::new(false, true, false, false).bitmask(), 0b0010);
+        assert_eq!(Bool4::new(false, false, true, false).bitmask(), 0b0100);
+        assert_eq!(Bool4::new(false, false, false, true).bitmask(), 0b1000);
+
+        assert_eq!(Bool4::new(false, true, false, true).bitmask(), 0b1010);
+        assert_eq!(Bool4::new(true, false, true, false).bitmask(), 0b0101);
+    }
+
+    #[test]
+    fn to_bools() {
+        assert_eq!(
+            Bool4::new(true, false, false, false).to_bools(),
+            [true, false, false, false]
+        );
+        assert_eq!(
+            Bool4::new(false, true, false, false).to_bools(),
+            [false, true, false, false]
+        );
+        assert_eq!(
+            Bool4::new(false, false, true, false).to_bools(),
+            [false, false, true, false]
+        );
+        assert_eq!(
+            Bool4::new(false, false, false, true).to_bools(),
+            [false, false, false, true]
+        );
+
+        assert_eq!(
+            Bool4::new(false, true, false, true).to_bools(),
+            [false, true, false, true]
+        );
+        assert_eq!(
+            Bool4::new(true, false, true, false).to_bools(),
+            [true, false, true, false]
+        );
+    }
+
+    #[test]
+    fn any() {
+        assert_eq!(Bool4::new(true, false, false, false).any(), true);
+        assert_eq!(Bool4::new(false, true, false, false).any(), true);
+        assert_eq!(Bool4::new(false, false, true, false).any(), true);
+        assert_eq!(Bool4::new(false, false, false, true).any(), true);
+
+        assert_eq!(Bool4::new(false, false, false, false).any(), false);
+    }
+
+    #[test]
+    fn all() {
+        assert_eq!(Bool4::new(false, true, true, true).all(), false);
+        assert_eq!(Bool4::new(true, false, true, true).all(), false);
+        assert_eq!(Bool4::new(true, true, false, true).all(), false);
+        assert_eq!(Bool4::new(true, true, true, false).all(), false);
+
+        assert_eq!(Bool4::new(true, true, true, true).all(), true);
+    }
+
+    #[test]
+    fn boolean_ops() {
+        let all = Bool4::new(true, true, true, true);
+        let none = Bool4::new(false, false, false, false);
+        let a = Bool4::new(true, false, true, false);
+        let b = Bool4::new(false, true, false, true);
+
+        // Not.
+        assert_eq!(!a, b);
+        assert_eq!(!b, a);
+        assert_eq!(!all, none);
+        assert_eq!(!none, all);
+
+        // And.
+        assert_eq!(a & b, none);
+        assert_eq!(all & none, none);
+        assert_eq!(all & all, all);
+        assert_eq!(none & none, none);
+
+        // Or.
+        assert_eq!(a | b, all);
+        assert_eq!(all | none, all);
+        assert_eq!(all | all, all);
+        assert_eq!(none | none, none);
+
+        // Xor.
+        assert_eq!(a ^ b, all);
+        assert_eq!(all ^ none, all);
+        assert_eq!(all ^ all, none);
+        assert_eq!(none ^ none, none);
+    }
+
+    #[test]
+    fn matches_fallback() {
+        fn tf1(n: Float4) -> [f32; 4] {
+            [n.a(), n.b(), n.c(), n.d()]
+        }
+        fn tf2(n: fallback::Float4) -> [f32; 4] {
+            [n.a(), n.b(), n.c(), n.d()]
+        }
+
+        let a1 = Float4::new(1.53245, 5.4234523, -424.432, 0.0004231);
+        let b1 = Float4::new(74.63, -9.65436, 3.0, -1003.3);
+        let c1 = Float4::new(-0.4216, -132.52, 8.9452, 42.0);
+
+        let a2 = fallback::Float4::new(1.53245, 5.4234523, -424.432, 0.0004231);
+        let b2 = fallback::Float4::new(74.63, -9.65436, 3.0, -1003.3);
+        let c2 = fallback::Float4::new(-0.4216, -132.52, 8.9452, 42.0);
+
+        assert_eq!(tf1(a1), tf2(a2));
+        assert_eq!(tf1(b1), tf2(b2));
+        assert_eq!(tf1(c1), tf2(c2));
+
+        assert_eq!(tf1(a1 + b1), tf2(a2 + b2));
+        assert_eq!(tf1(a1 - b1), tf2(a2 - b2));
+        assert_eq!(tf1(a1 * b1), tf2(a2 * b2));
+        assert_eq!(tf1(a1 / b1), tf2(a2 / b2));
+        assert_eq!(tf1(a1.mul_add(b1, c1)), tf2(a2.mul_add(b2, c2)));
+        assert_eq!(tf1(a1.min(b1)), tf2(a2.min(b2)));
+        assert_eq!(tf1(a1.max(b1)), tf2(a2.max(b2)));
+        assert_eq!(a1.min_element(), a2.min_element());
+        assert_eq!(a1.max_element(), a2.max_element());
+        assert_eq!(tf1(a1.recip()), tf2(a2.recip()));
+        assert_eq!(tf1(a1.abs()), tf2(a2.abs()));
+    }
+}

sub_crates/rmath/src/wide4/sse.rs

@@ -0,0 +1,391 @@
+use std::ops::{Add, BitAnd, BitOr, BitXor, Div, Index, Mul, Neg, Not, Sub};
+
+use std::arch::x86_64::{
+    __m128, _mm_add_ps, _mm_and_ps, _mm_castsi128_ps, _mm_cmpeq_ps, _mm_cmpge_ps, _mm_cmpgt_ps,
+    _mm_cmple_ps, _mm_cmplt_ps, _mm_div_ps, _mm_fmadd_ps, _mm_max_ps, _mm_min_ps, _mm_movemask_ps,
+    _mm_mul_ps, _mm_or_ps, _mm_set1_epi32, _mm_set1_ps, _mm_set_epi32, _mm_set_ps, _mm_setzero_ps,
+    _mm_shuffle_ps, _mm_storeu_ps, _mm_sub_ps, _mm_xor_ps,
+};
+
+use crate::FMulAdd;
+
+//=============================================================
+// Float4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Float4(__m128);
+
+impl Float4 {
+    /// Create a new `Float4` with the given components.
+    #[inline(always)]
+    pub fn new(a: f32, b: f32, c: f32, d: f32) -> Self {
+        Self(unsafe { _mm_set_ps(d, c, b, a) })
+    }
+
+    /// Create a new `Float4` with all elements set to `n`.
+    #[inline(always)]
+    pub fn splat(n: f32) -> Self {
+        Self(unsafe { _mm_set1_ps(n) })
+    }
+
+    /// Component-wise fused multiply-add.
+    ///
+    /// `(self * a) + b` with only one rounding error.
+    #[inline(always)]
+    pub fn mul_add(self, a: Self, b: Self) -> Self {
+        if is_x86_feature_detected!("fma") {
+            Self(unsafe { _mm_fmadd_ps(self.0, a.0, b.0) })
+        } else {
+            Self::new(
+                self.a().mul_add(a.a(), b.a()),
+                self.b().mul_add(a.b(), b.b()),
+                self.c().mul_add(a.c(), b.c()),
+                self.d().mul_add(a.d(), b.d()),
+            )
+        }
+    }
+
+    /// Vertical minimum.
+    #[inline(always)]
+    pub fn min(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_min_ps(self.0, rhs.0) })
+    }
+
+    /// Vertical maximum.
+    #[inline(always)]
+    pub fn max(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_max_ps(self.0, rhs.0) })
+    }
+
+    /// Horizontal minimum.
+    #[inline(always)]
+    pub fn min_element(self) -> f32 {
+        let a = self.a().min(self.b());
+        let b = self.c().min(self.d());
+        a.min(b)
+    }
+
+    /// Horizontal maximum.
+    #[inline(always)]
+    pub fn max_element(self) -> f32 {
+        let a = self.a().max(self.b());
+        let b = self.c().max(self.d());
+        a.max(b)
+    }
+
+    /// 1.0 / self
+    #[inline(always)]
+    pub fn recip(self) -> Self {
+        // The reciprocal intrinsic is not precise enough.
+        // Self(unsafe { std::arch::x86_64::_mm_rcp_ps(self.0) })
+
+        Self::splat(1.0) / self
+    }
+
+    #[inline(always)]
+    pub fn abs(self) -> Self {
+        Self(unsafe {
+            let abs_mask = _mm_castsi128_ps(_mm_set1_epi32(!(1 << 31)));
+            _mm_and_ps(self.0, abs_mask)
+        })
+    }
+
+    //-----------------------------------------------------
+    // Comparisons.
+
+    /// Less than.
+    #[inline(always)]
+    pub fn cmplt(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmplt_ps(self.0, rhs.0) })
+    }
+
+    /// Less than or equal.
+    #[inline(always)]
+    pub fn cmplte(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmple_ps(self.0, rhs.0) })
+    }
+
+    /// Greater than.
+    #[inline(always)]
+    pub fn cmpgt(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpgt_ps(self.0, rhs.0) })
+    }
+
+    /// Greater than or equal.
+    #[inline(always)]
+    pub fn cmpgte(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpge_ps(self.0, rhs.0) })
+    }
+
+    /// Equal.
+    #[inline(always)]
+    pub fn cmpeq(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpeq_ps(self.0, rhs.0) })
+    }
+
+    //-----------------------------------------------------
+    // Individual components.
+
+    #[inline(always)]
+    pub fn a(self) -> f32 {
+        self[0]
+    }
+
+    #[inline(always)]
+    pub fn b(self) -> f32 {
+        self[1]
+    }
+
+    #[inline(always)]
+    pub fn c(self) -> f32 {
+        self[2]
+    }
+
+    #[inline(always)]
+    pub fn d(self) -> f32 {
+        self[3]
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_a(self, n: f32) -> Self {
+        Self::new(n, self.b(), self.c(), self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_b(self, n: f32) -> Self {
+        Self::new(self.a(), n, self.c(), self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_c(self, n: f32) -> Self {
+        Self::new(self.a(), self.b(), n, self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_d(self, n: f32) -> Self {
+        Self::new(self.a(), self.b(), self.c(), n)
+    }
+
+    //-----------------------------------------------------
+    // Shuffles.
+
+    #[inline(always)]
+    pub fn aaaa(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b00_00_00_00) })
+    }
+
+    #[inline(always)]
+    pub fn bbbb(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b01_01_01_01) })
+    }
+
+    #[inline(always)]
+    pub fn cccc(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b10_10_10_10) })
+    }
+
+    #[inline(always)]
+    pub fn dddd(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_11_11_11) })
+    }
+
+    #[inline(always)]
+    pub fn bcad(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_00_10_01) })
+    }
+
+    #[inline(always)]
+    pub fn cabd(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_01_00_10) })
+    }
+}
+
+impl Index<usize> for Float4 {
+    type Output = f32;
+
+    #[inline(always)]
+    fn index(&self, idx: usize) -> &f32 {
+        let elements: &[f32; 4] = unsafe { std::mem::transmute(&self.0) };
+        match idx {
+            0 => &elements[0],
+            1 => &elements[1],
+            2 => &elements[2],
+            3 => &elements[3],
+            _ => panic!("Out of bounds access of Float4 elements."),
+        }
+    }
+}
+
+impl Add for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_add_ps(self.0, rhs.0) })
+    }
+}
+
+impl Sub for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_sub_ps(self.0, rhs.0) })
+    }
+}
+
+impl Mul for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_mul_ps(self.0, rhs.0) })
+    }
+}
+
+impl Mul<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: f32) -> Self {
+        Self(unsafe { _mm_mul_ps(self.0, _mm_set1_ps(rhs)) })
+    }
+}
+
+impl Div for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_div_ps(self.0, rhs.0) })
+    }
+}
+
+impl Div<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: f32) -> Self {
+        Self(unsafe { _mm_div_ps(self.0, _mm_set1_ps(rhs)) })
+    }
+}
+
+impl Neg for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self(unsafe {
+            let abs_mask = _mm_castsi128_ps(_mm_set1_epi32(1 << 31));
+            _mm_xor_ps(self.0, abs_mask)
+        })
+    }
+}
+
+impl FMulAdd for Float4 {
+    #[inline(always)]
+    fn fma(self, b: Self, c: Self) -> Self {
+        self.mul_add(b, c)
+    }
+}
+
+//=============================================================
+// Bool4
+
+#[derive(Copy, Clone)]
+#[repr(transparent)]
+pub struct Bool4(__m128);
+
+impl Bool4 {
+    #[inline(always)]
+    pub fn new(a: bool, b: bool, c: bool, d: bool) -> Self {
+        const ONES: i32 = unsafe { std::mem::transmute(0xffffffffu32) };
+
+        unsafe {
+            let ints = _mm_set_epi32(
+                d as i32 * ONES,
+                c as i32 * ONES,
+                b as i32 * ONES,
+                a as i32 * ONES,
+            );
+            Bool4(_mm_castsi128_ps(ints))
+        }
+    }
+
+    #[inline(always)]
+    pub fn new_false() -> Self {
+        Self(unsafe { _mm_setzero_ps() })
+    }
+
+    #[inline(always)]
+    pub fn to_bools(self) -> [bool; 4] {
+        let mut v = [0.0f32; 4];
+        unsafe { _mm_storeu_ps((&mut v[..]).as_mut_ptr(), self.0) }
+        [
+            v[0].to_bits() != 0,
+            v[1].to_bits() != 0,
+            v[2].to_bits() != 0,
+            v[3].to_bits() != 0,
+        ]
+    }
+
+    /// Note: `a` goes to the least significant bit.
+    #[inline(always)]
+    pub fn bitmask(self) -> u8 {
+        unsafe { _mm_movemask_ps(self.0) as u8 }
+    }
+
+    #[inline(always)]
+    pub fn any(self) -> bool {
+        self.bitmask() != 0
+    }
+
+    #[inline(always)]
+    pub fn all(self) -> bool {
+        self.bitmask() == 0b1111
+    }
+}
+
+impl BitAnd for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitand(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_and_ps(self.0, rhs.0) })
+    }
+}
+
+impl BitOr for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitor(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_or_ps(self.0, rhs.0) })
+    }
+}
+
+impl BitXor for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitxor(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_xor_ps(self.0, rhs.0) })
+    }
+}
+
+impl Not for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn not(self) -> Self {
+        Self(unsafe {
+            let ones = _mm_castsi128_ps(_mm_set1_epi32(!0));
+            _mm_xor_ps(self.0, ones)
+        })
+    }
+}

sub_crates/rmath/src/xform.rs

@@ -0,0 +1,305 @@
+#![allow(dead_code)]
+
+use std::ops::{Add, Mul};
+
+use crate::point::Point;
+use crate::sealed::Sealed;
+use crate::wide4::Float4;
+
+/// A forward affine transform.
+///
+/// Use this for working with transforms that still need to be
+/// manipulated or composed with other transforms, or for storing
+/// transforms more compactly.
+///
+/// Note: slightly counter-intuitively, even though this can perform
+/// forward (but not inverse) transforms on points and vectors, it is
+/// capable of *inverse* (but not forward) transforms on surface normals.
+/// This is because forward transforms on surface normals require the
+/// inverse transform matrix.
+///
+/// Convert to an [`XformFull`] for a larger-format type capable of
+/// efficiently performing both forward and inverse transforms on all
+/// types, but which is effectively "frozen" in terms of further
+/// manipulation of the transform itself.
+#[derive(Debug, Copy, Clone, PartialEq)]
+#[repr(C)]
+pub struct Xform {
+    /// Rotation/scale/shear matrix.
+    pub m: [Float4; 3],
+
+    /// Translation.
+    pub t: Float4,
+}
+
+impl Xform {
+    /// Creates a new affine transform with the specified values:
+    ///
+    /// ```text
+    /// a d g j
+    /// b e h k
+    /// c f i l
+    /// ```
+    ///
+    /// Where j, k, and l are the xyz translation component.
+    #[inline]
+    #[allow(clippy::many_single_char_names)]
+    #[allow(clippy::too_many_arguments)]
+    pub fn new(
+        a: f32,
+        b: f32,
+        c: f32,
+        d: f32,
+        e: f32,
+        f: f32,
+        g: f32,
+        h: f32,
+        i: f32,
+        j: f32,
+        k: f32,
+        l: f32,
+    ) -> Self {
+        Self {
+            m: [
+                Float4::new(a, b, c, 0.0),
+                Float4::new(d, e, f, 0.0),
+                Float4::new(g, h, i, 0.0),
+            ],
+            t: Float4::new(j, k, l, 0.0),
+        }
+    }
+
+    /// Creates a new identity transform.
+    #[inline]
+    pub fn identity() -> Self {
+        Self {
+            m: [
+                Float4::new(1.0, 0.0, 0.0, 0.0),
+                Float4::new(0.0, 1.0, 0.0, 0.0),
+                Float4::new(0.0, 0.0, 1.0, 0.0),
+            ],
+            t: Float4::splat(0.0),
+        }
+    }
+
+    #[inline]
+    pub fn from_location(loc: Point) -> Xform {
+        Self {
+            m: [
+                Float4::new(1.0, 0.0, 0.0, 0.0),
+                Float4::new(0.0, 1.0, 0.0, 0.0),
+                Float4::new(0.0, 0.0, 1.0, 0.0),
+            ],
+            t: loc.0,
+        }
+    }
+
+    /// Returns whether the matrices are approximately equal to each other.
+    /// Each corresponding element in the matrices cannot have a relative
+    /// error exceeding epsilon.
+    pub(crate) fn aprx_eq(&self, other: Xform, max_ulps: u32) -> bool {
+        let mut eq = true;
+        eq &= Float4::aprx_eq(self.m[0], other.m[0], max_ulps);
+        eq &= Float4::aprx_eq(self.m[1], other.m[1], max_ulps);
+        eq &= Float4::aprx_eq(self.m[2], other.m[2], max_ulps);
+        eq &= Float4::aprx_eq(self.t, other.t, max_ulps);
+        eq
+    }
+
+    /// Compute the "full" version of the transform.
+    #[inline]
+    pub fn to_full(&self) -> Option<XformFull> {
+        if let Some(inv_m) = Float4::invert_3x3(&self.m) {
+            Some(XformFull {
+                fwd: *self,
+                inv_m: inv_m,
+            })
+        } else {
+            None
+        }
+    }
+
+    /// Faster but less precise version of `to_full()`.
+    #[inline]
+    pub fn to_full_fast(&self) -> Option<XformFull> {
+        if let Some(inv_m) = Float4::invert_3x3_fast(&self.m) {
+            Some(XformFull {
+                fwd: *self,
+                inv_m: inv_m,
+            })
+        } else {
+            None
+        }
+    }
+
+    /// Composes two transforms together.
+    ///
+    /// The resulting transform is the same as doing `self` and then
+    /// `rhs` in sequence.
+    #[inline]
+    pub fn compose(&self, rhs: &Self) -> Self {
+        let (m, t) = Float4::affine_mul_affine(&self.m, self.t, &rhs.m, rhs.t);
+        Self { m: m, t: t }
+    }
+
+    /// Composes two transforms together.
+    ///
+    /// Faster but less precise version.
+    #[inline]
+    pub fn compose_fast(&self, rhs: &Self) -> Self {
+        let (m, t) = Float4::affine_mul_affine_fast(&self.m, self.t, &rhs.m, rhs.t);
+        Self { m: m, t: t }
+    }
+}
+
+impl Default for Xform {
+    fn default() -> Self {
+        Self::identity()
+    }
+}
+
+/// Multiply a matrix by a f32
+impl Mul<f32> for Xform {
+    type Output = Self;
+
+    #[inline]
+    fn mul(self, rhs: f32) -> Self {
+        Self {
+            m: [self.m[0] * rhs, self.m[1] * rhs, self.m[2] * rhs],
+            t: self.t * rhs,
+        }
+    }
+}
+
+/// Add two matrices together
+impl Add for Xform {
+    type Output = Self;
+
+    #[inline]
+    fn add(self, rhs: Self) -> Self {
+        Self {
+            m: [
+                self.m[0] + rhs.m[0],
+                self.m[1] + rhs.m[1],
+                self.m[2] + rhs.m[2],
+            ],
+            t: self.t + rhs.t,
+        }
+    }
+}
+
+impl AsXform for Xform {
+    #[inline(always)]
+    fn as_xform(&self) -> &Xform {
+        self
+    }
+}
+
+impl Sealed for Xform {}
+
+//-------------------------------------------------------------
+
+/// A combined forward/inverse affine transform.
+///
+/// Unlike [`Xform`], this can perform both forward and inverse
+/// transforms on all types.  However, it also takes up more space and
+/// is effectively "frozen" in terms of further manipulation.  Prefer
+/// [`Xform`] when manipulating or composing transforms, and also
+/// when storing transforms if space is a consideration.
+///
+/// Note: only the 3x3 part of the transform is stored inverted.  This
+/// is because it's both trivial and more numerically stable to reuse
+/// the forward translation vector to do inverse transforms, as
+/// `(point - fwd.t) * inv_m`.
+#[derive(Debug, Copy, Clone)]
+#[repr(C)]
+pub struct XformFull {
+    /// Forward transform.
+    pub fwd: Xform,
+
+    /// Inverse rotation/scale/shear matrix.
+    pub inv_m: [Float4; 3],
+}
+
+impl XformFull {
+    pub fn identity() -> Self {
+        Self {
+            fwd: Xform {
+                m: [
+                    Float4::new(1.0, 0.0, 0.0, 0.0),
+                    Float4::new(0.0, 1.0, 0.0, 0.0),
+                    Float4::new(0.0, 0.0, 1.0, 0.0),
+                ],
+                t: Float4::splat(0.0),
+            },
+            inv_m: [
+                Float4::new(1.0, 0.0, 0.0, 0.0),
+                Float4::new(0.0, 1.0, 0.0, 0.0),
+                Float4::new(0.0, 0.0, 1.0, 0.0),
+            ],
+        }
+    }
+}
+
+impl AsXform for XformFull {
+    #[inline(always)]
+    fn as_xform(&self) -> &Xform {
+        &self.fwd
+    }
+}
+
+impl Sealed for XformFull {}
+
+//-------------------------------------------------------------
+
+pub trait AsXform: Sealed {
+    fn as_xform(&self) -> &Xform;
+}
+
+//-------------------------------------------------------------
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn equality() {
+        let a = Xform::identity();
+        let b = Xform::identity();
+        let c = Xform::new(1.1, 0.0, 0.0, 0.0, 0.0, 1.1, 0.0, 0.0, 0.0, 0.0, 1.1, 0.0);
+
+        assert_eq!(a, b);
+        assert!(a != c);
+    }
+
+    #[test]
+    fn approximate_equality() {
+        let a = Xform::identity();
+        let b = Xform::new(
+            1.000001, 0.0, 0.0, 0.0, 1.000001, 0.0, 0.0, 0.0, 1.000001, 0.0, 0.0, 0.0,
+        );
+        let c = Xform::new(
+            1.000003, 0.0, 0.0, 0.0, 1.000003, 0.0, 0.0, 0.0, 1.000003, 0.0, 0.0, 0.0,
+        );
+
+        assert!(a.aprx_eq(b, 10));
+        assert!(!a.aprx_eq(b, 6));
+
+        assert!(a.aprx_eq(c, 27));
+        assert!(!a.aprx_eq(c, 23));
+    }
+
+    #[test]
+    fn compose() {
+        let a = Xform::new(1.0, 3.0, 9.0, 2.0, 6.0, 2.0, 2.0, 7.0, 11.0, 1.5, 8.0, 12.0);
+        let b = Xform::new(
+            1.0, 2.0, 3.0, 5.0, 6.0, 7.0, 9.0, 10.0, 11.0, 13.0, 14.0, 15.0,
+        );
+        let c = Xform::new(
+            97.0, 110.0, 123.0, 50.0, 60.0, 70.0, 136.0, 156.0, 176.0, 162.5, 185.0, 207.5,
+        );
+
+        assert_eq!(a.compose(&b), c);
+        assert_eq!(a.compose_fast(&b), c);
+    }
+}

sub_crates/rrand/Cargo.toml

@@ -0,0 +1,12 @@
+[package]
+name = "rrand"
+version = "0.1.0"
+edition = "2021"
+
+[dev-dependencies]
+bencher = "0.1.5"
+rand = "0.6"
+
+[[bench]]
+name = "bench"
+harness = false

sub_crates/rrand/benches/bench.rs

@@ -0,0 +1,67 @@
+use bencher::{benchmark_group, benchmark_main, black_box, Bencher};
+use rrand::{mix32, mix32_seed, mix64, mix64_seed, Rng};
+
+//----
+
+fn rng_u32_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        let mut rng = Rng::new(black_box(0));
+        for _ in 0..100000 {
+            black_box(rng.u32());
+        }
+    });
+}
+
+fn rng_u64_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        let mut rng = Rng::new(black_box(0));
+        for _ in 0..100000 {
+            black_box(rng.u64());
+        }
+    });
+}
+
+fn mix32_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        for i in 0..100000 {
+            black_box(mix32(black_box(i)));
+        }
+    });
+}
+
+fn mix64_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        for i in 0..100000 {
+            black_box(mix64(black_box(i)));
+        }
+    });
+}
+
+fn mix32_seed_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        for i in 0..100000 {
+            black_box(mix32_seed(black_box(i), black_box(0)));
+        }
+    });
+}
+
+fn mix64_seed_100000(bench: &mut Bencher) {
+    bench.iter(|| {
+        for i in 0..100000 {
+            black_box(mix64_seed(black_box(i), black_box(0)));
+        }
+    });
+}
+
+//----
+
+benchmark_group!(
+    benches,
+    rng_u32_100000,
+    rng_u64_100000,
+    mix32_100000,
+    mix64_100000,
+    mix32_seed_100000,
+    mix64_seed_100000,
+);
+benchmark_main!(benches);

sub_crates/rrand/src/lib.rs

@@ -0,0 +1,127 @@
+//! Sources of deterministic "randomness" for rendering applications.
+
+/// Convert a `u32` to a float in [0.0, 1.0).
+///
+/// Use for getting f32 values from random u32 sources.
+///
+/// Note: this is a linear mapping from [0, int_max] to [0.0, 1.0).
+#[inline(always)]
+pub fn u32_to_f32_norm(n: u32) -> f32 {
+    f32::from_bits((n >> 9) | 0x3f800000) - 1.0
+}
+
+//-------------------------------------------------------------
+
+/// A fast RNG.
+///
+#[derive(Debug, Copy, Clone)]
+pub struct Rng {
+    state: u64,
+}
+
+impl Rng {
+    /// Creates a new Rng from a seed.
+    ///
+    /// A seed of zero is perfectly fine, and does not affect the quality
+    /// of the generator.
+    #[inline]
+    pub fn new(seed: u64) -> Self {
+        Self { state: seed }
+    }
+
+    /// Gets the nth relative RNG stream from this one.
+    ///
+    /// The returned stream will be at the same point in its sequence as
+    /// this one.
+    #[inline]
+    pub fn nth_stream(&self, n: u64) -> Self {
+        Self {
+            // We just jump forward 2^40*n states.  This gives us 2^24
+            // unique streams, each of which is 2^40 numbers long.
+            state: self
+                .state
+                .wrapping_add(0xa0761d6478bd642f_u64.wrapping_mul(1 << 40).wrapping_mul(n)),
+        }
+    }
+
+    /// Returns a random u32 in [0, int_max].
+    #[inline(always)]
+    pub fn u32(&mut self) -> u32 {
+        self.u64() as u32
+    }
+
+    /// Returns a random u64 in [0, int_max].
+    #[inline(always)]
+    pub fn u64(&mut self) -> u64 {
+        // The wyrand RNG.
+        self.state = self.state.wrapping_add(0xa0761d6478bd642f);
+        let t = (self.state as u128).wrapping_mul(self.state as u128 ^ 0xe7037ed1a0b428db);
+        ((t >> 64) ^ t) as u64
+    }
+
+    /// Returns a random f32 in [0.0, 1.0).
+    #[inline(always)]
+    pub fn f32(&mut self) -> f32 {
+        u32_to_f32_norm(self.u32())
+    }
+}
+
+//-------------------------------------------------------------
+
+/// A fast 32-bit mixing function.
+///
+/// Scrambles the input number to produce a different deterministic
+/// "random" number.
+#[inline(always)]
+pub fn mix32(mut n: u32) -> u32 {
+    // From https://github.com/skeeto/hash-prospector
+    n ^= n >> 16;
+    n = n.wrapping_mul(0x21f0aaad);
+    n ^= n >> 15;
+    n = n.wrapping_mul(0xd35a2d97);
+    n ^= n >> 15;
+
+    // Xor by a random number so input zero doesn't map to output zero.
+    // The particular number used here isn't special.
+    n ^ 0xe6fe3beb
+}
+
+/// A fast seedable 32-bit mixing function.
+///
+/// Same as `mix32()` but takes a seed.
+#[inline(always)]
+pub fn mix32_seed(n: u32, seed: u32) -> u32 {
+    // We rotate the bits of `seed` so it's unlikely to interact with `n`
+    // in bad ways if they're both e.g. incrementing.  The particular
+    // rotation constant used here isn't special.
+    mix32(n ^ seed.rotate_left(23))
+}
+
+/// A fast 64-bit mixing function.
+///
+/// Scrambles the input number to produce a different deterministic
+/// "random" number.
+#[inline(always)]
+pub fn mix64(mut n: u64) -> u64 {
+    // From https://zimbry.blogspot.com/2011/09/better-bit-mixing-improving-on.html
+    n ^= n >> 30;
+    n = n.wrapping_mul(0xbf58476d1ce4e5b9);
+    n ^= n >> 27;
+    n = n.wrapping_mul(0x94d049bb133111eb);
+    n ^= n >> 31;
+
+    // Xor by a random number so input zero doesn't map to output zero.
+    // The particular number used here isn't special.
+    n ^ 0x4acc3f27cc712c9d
+}
+
+/// A fast seedable 64-bit mixing function.
+///
+/// Same as `mix64()` but takes a seed.
+#[inline(always)]
+pub fn mix64_seed(n: u64, seed: u64) -> u64 {
+    // We rotate the bits of `seed` so it's unlikely to interact with `n`
+    // in bad ways if they're both e.g. incrementing.  The particular
+    // rotation constant used here isn't special.
+    mix64(n ^ seed.rotate_left(47))
+}

sub_crates/spectral_upsampling/Cargo.toml

@@ -10,4 +10,4 @@ name = "spectral_upsampling"
 path = "src/lib.rs"
 
 [dependencies]
-glam = "0.15"
+rmath = { path = "../rmath" }

sub_crates/spectral_upsampling/src/jakob.rs

@@ -6,7 +6,9 @@
 /// The provides similar color matching as full Jakob, at the expense of
 /// somewhat lower quality spectrums, and the inability to precalculate
 /// the coefficents for even more efficient evaluation later on.
-use glam::Vec4;
+use rmath::wide4::Float4;
+
+pub const EQUAL_ENERGY_REFLECTANCE: f32 = 1.0;
 
 /// How many polynomial coefficients?
 const RGB2SPEC_N_COEFFS: usize = 3;
@@ -15,7 +17,7 @@ const RGB2SPEC_N_COEFFS: usize = 3;
 include!(concat!(env!("OUT_DIR"), "/jakob_table_inc.rs"));
 
 #[inline]
-pub fn rec709_to_spectrum_p4(lambdas: Vec4, rgb: (f32, f32, f32)) -> Vec4 {
+pub fn rec709_to_spectrum_p4(lambdas: Float4, rgb: (f32, f32, f32)) -> Float4 {
     small_rgb_to_spectrum_p4(
         REC709_TABLE,
         REC709_TABLE_RES,
@@ -26,7 +28,7 @@ pub fn rec709_to_spectrum_p4(lambdas: Vec4, rgb: (f32, f32, f32)) -> Vec4 {
 }
 
 #[inline]
-pub fn rec2020_to_spectrum_p4(lambdas: Vec4, rgb: (f32, f32, f32)) -> Vec4 {
+pub fn rec2020_to_spectrum_p4(lambdas: Float4, rgb: (f32, f32, f32)) -> Float4 {
     small_rgb_to_spectrum_p4(
         REC2020_TABLE,
         REC2020_TABLE_RES,
@@ -37,7 +39,7 @@ pub fn rec2020_to_spectrum_p4(lambdas: Vec4, rgb: (f32, f32, f32)) -> Vec4 {
 }
 
 #[inline]
-pub fn aces_to_spectrum_p4(lambdas: Vec4, rgb: (f32, f32, f32)) -> Vec4 {
+pub fn aces_to_spectrum_p4(lambdas: Float4, rgb: (f32, f32, f32)) -> Float4 {
     small_rgb_to_spectrum_p4(
         ACES_TABLE,
         ACES_TABLE_RES,
@@ -56,9 +58,9 @@ fn small_rgb_to_spectrum_p4(
     table: &[[(f32, f32, f32); 2]],
     table_res: usize,
     table_mid_value: f32,
-    lambdas: Vec4,
+    lambdas: Float4,
     rgb: (f32, f32, f32),
-) -> Vec4 {
+) -> Float4 {
     // Determine largest RGB component, and calculate the other two
     // components scaled for lookups.
     let (i, max_val, x, y) = if rgb.0 > rgb.1 && rgb.0 > rgb.2 {
@@ -71,7 +73,7 @@ fn small_rgb_to_spectrum_p4(
     if max_val == 0.0 {
         // If max_val is zero, just return zero.  This avoids NaN's from
         // divide by zero.  This is also correct, since it's black.
-        return Vec4::splat(0.0);
+        return Float4::splat(0.0);
     }
     let x = x * 63.0 / max_val;
     let y = y * 63.0 / max_val;
@@ -91,20 +93,20 @@ fn small_rgb_to_spectrum_p4(
 
     // Convert to SIMD format for faster interpolation.
     let a0 = [
-        Vec4::new(a0[0].0, a0[0].1, a0[0].2, 0.0),
-        Vec4::new(a0[1].0, a0[1].1, a0[1].2, 0.0),
+        Float4::new(a0[0].0, a0[0].1, a0[0].2, 0.0),
+        Float4::new(a0[1].0, a0[1].1, a0[1].2, 0.0),
     ];
     let a1 = [
-        Vec4::new(a1[0].0, a1[0].1, a1[0].2, 0.0),
-        Vec4::new(a1[1].0, a1[1].1, a1[1].2, 0.0),
+        Float4::new(a1[0].0, a1[0].1, a1[0].2, 0.0),
+        Float4::new(a1[1].0, a1[1].1, a1[1].2, 0.0),
     ];
     let a2 = [
-        Vec4::new(a2[0].0, a2[0].1, a2[0].2, 0.0),
-        Vec4::new(a2[1].0, a2[1].1, a2[1].2, 0.0),
+        Float4::new(a2[0].0, a2[0].1, a2[0].2, 0.0),
+        Float4::new(a2[1].0, a2[1].1, a2[1].2, 0.0),
     ];
     let a3 = [
-        Vec4::new(a3[0].0, a3[0].1, a3[0].2, 0.0),
-        Vec4::new(a3[1].0, a3[1].1, a3[1].2, 0.0),
+        Float4::new(a3[0].0, a3[0].1, a3[0].2, 0.0),
+        Float4::new(a3[1].0, a3[1].1, a3[1].2, 0.0),
     ];
 
     // Do interpolation.
@@ -133,22 +135,22 @@ fn small_rgb_to_spectrum_p4(
 // Coefficient -> eval functions
 
 #[inline(always)]
-fn rgb2spec_fma_4(a: Vec4, b: Vec4, c: Vec4) -> Vec4 {
-    (a * b) + c
+fn rgb2spec_fma_4(a: Float4, b: Float4, c: Float4) -> Float4 {
+    a.mul_add(b, c)
 }
 
-fn rgb2spec_eval_4(coeff: [f32; RGB2SPEC_N_COEFFS], lambda: Vec4) -> Vec4 {
-    let co0 = Vec4::splat(coeff[0]);
-    let co1 = Vec4::splat(coeff[1]);
-    let co2 = Vec4::splat(coeff[2]);
+fn rgb2spec_eval_4(coeff: [f32; RGB2SPEC_N_COEFFS], lambda: Float4) -> Float4 {
+    let co0 = Float4::splat(coeff[0]);
+    let co1 = Float4::splat(coeff[1]);
+    let co2 = Float4::splat(coeff[2]);
 
     let x = rgb2spec_fma_4(rgb2spec_fma_4(co0, lambda, co1), lambda, co2);
 
     let y = {
         // TODO: replace this with a SIMD sqrt op.
-        let (x, y, z, w) = rgb2spec_fma_4(x, x, Vec4::splat(1.0)).into();
-        Vec4::new(x.sqrt(), y.sqrt(), z.sqrt(), w.sqrt()).recip()
+        let (x, y, z, w) = rgb2spec_fma_4(x, x, Float4::splat(1.0)).into();
+        Float4::new(x.sqrt(), y.sqrt(), z.sqrt(), w.sqrt()).recip()
     };
 
-    rgb2spec_fma_4(Vec4::splat(0.5) * x, y, Vec4::splat(0.5))
+    rgb2spec_fma_4(Float4::splat(0.5) * x, y, Float4::splat(0.5))
 }

sub_crates/spectral_upsampling/src/meng.rs

@@ -6,7 +6,7 @@
 
 use std::f32;
 
-use glam::Vec4;
+use rmath::wide4::Float4;
 
 mod meng_spectra_tables;
 
@@ -174,7 +174,7 @@ pub fn spectrum_xyz_to_p(lambda: f32, xyz: (f32, f32, f32)) -> f32 {
 ///
 /// Works on 4 wavelengths at once via SIMD.
 #[inline]
-pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
+pub fn spectrum_xyz_to_p_4(lambdas: Float4, xyz: (f32, f32, f32)) -> Float4 {
     assert!(lambdas.min_element() >= SPECTRUM_SAMPLE_MIN);
     assert!(lambdas.max_element() <= SPECTRUM_SAMPLE_MAX);
 
@@ -184,7 +184,7 @@ pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
         if norm < f32::MAX {
             norm
         } else {
-            return Vec4::splat(0.0);
+            return Float4::splat(0.0);
         }
     };
 
@@ -197,7 +197,7 @@ pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
         || uv.1 < 0.0
         || uv.1 >= SPECTRUM_GRID_HEIGHT as f32
     {
-        return Vec4::splat(0.0);
+        return Float4::splat(0.0);
     }
 
     let uvi = (uv.0 as i32, uv.1 as i32);
@@ -214,11 +214,11 @@ pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
 
     // If the cell has no points, nothing we can do, so return 0.0
     if num == 0 {
-        return Vec4::splat(0.0);
+        return Float4::splat(0.0);
     }
 
     // Normalize lambda to spectrum table index range.
-    let sb: Vec4 = (lambdas - Vec4::splat(SPECTRUM_SAMPLE_MIN))
+    let sb: Float4 = (lambdas - Float4::splat(SPECTRUM_SAMPLE_MIN))
         / (SPECTRUM_SAMPLE_MAX - SPECTRUM_SAMPLE_MIN)
         * (SPECTRUM_NUM_SAMPLES as f32 - 1.0);
     debug_assert!(sb.min_element() >= 0.0);
@@ -226,7 +226,7 @@ pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
 
     // Get the spectral values for the vertices of the grid cell.
     // TODO: use integer SIMD intrinsics to make this part faster.
-    let mut p = [Vec4::splat(0.0); 6];
+    let mut p = [Float4::splat(0.0); 6];
     let sb0: [i32; 4] = [sb[0] as i32, sb[1] as i32, sb[2] as i32, sb[3] as i32];
     assert!(sb0[0].max(sb0[1]).max(sb0[2].max(sb0[3])) < SPECTRUM_NUM_SAMPLES);
     let sb1: [i32; 4] = [
@@ -235,27 +235,27 @@ pub fn spectrum_xyz_to_p_4(lambdas: Vec4, xyz: (f32, f32, f32)) -> Vec4 {
         (sb[2] as i32 + 1).min(SPECTRUM_NUM_SAMPLES - 1),
         (sb[3] as i32 + 1).min(SPECTRUM_NUM_SAMPLES - 1),
     ];
-    let sbf = sb - Vec4::new(sb0[0] as f32, sb0[1] as f32, sb0[2] as f32, sb0[3] as f32);
+    let sbf = sb - Float4::new(sb0[0] as f32, sb0[1] as f32, sb0[2] as f32, sb0[3] as f32);
     for i in 0..(num as usize) {
         debug_assert!(idx[i] >= 0);
         let spectrum = &SPECTRUM_DATA_POINTS[idx[i] as usize].spectrum;
-        let p0 = Vec4::new(
+        let p0 = Float4::new(
             spectrum[sb0[0] as usize],
             spectrum[sb0[1] as usize],
             spectrum[sb0[2] as usize],
             spectrum[sb0[3] as usize],
         );
-        let p1 = Vec4::new(
+        let p1 = Float4::new(
             spectrum[sb1[0] as usize],
             spectrum[sb1[1] as usize],
             spectrum[sb1[2] as usize],
             spectrum[sb1[3] as usize],
         );
-        p[i] = p0 * (Vec4::splat(1.0) - sbf) + p1 * sbf;
+        p[i] = p0 * (Float4::splat(1.0) - sbf) + p1 * sbf;
     }
 
     // Linearly interpolate the spectral power of the cell vertices.
-    let mut interpolated_p = Vec4::splat(0.0);
+    let mut interpolated_p = Float4::splat(0.0);
     if inside {
         // Fast path for normal inner quads:
         let uv2 = (uv.0 - uvi.0 as f32, uv.1 - uvi.1 as f32);