f84d093f66
PsychoBlend now updates render in realtime within Blender.
...
Also, cancelling renders is much more responsive now, especially
during long exports.
2017-06-03 00:00:19 -07:00
3beffab507
Update to latest openexr-rs, and switch to pure Rust PNG output crate.
2017-05-28 17:51:01 -07:00
f0604a69ae
Updated to latest openexr crate, and switched to 16-bit float EXR's.
2017-05-20 23:48:01 -07:00
4e2d3e64c0
Switched to using the openexr crate on crates.io.
2017-05-17 00:36:35 -07:00
edb71864e8
LDS sampling falls back on random sampling at higher dimensions.
...
This is more a piece-of-mind thing than anything else. But it
also lets us make the number of LDS dimensions lower without
worrying, which in turn makes the code smaller.
2017-05-14 16:06:54 -07:00
322627641a
Broke render job out into its own method.
...
The left-drift of having it inline was getting cumbersome.
2017-05-14 14:51:55 -07:00
922e33ec3f
Reformat code with latest rustfmt and custom config.
2017-05-14 13:43:51 -07:00
993ba719d7
Fixed bug that was causing spurious NaN pixels.
...
It was because sometimes the PDF's for a sample could be zero,
leading to a divide by zero. PDF = 0.0 is now checked for.
2017-05-14 12:45:43 -07:00
567b658b6c
Removed blue noise dithered sampling.
2017-05-14 12:37:01 -07:00
d71fd3b5c8
Implemented blue noise dithered sampling. Temporary.
...
After implementation, it does appear to make rendering slower
by a noticable bit compared to what I was doing before. At very
low sampling rates it does provide a bit of visual improvement,
but by the time you get to even just 16 samples per pixel its
benefits seem to disappear.
Due to the slow down and the minimal gains, I'll be removing
this in the next commit. But I want to commit it so I don't
lose the code, since it was an interesting experiment with
some promising results.
2017-05-14 12:25:01 -07:00
d8a33c7bfa
Added a blue noise mask generator as a sub-crate.
2017-05-13 03:23:55 -07:00
172e2f19ef
Removed BVH4 and the related bitstack from AccelRay.
...
I couldn't make the BVH4 faster than the BVH, and the bitstack
was bloating the AccelRay struct. Removing the bitstack gives
a small but noticable speedup in rendering.
2017-05-12 21:07:40 -07:00
88578b9eae
Moved Float4 and Matrix and Vector math into their own sub-crates.
2017-05-12 21:01:54 -07:00
04e8a6ca73
Added a Rust port of Leonhard Grünschloß's Sobol sampler.
...
The Halton sampler appears to be better, but it was fun to add
this anyway!
2017-05-11 23:56:12 -07:00
b698a52f6c
Split some more things out into sub-crates.
...
Also translated the Halton generator to rust and made it a crate
where the code is generated by a build.rs file.
2017-05-11 22:57:59 -07:00
96db00b10a
Refactored some parts of Renderer to take up less memory and be faster.
...
Specifically, LightPath is now significantly smaller, and
resultingly faster to process.
Also finally fixed the bug where without light sources the light
from the sky wouldn't affect surfaces.
2017-05-11 20:23:19 -07:00
a3a19e53ef
Some misc small optimizations.
2017-05-11 01:22:37 -07:00
dcf25b92af
Fiddled with BVHNode to make it 32 bytes instead of 40 bytes.
...
This helps reduce the total size of BVHs since both BBoxes and
and BVHNodes are now 32 bytes and can be properly packed together.
2017-05-08 19:47:10 -07:00
d8e31750a8
Small optimization in BVH traversal.
2017-04-26 01:03:16 -07:00
8b0c422630
Small optimization in BBox intersection and bugfix in BBox4 intersection.
2017-04-26 00:28:43 -07:00
c92a8c4da0
During BVH construction, merge BBox time samples based on a threshold.
...
If the average surface area of all the time samples is close enough
to the surface area of their union, just take the union and use that.
This both makes the BVH smaller in memory (time samples don't
propigate up the tree beyond their usefulness) and makes it
faster since traversal can avoid interpolating BBoxes when there's
only one BBox for a node.
2017-04-23 23:15:31 -07:00
b135e8beb8
Fixed bug in new BVH4 traversal code.
2017-04-22 22:32:36 -07:00
a940630a27
Reduced max bvh tree depth.
...
Reduced from 64 to 42. This still allows each BVH to hold 4.4
trillion elements, but it guarantees that the accel ray's
traversal bitstack can accommodate at least two nested max-depth
trees.
2017-04-22 12:51:35 -07:00
0cbf851451
Fixed a commandline argument that was broken.
...
Also removed a redundant statistics printout.
2017-04-22 12:50:28 -07:00
fdf2b4babf
Print stats for BVH and BVH4 traversal, and working on speeding up BVH4 traversal.
2017-04-22 11:18:29 -07:00
597bcf0518
Sped up splitting rays by direction.
...
Turns out the is_sign_positive() method is slower than a simple
comparison.
2017-04-22 11:15:17 -07:00
8b14e46bc1
Added more detailed render stats.
2017-04-22 10:31:27 -07:00
7884cff604
Added unit tests for BitStack128.
2017-04-16 23:13:34 -07:00
725bbca6a1
Eliminated some unnecessary left drift in BVH4 code.
...
IMO this is much easier to read.
2017-04-16 17:27:29 -07:00
92fd83a766
Actually using the None match arm in BVH4 traversal now.
...
It seems to maybe be just a hair faster. But, also, it sets the
code up to try some more optimizations.
2017-04-16 17:22:51 -07:00
f93dde1810
Changed the BBox4 data layout to give a small speed boost.
2017-04-16 15:54:02 -07:00
2c4f103db5
Factored some calculations into the outer loop of BVH4 traversal.
...
Provides a small but noticable speed boost.
2017-04-16 00:51:13 -07:00
7d5bda2a3b
Print newline before exit.
...
Makes it easier to scan commandline history for multiple renders.
2017-04-14 23:22:38 -07:00
a6d8de43d7
Make BVH4 calculate and store its node count.
...
Useful for debugging.
2017-04-14 22:14:52 -07:00
0296fd2d34
Fixed a bunch of problems that came up when running debug build.
...
The worst offender was an obvious and nasty bug in the bitstack
implementation. Hooray for debug asserts!
2017-04-14 21:37:29 -07:00
44e3ee8b4b
Fixed nasty bug in BVH4 building.
...
In practice it worked fine, but only by accident. NaN's were
being passed to the lerp_slice function, which led to the
correct result in this case but is icky and dependant
on how lerp_slice is implemented.
2017-04-14 20:25:12 -07:00
f8d47dc24a
Fixed some things for building with SIMD support.
2017-04-13 23:28:26 -07:00
7f761a9489
Added custom aligned allocations to MemArena.
...
Used in BVH4.
2017-04-13 23:06:56 -07:00
8b1e3a3cdb
Added some dev printing info.
2017-04-13 22:50:47 -07:00
573c5da5ab
Got BVH4 working. Woo hoo!
2017-04-13 22:44:34 -07:00
53a14996c9
Completed first attempt at implementing BVH4.
...
Completely untested, so almost certainly non-functional. But it
compiles!
2017-04-12 23:33:34 -07:00
e0c6dfbd76
Beginnings of BVH4 implementation.
...
- Created a traversal-order table.
- Created a function to help index into that table.
2017-04-11 23:10:54 -07:00
a2d7f9e149
Added BBox4 and BVH4 types.
...
BVH4 is just a skeleton (duplicated BVH) right now. But BBox4
is complete (pending testing).
2017-04-11 14:22:11 -07:00
c05bae2636
Created a BitStack128 type and added it to Ray.
...
This will be used for BVH4 traversal.
2017-04-11 11:13:23 -07:00
f5ba119562
Switched to clap for command line argument parsing.
2017-04-11 10:15:32 -07:00
d9564bc005
Refactor and rework of BVH.
...
The BVH building code is now largely split out into a separate
type, BVHBase. The intent is that this will also be used by
the BVH4 when I get around to it.
The BVH itself now uses references instead of indexes, allocating
and pointing directly into the MemArena. This allows the nodes
to all be right next to their bounding boxes in memory.
2017-04-10 23:41:38 -07:00
db3e4ad129
MemArena now slowly increases the size of each subsequent block.
...
This seems to work more nicely than a fixed block size, because
it adapts to how much memory is being requested over-all. For
example, a small scene won't allocate large amounts of RAM,
but a large scene with large data won't be penalized with a
lot of tiny allocations.
2017-04-10 20:49:55 -07:00
70f715ec2d
MemArena now collects statistics about its usage.
...
You can print out statistics about the main arena used in
Psychopath by passing the new --stats flag.
2017-04-10 18:34:08 -07:00
6623443e2e
Improved .psy file parsing error messages.
...
Biggest improvement: it gives you line numbers. But also progress
on better descriptions.
2017-04-10 14:03:01 -07:00
d3b3a50a4d
LightTree and Assemblies use MemArena now.
2017-04-10 00:03:49 -07:00
c82c821b31
BVH and objects now use MemArena.
2017-04-09 23:33:36 -07:00
e9e202933f
Assembly type now partially uses MemArena.
2017-04-09 18:27:01 -07:00
fc15fa9192
World type uses memory arena now.
2017-04-09 17:56:06 -07:00
57aa9be496
First steps towards using the memory arena allocator.
...
Changed Camera to us the memory arena allocator, and worked out a
bunch of stuff through doing that.
2017-04-09 17:31:17 -07:00
f6ffd8d30b
Fixed a typo in a comment.
2017-03-29 13:20:58 -07:00
7223f69151
Implemented a more efficient orthonormal basis function.
...
Algorithm taken from the paper "Building an Orthonormal
Basis, Revisited" by Duff et al.
2017-03-29 13:19:51 -07:00
4a86c4122a
Laying some groundwork for a simple shading system.
2017-03-14 00:27:25 -07:00
114f11c583
Fixed bug in GTRClosure that was resulting in all NaN outputs.
2017-03-12 00:33:36 -08:00
0b05d364e4
Added GTR surface closure.
...
Not tested yet, just a straightforward conversion from the C++
Psychopath codebase. So there are probably bugs in it from the
conversion. But it compiles!
2017-03-07 08:31:57 -08:00
f4445417dc
Reorganizing the module tree.
...
Enough things had accumulated that it seemed to make sense to
group some stuff together. So here it is.
2017-02-14 00:14:08 -08:00
d504ca5e6a
Got DistantDiscLights working! Yay!
...
This involved fixing a bug in PsychoBlend. It was including
translation in how it was transforming the sun lights' direction
vector.
2017-02-13 00:03:34 -08:00
3cbb816d4b
Added DistantDiskLight (a.k.a. sun light) parsing and data structures.
...
Also created a proper World struct in the process, to store all
infinite-extent type stuff.
Note that I goofed and did a new rustfmt pass but forgot to
commit before making these changes, so there's a lot of
formatting changes in this too. *sigh*
2017-02-12 20:29:08 -08:00
746b3b0c1f
Replaced try!() macros with new ? operator.
2017-02-06 16:20:58 -08:00
0e8d708ff0
Made samples per bucket configurable.
2016-10-16 14:55:00 -07:00
bb4e57795f
Added support for OpenEXR output.
2016-10-09 20:34:58 -07:00
0d189f763d
Changed LightTree traversal to be a bit better in its choices.
...
After some experimentation, it's pretty clear that the LightTree
performs a lot better with a model of spherical _volume_ light
sources. This makes sense considering that generally they
represent a distribution of other lights in space.
This is a quick hack to make it behave a bit more like that. But
the long-term solution will be to adjust how
estimate_eval_over_solid_angle() of surface closures is implemented.
2016-08-21 15:27:15 -07:00
7f1ab59c5e
Implemented alternative SAH split that uses off-axis split planes.
...
It doesn't seem to actually perform better, generally, but I'm
leaving it in for future reference for other things.
2016-08-21 15:26:47 -07:00
c71b00ca31
LightTree now works with lights in hierarchical instancing.
2016-08-21 00:19:43 -07:00
97b5ef77f8
Code cosmetics: organize and alphabetize crate/mod/use statements.
2016-08-20 18:10:55 -07:00
f75c0c78fe
Moved TransformStack into its own module.
...
It's clear at this point that it will be useful outside of Tracer.
2016-08-20 17:43:26 -07:00
c997c55739
Separate timing of rendering and image writing in print out.
2016-08-20 16:25:12 -07:00
cf49cdbb02
Moved int hash functions into own file, and use in quick_select().
2016-08-14 16:03:17 -07:00
fbeadfce29
Use quick select for balanced BVH building.
...
This gets rid of a crate dependency, and also is faster.
2016-08-14 00:51:21 -07:00
024ca560af
Minor optimization in the transform stack.
...
Got rid of the scratch space by writing directly into the main
stack vector when merging transforms. Involves a small bit of
unsafe code.
2016-08-06 01:51:20 -07:00
08576b9934
Cleaned up console output.
2016-07-31 12:27:06 -07:00
c6f6266a38
Added timer output so you can see how long things take.
2016-07-31 12:05:28 -07:00
d12b76a933
Minor tweak to horizontal add and multiply for Float4.
...
In theory this should make it more pipeline-able.
2016-07-31 11:49:01 -07:00
b4939f88c8
Silence some warnings.
2016-07-31 11:19:30 -07:00
3e7b142cd8
Implemented light tree sampling, for better sampling of many lights.
2016-07-31 11:14:33 -07:00
1f94791b6b
Moved SAH splitting code into its own function.
...
This is in prep for the light tree, where we'll want to use that
code.
2016-07-30 21:20:01 -07:00
7ef5846c89
Got rid of indexing into Float4 and 3d vector elements.
...
This assures correct semantics for SIMD usage.
2016-07-30 19:36:54 -07:00
e4c94d0c58
Optional use of SIMD intrinsics, via feature in cargo.
...
This is disabled by default right now, because it only builds on
nightly.
2016-07-16 19:58:39 -07:00
4db96bc758
Use simple if-statement-based min/max functions in hot code.
...
Again, for whatever reason this is notably faster than the stdlib
min/max functions.
2016-07-16 17:55:18 -07:00
7401c6fb99
Use fast approximate exp functions for XYZ color curves.
...
This brings a significant performance improvement, and doesn't
appear to have any visual impact.
2016-07-16 17:53:50 -07:00
177bb95eff
Got rid of UnsafeCell in Tracer.
...
To do this, I moved the accel rays into an outer struct, which
then calls the main tracing code as an inner struct.
2016-07-16 16:26:34 -07:00
1571c1634e
Sped up min and max functions in Float4.
...
Turns out that the standard min/max functions were slow for
some reason, and simple if statements are much faster. This
simple change improves render times by over 30%. Crazy.
2016-07-16 15:56:14 -07:00
76f074f866
Added a TODO comment.
2016-07-16 15:10:07 -07:00
ec75f11206
Render buckets in hilbert curve order.
...
Provides a slight speed-up.
2016-07-16 14:53:03 -07:00
5bc97f69b8
Actually use the background color from the scene file.
2016-07-16 13:54:41 -07:00
8cd445813a
Removed some dead code.
2016-07-16 13:36:34 -07:00
9d47dfdeb2
Forgot one place where bounds need to be merged for motion blur.
2016-07-16 12:28:08 -07:00
2e3851837d
Implemented SAH splitting for the BVH. Also fixed crash bug.
...
The bug was in the previous commit, where I thought I was
preventing out-of-bounds access during traversal by limiting
the tree depth. While the idea was correct, I forgot that the
traversal stack needs _2_ extra slots on top of the tree depth,
not just 1. Fixed.
2016-07-16 03:21:57 -07:00
35ef58b09c
Special case BVH building when we get too close to max depth.
...
This avoids exceeding max BVH depth even in pathological cases.
Still need improve non-worst case building, but this at least
prevents crashes in worst case.
2016-07-16 00:05:23 -07:00
e2ef44f666
Finished implementing motion blur.
2016-07-10 19:20:37 -07:00
c2eb421fd8
Actually use seed in renders.
...
This is important for animations, where you don't want the same
noise pattern every frame.
2016-07-10 17:56:42 -07:00
e4f9a54e48
Leave console output on a newline.
2016-07-10 17:47:28 -07:00
c466e46138
Make BVH traversal stack bigger.
...
This is a stop-gap for a less-than ideal BVH construction. It
should be reduced again once the BVH building is made less stupid.
2016-07-10 17:45:10 -07:00
5b461220a5
PsychoBlend now exports meshes as meshes, and Psychopath renders to PNGs.
2016-07-10 17:42:28 -07:00
e8ee371423
Silence some warnings.
2016-07-10 16:46:14 -07:00
da46878243
Print render progress and auto-determine bucket size.
...
The bucket size is automatically determined based on a target
number of samples per bucket.
2016-07-10 16:40:30 -07:00
