c48839d5f3
An even better Owen scrambling hash.
2021-01-23 00:55:54 +09:00
c32281b04a
Significantly better Owen scrambling hash.
2021-01-13 22:25:16 +09:00
105d6e52b5
Add basic benchmark to Sobol sampler.
2021-01-10 00:20:22 +09:00
285f714d02
Switch to a significantly better LK hash.
...
With just one round, this hash is better than the previous one with
two rounds. And at 2-3 rounds it seems to be just as good quality
as a slow, per-bit hashing approach, which I've been using as
ground-truth for testing.
2021-01-04 15:16:04 +09:00
776a2c1456
Add PR template explaining that contributions aren't currently accepted.
...
Ideally I would just disable PRs on the repo, but github doesn't
currently allow that.
2020-12-31 13:00:46 +09:00
16e5f539ea
Update to latest openexr-rs.
2020-12-31 12:25:22 +09:00
d19305d23b
Update sobol sampler to properly credit Brent Burley.
...
I was just waiting for his paper to get published, which happened today!
2020-12-30 21:44:25 +09:00
3d85ce8005
Consolidate all of the compact value storage formats into one crate.
2020-12-30 20:41:56 +09:00
6586b38dc9
Use an external crate for fast approximate math functions.
2020-12-26 10:03:35 +09:00
4cfc051260
Change overall project license to GPLv3.
2020-12-26 09:20:55 +09:00
646e6a4e18
More readme updates.
2020-12-21 18:55:00 +09:00
73cf201bf6
Update readme.
2020-12-21 18:42:23 +09:00
6d6904a615
FLuv32: increase dynamic range, and decrease precision.
...
This still exceeds the precision of LogLuv, but lets us match its
dynamic range.
2020-09-22 11:06:40 +09:00
9cf5ebdf91
Cleaning up the code in fluv32 a bit.
2020-09-21 11:54:14 +09:00
066105b20a
Fluv32: slightly tweak the u/v scaling constants.
...
This allows perfect representation of E (equal energy spectrum).
It's not important from a perceptual standpoint, but it provides
a simple way for Psychopath to represent E when needed for other
purposes.
2020-09-21 09:45:47 +09:00
05f9621ac5
Added a FloatLuv decode function to decode to Yuv instead of XYZ.
...
This is useful because it's super fast, and chromaticity lookups
are typical for spectral upsampling anyway, so this will likely
enable cutting out a bunch of unecessary intermediate calculations.
2020-09-20 15:16:20 +09:00
0ba1acc42d
Add oct32 encode_precise() function.
...
This is significantly slower than `encode()`, but a little more precise.
2020-09-20 14:36:50 +09:00
8c738b2f39
Use FloatLuv32 in Psychopath for encoding XYZ colors.
2020-09-20 11:04:37 +09:00
161e47fc44
Clean up oct32norm code a bit.
2020-09-20 10:37:26 +09:00
3eff608493
More FloatLuv32 optimizations, and general code cleanup.
...
This gives another little speed boost to decoding, but gives a
massive (over 3x) speed boost to encoding.
2020-09-20 10:07:02 +09:00
f4ef11f9f3
Fix some names in fluv32 error messages.
2020-09-20 03:02:56 +09:00
8dee53d1fc
Additional optimization to fluv32 decoding.
...
Tiny change but with a nice speed bump.
2020-09-20 02:35:37 +09:00
f20567247d
Go all in with the fluv32 naming.
2020-09-20 02:02:35 +09:00
cda9156af8
Optimize FloatLuv decoding.
...
Speed ups of over 20%.
2020-09-20 01:43:36 +09:00
03bedcb594
Cleanup, tweaks, and better documentation for the 32-bit Luv format.
2020-09-19 23:57:59 +09:00
485da9f918
Add a new 32-bit Luv color format.
...
It's based on the SGI LogLuv format, but using a floating point
instead of log encoding for the luminance.
2020-09-19 19:51:40 +09:00
66e9abe66e
Make signed48 trifloat encoding also a byte array.
...
Same reason as for the unsigned40 encoding in an earlier commit.
2020-09-19 12:22:51 +09:00
49c97bf0fe
Cleaned up the signed48 trifloat code.
2020-09-19 08:55:35 +09:00
fd98b33333
Make unsigned40 trifloat encoding a byte array.
...
The whole point of these formats is to compress down to less
space, so let's not leave actually putting it in the space-saving
form on the client code.
2020-09-19 08:28:59 +09:00
9fb3c95171
Added an unsigned 40-bit trifloat format.
...
It is identical to the 32-bit format, except with more precision
and range due to using more bits. This format should comfortably
store any color information with precision easily exceeding the
limits of human vision.
2020-09-18 22:12:12 +09:00
96b8dd84b9
Cleaned up the u32 trifloat implementation.
...
This also makes encoding faster. However, it no longer does
rounding to the nearest precision when encoding, and insead does
flooring. This seems like a reasonable tradeoff: if you want more
precision... you should use a format with more precision.
2020-09-18 21:04:16 +09:00
f13ffac7bd
Removed the experimental luma-chroma color format.
...
It was a worthwhile experiment, but for it to really work it needs
a really proper luma-chroma separation, which is both slower than
I really want, and requires knowing the colorspace being used.
I might make another go at this based on the TIFF LogLUV color
format, requiring XYZ as input.
2020-09-18 17:57:13 +09:00
c1f516c2b6
Use a better chroma formula for the RGB32 format.
...
This makes much better use of the bit space.
2020-09-13 11:22:48 +09:00
bd6cf359b4
Some code clean-up in the RGB32 encoding/decoding code.
2020-09-13 06:44:42 +09:00
d6ab9d06be
More work on the packed HDR RGB 32-bit format.
...
Switched to a different chroma encoding, which is notably faster
and never produces negative values when decoded.
2020-09-11 21:57:43 +09:00
339568ec0c
Remove stale comments.
2020-09-10 22:42:20 +09:00
7066c38189
Implement an experimental packed HDR RGB 32-bit storage format.
2020-09-10 22:36:20 +09:00
99b6ddfa54
Run clippy on Sobol sampler, and fix/silence warnings.
2020-05-01 16:07:19 +09:00
78acaa7b63
Make Sobol SIMD code work on all x86-64 platforms.
...
Before this it needed SSE 4.1, which is not strictly present on
all x86-64 platforms. This will still compile the faster path if
SSE 4.1 is available, but has an alternate path as well for all
x86-64 platforms.
2020-05-01 15:32:18 +09:00
fd75a72655
More code tidying of the Sobol sampler.
...
In particular, clearer and more concise documentation.
2020-05-01 10:42:02 +09:00
45241784fb
Code tidying on the Sobol sampler.
...
Also swapped the sample index and dimension paramater in the function
signature. This feels more intuitive.
2020-04-30 22:46:47 +09:00
1f75e7854e
Properly hash all four scramble values in the 4d Sobol sampler.
2020-04-25 18:12:35 +09:00
72adbedbb4
Accelerate the Sobol sampler with SIMD on x86_64.
2020-04-24 23:32:43 +09:00
0dfe916523
Preparing for SIMD accelerated Sobol sampling.
...
This implements the 4-wide API, and moves the renderer over to it.
But the actual implementation is still scalar.
2020-04-24 21:05:29 +09:00
78b5cf4c53
Faster lk hash for the sobol sampler.
...
This gets about the same quality as the previous hash, but in
much fewer operations.
2020-04-24 10:47:53 +09:00
2c68e18bd6
Use golden ratio sampling for wavelength again.
...
Now that the Sobol sequence can be decorrelated, this doesn't
introduce sampling correlation problems anymore, so it's mostly
just a win.
2020-04-23 14:46:21 +09:00
3ffaf7bee5
Forgot to use a wrapping multiply instead of a straight multiply.
2020-04-23 13:52:04 +09:00
29fcc69ae1
Move seed hashing into the lk_scramble function.
2020-04-23 13:36:04 +09:00
c0a8c955c4
Make LightPath::new() harder to mis-use.
2020-04-23 08:32:44 +09:00
b776bf56b8
Improved lk scrambling function.
...
This actually gets very close to the behavior of a full per-bit
hash, except that you still need a fairly random seed.
2020-04-23 08:31:03 +09:00
