mirrors/blender
1
0
Fork 0
mirror of https://projects.blender.org/blender/blender.git synced 2025-01-22 07:22:12 -05:00
Code Issues Projects Releases Packages Wiki Activity

Color management: optimize OCIO predivide processing

Browse source 
cpuProcessorApply_predivide was doing, for each pixel:
- Un-premultiply pixel to straight alpha
- Call OCIO processor on that one pixel
- Premultiply pixel back

This is not great due to just function call overhead, and probably
prevents whatever "batch processing SIMD optimizations" that OCIO
migth have.

Instead, do this:
- Un-premultiply whole input image,
- Call OCIO on the whole image to do whatever it does,
- Premultiply whole image back.

Doing cpuProcessorApply_predivide on a 4K resolution, float4 image
on Ryzen 5950X (Win10/VS2022) on one thread: 128ms -> 69ms

Pull Request: https://projects.blender.org/blender/blender/pulls/127307
This commit is contained in:
Aras Pranckevicius 2024-09-09 13:14:55 +02:00 committed by Aras Pranckevicius
parent a904db3ee7
commit d0ea251c33
1 changed files with 26 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
intern/opencolorio/ocio_impl.cc
View file

@ -578,22 +578,36 @@ void OCIOImpl::cpuProcessorApply_predivide(OCIO_ConstCPUProcessorRcPtr *cpu_proc
int channels = img->getNumChannels();
if (channels == 4) {
/* Convert from premultiplied alpha to straight alpha. */
assert(img->isFloat());
float *pixels = (float *)img->getData();
size_t width = img->getWidth();
size_t height = img->getHeight();
for (int y = 0; y < height; y++) {
for (int x = 0; x < width; x++) {
float *pixel = pixels + 4 * (y * width + x);
cpuProcessorApplyRGBA_predivide(cpu_processor, pixel);
float *pixel = (float *)img->getData();
size_t pixel_count = img->getWidth() * img->getHeight();
for (size_t i = 0; i < pixel_count; i++, pixel += 4) {
float alpha = pixel[3];
if (alpha != 0.0f && alpha != 1.0f) {
float inv_alpha = 1.0f / alpha;
pixel[0] *= inv_alpha;
pixel[1] *= inv_alpha;
pixel[2] *= inv_alpha;
}
}
}
else {
(*(ConstCPUProcessorRcPtr *)cpu_processor)->apply(*img);
(*(ConstCPUProcessorRcPtr *)cpu_processor)->apply(*img);
if (channels == 4) {
/* Back to premultiplied alpha. */
assert(img->isFloat());
float *pixel = (float *)img->getData();
size_t pixel_count = img->getWidth() * img->getHeight();
for (size_t i = 0; i < pixel_count; i++, pixel += 4) {
float alpha = pixel[3];
if (alpha != 0.0f && alpha != 1.0f) {
pixel[0] *= alpha;
pixel[1] *= alpha;
pixel[2] *= alpha;
}
}
}
}
catch (Exception &exception) {