ColorManagement: Add .getLuminanceCoefficients

[donmccurdy]

Follow-up to #28861. Adds a method to ColorManagement, to produce luminance coefficients for a given color space. Currently supports sRGB, Linear sRGB, Display P3, and Linear Display P3. For sRGB and Linear sRGB I used the (more common?) 4-decimal precision, 0.2126, 0.7152, 0.0722, simply or consistency with OCIO, Blender, and ASC specifications. I expect the difference will be ~zero, but if we see E2E failures as a result, I can switch back. I believe it's fair to call these either "luma coefficients" or "luminance coefficients", I don't have a preference other than internal consistency.

Certain post-processing effects were using luminance coefficients from Rec. 601, which we do not otherwise support. I've corrected those coefficients.

One notable exception on ColorManagement's use, common.glsl.js contains hard-coded luminance coefficients. Ideally they'd be based on the working color space, but I'm unsure if it's a good idea to put a new global uniform into this shader chunk. Suggestions welcome. In terms of working toward support for wide-gamut PBR rendering, it might be more practical to focus on WebGPURenderer for now. Fixed. ✅

Finally, note that while the luminance coefficients are the same for Linear sRGB vs. sRGB, and for Linear Display P3 vs. Display P3, the dot product gives different results with different meanings given linear vs. non-linear components.

/cc @Mugen87 @WestLangley

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

Filesize dev	Filesize PR	Diff
683.5 kB (169.2 kB)	683.9 kB (169.3 kB)	+397 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

Filesize dev	Filesize PR	Diff
460.7 kB (111.1 kB)	461.1 kB (111.2 kB)	+397 B

[WestLangley]

I believe it's fair to call these either "luma coefficients" or "luminance coefficients", I don't have a preference other than internal consistency.

Can we please use luninanceCoefficients and stay away from luma? luninanceCoeffs is OK, too, if you want.

[Mugen87]

Ideally they'd be based on the working color space, but I'm unsure if it's a good idea to put a new global uniform into this shader chunk. Suggestions welcome.

I would say yes so ideally components like LuminosityHighPassShader could directly use the luminance() function from commons.glsl. That should make the code more readable.

When thinking about it, we can have a solution without a uniform at all if you update WebGLProgram instead. You can embed the coefficients directly in the fragment shader similar to the tone mapping and color space logic. Use getTexelEncodingFunction() as a reference.

[donmccurdy]

Can we please use luninanceCoefficients and stay away from luma? luninanceCoeffs is OK, too, if you want.

Ok, works for me. 👍🏻

You can embed the coefficients directly in the fragment shader similar to the tone mapping and color space logic.

Will look into this – thanks!

Moving this PR to 'draft' as I address feedback.

[donmccurdy]

I believe all feedback has been addressed, moving the luminance function to WebGLProgram works well. ✅