ColorAdjustmentNode: Correct lumaCoeffs.

[Mugen87]

Related issue: -

Description

This PR corrects the luma coefficients to the values previously used in WebGLRenderer.

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

Filesize dev	Filesize PR	Diff
683.5 kB (169.2 kB)	683.5 kB (169.2 kB)	+0 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)	460.7 kB (111.1 kB)	+0 B

[Mugen87]

@WestLangley @donmccurdy Looking at https://en.wikipedia.org/wiki/Rec._709#Luma_coefficients, it seems 0.2125, 0.7154, 0.0721 are actually the more recent luma coefficients.

Should we update the formula in common.js/ WebGLRenderer instead?

[WestLangley]

I don't think Luma is the correct term. Luma is computed from components in sRGB color space -- not linear sRGB color space. I'd suggest renaming to LuminanceCoeffs -- if you want to retain that constant. /ping @Mugen87

https://en.wikipedia.org/wiki/Relative_luminance

[donmccurdy]

I think it's more about consistency than correctness for our purposes – I've seen the 4-digit coefficients 0.2126, 0.7152, 0.0722 most often, that's closer to what we've historically used, Blender uses them, and that's also what the ASC CDL requires... I would be inclined not to update to the newer BT.709-1 coefficients until we observe other software relevant to our users doing the same.

Ideally (not necessarily in this PR, I could look at this later) the coefficients should be color space dependent, and could throw an error for anything but Linear-sRGB and sRGB.

const luminanceCoeffs = ColorManagement.getLuminanceCoefficients(
  ColorManagement.workingColorspace
);

My understanding would be that the coefficients are applicable to both sRGB and Linear-sRGB components, but produce different results: luma (for sRGB components) and relative luminance (for Linear-sRGB components)

[Mugen87]

I would be inclined not to update to the newer BT.709-1 coefficients until we observe other software relevant to our users doing the same.

Okay, in this case I merge so we are consistent with WebGLRenderer.

[WestLangley]

the coefficients should be color space dependent

Yes, that is why I included the comment in common.glsl.js:

float luminance( const in vec3 rgb ) {

	// assumes rgb is in linear color space with sRGB primaries and D65 white point

	const vec3 weights = vec3( 0.2126729, 0.7151522, 0.0721750 );

	return dot( weights, rgb );

}

I could look at this later

Great! These coefficients are simply the 2nd row in the conversion matrix from Linear-sRGB to XYZ color space, assuming a D65 white point.

[Mugen87]

BTW: There is a shader which is in used by the unreal bloom pass which relies on even different values:

three.js/examples/jsm/shaders/LuminosityHighPassShader.js

Lines 52 to 54 in b999622

	vec3 luma = vec3( 0.299, 0.587, 0.114 );
	float v = dot( texel.xyz, luma );

These seems to be Rec. 601 luma luma coefficients. Would it make sense to change the code to the following so 709 is used? Related https://en.wikipedia.org/wiki/Luma_(video)#Rec._601_luma_versus_Rec._709_luma_coefficients

const v = luminance( texel.xyz );

[donmccurdy]

Sounds good! Follow-up PR:

• ColorManagement: Add .getLuminanceCoefficients #28880