MeshPhongMaterial: Assign color space of default specular value

[donmccurdy]

Assigns a color space for the default specular value, to ensure that it behaves the same way with or without THREE.ColorManagement. Without color management, all inputs are assumed to be Linear-sRGB. With color management, hexadecimal colors are considered sRGB, so we need to specify exceptions.

Related issue:

• Roadmap for a color-managed workflow in three.js #23614

[donmccurdy]

I missed this when updating the example earlier in 3b5cbe0. Further changes to lighting would restore the original appearance too, but it was much easier just to keep the specular value, and has little bearing on the example.

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

Filesize dev	Filesize PR	Diff
633.9 kB (157.1 kB)	633.9 kB (157.1 kB)	+24 B

🌳 Bundle size after tree-shaking

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

Filesize dev	Filesize PR	Diff
425.1 kB (103 kB)	425.1 kB (103.1 kB)	+11 B

[donmccurdy]

Hm, let me take a closer look at this. I'm not sure why this PR breaks the webgl_camera_arrays screenshot while enabling color management in the first place (13397a8) did not.

[donmccurdy]

It seems best to skip this change. I'm still a bit confused about why, but perhaps we should think of Phong's .specular property as more like a factor than a color, like a three-channel version of MeshPhysicalMaterial's .specularIntensity... In any case, adding this Linear-to-sRGB conversion is increasing the default specular highlight in a way we probably do not want.

[WestLangley]

@donmccurdy

material.color is the diffuse reflectance of the material.

material.specular is the specular reflectance of the material.

They are both unit-less, and they are both colors.

I don't think it would make sense from a physical modeling standpoint for the two colors to be associated with different color spaces -- or for one to be a color, and the other not.

[donmccurdy]

@WestLangley I agree that we cannot consider the two to be associated with different color spaces in the renderer. I believe we are agreed that this PR can remain closed, then, even if my speculative reasons in #25872 (comment) are incorrect?

[WestLangley]

@donmccurdy Yes, I think this PR should remain closed.