Dedicated test for correct pixel transfer functions

[lexaknyazev]

As of today, only four glTF color textures use sRGB encoding:

• emissiveTexture
• pbrMetallicRoughness.baseColorTexture
• KHR_materials_pbrSpecularGlossiness.specularGlossinessTexture
• KHR_materials_sheen.sheenColorTexture

All other textures are sampled linearly. We need dedicated test models that would highlight potential implementation errors in that area.

It's important to note, that working with linear 8-bit values in DCC tools could be quite challenging as some editors may misrepresent such content on screen and show misleading color-picker values.

[donmccurdy]

If we create test assets for these cases, I would prefer to use KHR_materials_specular (KhronosGroup/glTF#1719) over KHR_materials_pbrSpecularGlossiness. Fine to have both, too, but the emphasis will be more on the first moving forward.

[emackey]

Is this resolved now?

[lexaknyazev]

I'd prefer to keep this open until we have similar tests for:

• KHR_materials_pbrSpecularGlossiness.diffuseTexture
• KHR_materials_pbrSpecularGlossiness.specularGlossinessTexture
• KHR_materials_sheen.sheenColorTexture
• KHR_materials_specular.specularColorTexture

Although the SpecGloss is deprecated, Sample-Viewer had a bug with it and the sheen color: KhronosGroup/glTF-Sample-Viewer#295.

[DRx3D]

This issue has been opened for almost 3 years. Since the repo has moved and there is a request to keep it open until certain things are done, should this be transferred to Sample-Assets? If there is no response, the issue will be transferred to Sample Assets by 17 Nov.