MeshPhysicalMaterial: Support iridescence / thin-film materials #23869
Conversation
|
Preview Link:
|
donmccurdy
changed the title
donmccurdy
changed the title
|
Ops... the builds are already conflicting. Could you remove them from the PR? |
@mrdoob Yeah, didn't think of this conflict. Removed the build files from the PR. ;) |
|
Thanks @PascalSchoen! Looks good to me, though I'd like to know what @WestLangley thinks. I'd still like to refactor multiscattering, since I don't think it's right that we make up the energy difference with the diffuse lobe. Would be nice to simplify the code there too, but that can happen separately from this PR. |
|
@PascalSchoen Lookin good! Is there any sample model we can use in the examples that uses this? |
|
Just a heads-up, Khronos PBR TSG may need to change The following models have been updated to the new casing:
|
|
IridescenceMetallicSpheres looks cool but looks like testing purpose too much? Personally I'm ok with it but if we don't really prefer testing purpose models, probably Suzanne may be the only option in them. |
| @@ -40,7 +40,7 @@ class MeshPhysicalMaterial extends MeshStandardMaterial { | |||
| } ); | |||
|
|
|||
| this.iridescenceMap = null; | |||
| this.iridescenceIOR = 1.3; | |||
There was a problem hiding this comment.
@PascalSchoen sorry for the churn here, but glTF's camelCase convention for acronyms like XYZ (fooBarXyz) is different from what we follow in three.js (fooBarXYZ). I think for the public API of MeshPhysicalMaterial, this should remain iridescenceIOR. Feel free to use whichever you prefer for internal code in the rest of the PR. 🙏
There was a problem hiding this comment.
@donmccurdy I reverted the change and only kept it for the glTF extension parts. 🙂
There was a problem hiding this comment.
@donmccurdy You prefer using .ior and .iridescenceIOR?
Note, we also have the function applyIorToRoughness().
There was a problem hiding this comment.
I think so, yes — for example we use:
- attribute.getXYZ not .getXyz
- attribute.x not attribute.X
- GLTFLoader not GltfLoader
I'm comfortable with either convention but this seems to be the norm in three.js today, and (I think?) more common/idiomatic in JavaScript generally.
There was a problem hiding this comment.
OK, I don't have a strong feeling on this one. Maybe applyIORToRoughness()?
There is also the option of .refractiveIndex instead of .ior.
There was a problem hiding this comment.
Slight preference for applyIORToRoughness, yes - but not as worried about internal methods.
Somehow .ior does not bother me, feels similar to vec.xyz in GLSL. 🤷♂️
|
Thanks! |
|
Depending on the model the shader could crash because of |
@takahirox Khronos has a new photo-realistic model now from Wayfair: IridescenceLamp. |
Thanks. That looks much nicer. If we add an example for iridescence / thin-film materials we can use it. |
|
Yes, I noticed. I was going to do an example today 😊
|
…ob#23869) * Add iridescence parameters to Physical Material * Add iridescence fragment shader code * Iridescence shader integration and glTF loading * Update iridescence default values * Add KHR_materials_iridescence to supported extensions in GLTFLoader docu * Enable iridescence in Editor * Update build results * Remove build files from PR * Honor iridescence parameters in program cache * Use range for iridescence thin-film thickness * Fixed linting errors * Fix always-true conditional * Add iridescence to glTF export * Instantiate iridescenceThicknessRange in GLTFLoader if undefined * Make iridescence Fresnel consistent for IBL * Rename iridescenceIOR to iridescenceIor * Rename iridescenceIor back to iridescenceIOR except for glTF extension
Description
Implementation of the KHR_materials_iridescence glTF extension.