r152 multiple UV mapping bug

[RemusMar]

Description

@mrdoob @donmccurdy @Mugen87
Guys, I'm not a big fan of the r152 changes (is never a good idea to change the default values for "something"), but let's ignore that.
There is a big issue in r152 (and maybe r151 as well) related to a material with multiple UV Mappings.
In r152 is used the same UV mapping for all the texture layers.

Screenshots:
r150 (good):

r152 (bad):

Reproduction steps

r150 = good
r152 = bad

Code

// I'm sure you are aware of the code changes

Live example

r150 (good):
https://necromanthus.com/Test/html5/SMC.html

Screenshots

No response

Version

152

Device

Desktop, Mobile

Browser

Chrome, Firefox, Safari, Edge

OS

Windows

[mrdoob]

It's probably this change: #25943

[RemusMar]

It's probably this change: #25943

The GLTF ids are not correct anyway.
If the first one is "uv", the second one cannot be "uv1".

[Mugen87]

Unfortunately, the SEA3D library you are using is not compatible to r152. The respective loader produces a file with uv and uv2 which means it does not honor the latest uv naming conventions. As long as the SEA3D loader isn't updated, you have to change the uv attributes by yourself.

Regarding #25943 (comment), I have to admit the situation is not ideal that we can't provide any sort of backwards compatibility. But as mentioned in the comment, the potential side effects of the suggested fallback are not clear to me, too.

@RemusMar Can you please test if renaming the uv2 attribute to uv1 fixes the issue. I know you are not happy with that but it's important for us to know whether this fixes the issue or not.

[donmccurdy]

The respective loader produces a file with uv and uv2...

this is technically still ok, right? as long as texture.channel=2 is specified (or can be inferred?).

[RemusMar]

Can you please test if renaming the uv2 attribute to uv1 fixes the issue.

Michael, that's the 1st thing I did after Ricardo's response.
The same wrong result in r152.
And the situation is even worse:
I get the same wrong result in r152 even with the GLTF version (here is the good 150 one):
https://necromanthus.com/Test/html5/SMC_gltf.html

[Mugen87]

as long as texture.channel=2 is specified (or can be inferred?).

I forgot about channels.

Can we first clarify if the app already breaks with r151?

[bhouston]

I am running into glTF files exported from blender 3.5.1 (the latest) are no longer loaded in Three.js after updating to the latest Three.js. I do not know the exact issue, it may be some interaction as when I try to simplify the glTF the issue goes away.

[bhouston]

I can confirm it loads properly with 0.150 and 0.151, but not with 0.152. So I will just stick with the older version for now.

[mrdoob]

@bhouston can you share one of those files?

[mrdoob]

@donmccurdy

this is technically still ok, right? as long as texture.channel=2 is specified (or can be inferred?).

That's right yeah.

@sunag would it be possible to update SEA3DLoader with support for texture.channel?

[sunag]

@RemusMar I believe that renaming the string uv2 to uv1 in this line should solve it:

https://github.com/sunag/sea3d/blob/0533fb0e1192915a50f2d0c38d7570d64dc37be2/Source/Three.JS/SEA3DLoader.js#L1899

[sunag]

And add mat.lightMap.channel = 1 or mat.aoMap.channel = 1 here.

https://github.com/sunag/sea3d/blob/0533fb0e1192915a50f2d0c38d7570d64dc37be2/Source/Three.JS/SEA3DLoader.js#L2758

[RemusMar]

I believe that renaming the string uv2 to uv1 in this line should solve it

I did that in the first place and it didn't change anything.

And add mat.lightMap.channel = 1 or mat.aoMap.channel = 1 here

Yes, that did the trick.
Well done!

But guys, we still have a big problem.
This GLB (GLTF) sample still gives wrong results in r152:
https://necromanthus.com/Test/html5/SMC_gltf.html
Did you make any change in the GLTF loader in r151 or r152 (related to multiple UV mapping)?

[mrdoob]

Yes, make sure you're using the latest version of the library and also the latest version of GLTFLoader.

[RemusMar]

Yes, make sure you're using the latest version of the library and also the latest version of GLTFLoader.

Everything updated to r152bis
All good now.
152 SEA3D: https://necromanthus.com/Test/html5/SMC.html
152 GLTF: https://necromanthus.com/Test/html5/SMC_gltf.html