Odd Texture Issue in r72

[titansoftime]

I'm not really sure what's going on here, but you can see in this fiddle:

http://jsfiddle.net/titansoftime/pawthzr8/

that the texture is a little distorted. This is the only one (that I know of) out of several hundred of our models that has been affected.

This started in r72, it was fine previously.

You can get the model here: http://www.titansoftime.com/utils.php?task=getModel&id=71
And the texture: http://www.titansoftime.com/textures/71_1024.dds

[mrdoob]

Is it a problem with the model or the texture? Have you tried with a non-dds texture instead?

[titansoftime]

I will give it a shot when I get home from work. Interesting to test, though in production not using a compressed texture is not really an option.

[mrdoob]

Something I've just tried... I've tried loading the model in the editor and use a random texture in the map slot. Then, in the project panel, I switched the renderer to CanvasRenderer.

CanvasRenderer doesn't do the new THREE.BufferGeometry().fromGeometry() so I was double checking if there was a bug there... Seems like that part is fine.

Did you updated the DDSLoader?

[titansoftime]

I am using this loader in the fiddle: https://rawgit.com/mrdoob/three.js/dev/examples/js/loaders/DDSLoader.js

[mrdoob]

Ah, true.

[titansoftime]

This issue does not exist with skinning disabled.

http://jsfiddle.net/titansoftime/53rqpdfv/

Just like #7111. Hmm...

[mrdoob]

Updated links to use r72 instead of r73dev:

http://jsfiddle.net/pawthzr8/2/
http://jsfiddle.net/53rqpdfv/2/

[titansoftime]

@mrdoob Do you have any idea what caused this bug?

All I know (thanks to the observation of @arose) is that this issue only affects a material with skinning enabled. Please whoever broke this, take a look at it. I don't even know where to begin, not to mention r72 has a billion commits.

Can we at least get a bug label on this?

[mrdoob]

@titansoftime have you tried updating the code to the new animation system?

[titansoftime]

I have and unfortunately the issue remains (74dev). It's been a little while so I'd like to reiterate that setting material.skinning to false "fixes" this and #7111.

[mrdoob]

Do you mind updating the jsfiddles to r74dev? 😇

[titansoftime]

Here you are: http://jsfiddle.net/eo0hpjk5/

Old expected behaviour (r71): http://jsfiddle.net/fnanr01f/

I updated the fiddles for #7111 as well.

[WestLangley]

Can you reduce your model to just the flag? Or better yet, reproduce the issue with a single quad?

[titansoftime]

Can you reduce your model to just the flag? Or better yet, reproduce the issue with a single quad?

Unfortunately no. I am not an animator or a modeler. I paid for all my models / animations.

[mrdoob]

If you don't convert to BufferGeometry it seems to render just fine:

http://jsfiddle.net/eo0hpjk5/1/

[mrdoob]

My bet is that SkinnedMesh.normalizeSkinWeights() expects skinWeights to be normalized when using BufferGeometry.

[WestLangley]

Looks much better if weights are normalized prior to converting to BufferGeometry.

r.74dev: http://jsfiddle.net/eo0hpjk5/2/
r.71: http://jsfiddle.net/fnanr01f/

However, now notice that there are slight differences in the animations between versions.

[titansoftime]

Closing due to the cause of issue being on track to be fixed in #7111.