about two weeks ago
now
of cource
link to model 1
https://drive.google.com/file/d/1ZatGJfzXg8_fKB-R8JbVXwUFdYxqo8Va/view?usp=sharing
I have another model with slightly different export settings and it has other problems that weren’t there before either
link to model 2
https://drive.google.com/file/d/1ucxI1KDyDsEUS67UgggWgg2E9dbSDaJt/view?usp=sharing
I also want to add that everything is fine in other viewers
Will you please grant access for me to download the files? Thanks! Request sent.
sorry forgot to open it, i made available to anyone with the link 
1 Like
Yes, i know what before re-export blender fix model, but I wonder why the old model broke, what has changed in the engine…
Fair point. It works with 5.0.0-beta.7 but doesn’t work with 5.0.0-beta.8 or 9
Thanks for reporting!
Hey @Deltakosh! Bug confirmed.
beta.7
beta.8
Beta 9 has the same issue. Tested.
The main problem is that this asset has invalid skin.skeleton values which are supposed to be the common root node of the skeleton joints. You can check using the glTF Validator. After beta.8, the code is now using the new way of handling glTF skins which reads this value. It wasn’t being used in prior versions. This property is optional in the glTF. If it’s optional, the loader will calculate the nearest common ancestor for this value as per spec.
We have a few options:
- Keep it as is. Some models will break because they don’t have valid
skin.skeletonvalues. - Change the loader to always compute the nearest common ancestor. Performance will suffer slightly as the loader will have to compute the nearest common ancestor.
- Add a flag to force compute the nearest common ancestor.
I’m leaning towards 3 but it will be off by default (for perf reasons). Comments?
3 Likes
Agree on 3!
3 Likes
Flag added in this PR as well as a couple of other fixes: Various glTF fixes by bghgary · Pull Request #11996 · BabylonJS/Babylon.js (github.com)
3 Likes
Thank you all!!!
I’ll try to write a script to fix this problem with skin.skeleton…
2 Likes