Local Facial Layered Rig

by user9 posted 05-02-2018 | 3 comments

I am currently develop my own facial rigger scripts. Which one do you prefer when it comes to setting up a facial rig?

Do you parent everything under the head joint then skin the joint to the main mesh or create a local rig then transfer the deformation using blendshape?

What is the pros and cons?

Thanks

by user3 posted 06-02-2018

I prefer to stay away from local rigs as much as possible. I find them not very nice to work with (as you have multiple setups) and as in the article hazmondo mentioned they can lead to some slow down. Of course, if you follow the technique there you can counter that.

That being said, I try to do as much as I can in the world/global rig and I don't remember the last time I needed to do anything more than add some blendshapes front of chain. The evaluation works great that way, as there aren't any complex local rigs that need to be evaluated first.

by user129 posted 05-02-2018

I think the best thing to do is layer your deformers, you can even do this with skin clusters. You should definitely read this article: https://medium.com/@charles_76959/deformation-layering-in-mayas-parallel-gpu-world-15c2e3d66d82

by user128 posted 05-02-2018

I don't always build blendshapes but when I do I'm doing it at front-of-chain...

So everything is done in place with BS and skinning comes after that in history. Faster because of fewer deform nodes, less intermediate shapes in memory.