Doubts about control layers

Hey guys, i am interested in learn more about tweaks controlled by bind pre matrix and stuffs like that, because i only feel safe creating tweaks when using the default setup of a geometry with blendshape on the original mesh, and offset joints to keep them zeroed-out, but i didn’t find any resources that shows how works behind the scenes about that stuffs, i have seen some rigs that all the tweaks joints move along the geometry and that works pretty good for blendshapes and skin, is that useful for animators, or they don’t really care about? And how can i learn or search about that?

They’re definitely useful for animators, especially at the finalling stage. However as you’ve identified, they’re not the easiest things to do well.

It’s good to have a general solution to “things following along with a skinned / deforming mesh” - this is usually done by having a follicle driving a parent group, with the control transform as its child. Take the local matrix of the control transform and that’s your input to your joints or your softmod.

However, this system requires 2 layers of skincluster, with the intermediate driving the follicles, and the final result coming from the surface controls - this is slow. If that doesn’t matter, you’re fine - if it does, it gets a bit more tricky to avoid pulling the mesh off gpu, replicating the skin deformation with parent constraints to the control groups, etc.

It’s likely that the new deformation tags and live weights can make this much easier, but sadly I haven’t kept up with it too well - see if any of those tools fit your needs first.

1 Like

Thanks for your answer Ed, i do know how to move my controls and joints along the mesh using that method of follicle, i think i need to get better in math before proceed to find a better way to do that, because as you said, when we use a follicle for each of’em being moved by the mesh, things can be really heavy

Ugh, I HATE that people are still using follicles. If you can, please stop.

If you’re in 2020 or later, definitely use the proximity pin: It’s one node that can drive multiple transforms, and because of that, it can be way more efficient.