Best eyelid setup in Maya


Im working on a new facial rig system.
I was looking some reel and I found this amazing reel from Deng:

Anybody knows what setup is using for the eyelids?


Based on the gray shaded model view, it looks like he referenced an e-book called the Art of Moving Points by Brian Tindall. It’s an iBook. The resulting shapes also appear to follow the instruction of that book.


Awesome book.
But Im looking how to setup in Maya the behavior on the lids at the beginning.
I think he is using blendshapes


Judging from the Po Dad example, I’m assuming he was a part of the Oriental Dreamworks team on KFP3. That film was localized for China.

You can take a look at Stephan Candell’s stuff (former Lead Face Rigger at DW) on Vimeo where he talks about similar concepts. Joints riding on Nurbs surfaces in UV space driving curves driving joint systems. These can then talk to each other and generate these interactions. Pretty tricky with lots of systems working together to each other to get the shapes you are seeing, but possible.


Thanks Weider!!
Which Facial Rig approach do you think is the best for the animator?
Because the rig looks amazing but animator-wise maybe is too complicated.


If you have the R+D time to develop this, I think it pays off in the long run for animators. To be able to predictably close lids/lips and offset as needed is super handy. As far an an animator interface, what Deng is doing seems straightforward. A layer of macro controls

Corner -> L -> M -> R -> Corner

that’s fed into a seal/sharpen/soften control sliders followed by micro offset controls to further tweak if necessary and if topology allows. Blendshapes to further art direct lays on top as well. I can tell you that the above models are likely highly iterated and scrutinized to allow just the right amount of joint influences to get these shapes.

Again, from an animator perspective this will feel pretty simple, but powerful. Under the hood? So many rigging gnomes running around making it look great :grin: Up to budgets/R+D time to get this up and running so I’d base the complexity on the project scope.


Im working on my demo reel, I really want that behavior is there a technical paper or a tutorial around.


self collision example