If you don't see facial motion capture everywhere in movies and video games, there's probably a good reason for it. In order to train a capture system to match expressions, you frequently have to record the actor making all kinds of faces in different conditions... that's a lot of effort, especially for studios that don't have much time or money. Disney Research has a better way. It recently developed a face capture system that uses only a tiny number of recordings and synthetically generates the data you need to train the system. The solution is a uniformly-lit multi-camera rig (only one camera is active, the rest are for lighting) that creates an adjustable 3D model of the actor's face that you can use to create the info you need.