We don’t see much of the facial performance capture in movies as it requires a lot of effort. You have to record an actor making all kinds of facial expressions in different conditions frequently for training a capture system to match expressions, which is kind of difficult for a studio that doesn’t have much money. Disney Research has found a solution for this. It has developed a facial capture system that would dramatically reduce effort and time to capture facial expressions.
Facial Capture System
With the facial capture system, researchers could use a tiny number of recordings rather than comprehensively recording the actor showing different facial expressions under many combinations of lighting and camera viewpoints. Then they synthetically generate the data required to train the system.
They used a multi-camera capture setup and uniformly-lit rig, which generates an adjustable 3D model of the actor’s face. You can use that model to create the info you need.
The gains are actually pretty big. The amount of training data you need to collect is ten to hundreds of times smaller, without compromising facial capture accuracy. It means studios now have more time to focus on recording scenes, rather than working hard for preparation.
This facial capture system has widespread, as well as complicated uses with numerous actors. It’s not sure whether Disney will release its technology to some other companies or not. Even if they don’t, you may at least see some Disney movies and games in the future, which will come with an extra dash of realism.