Face2Face is a software project by a team at Stanford University led by assistant professor Matthias Nießner that allows users to control the faces of people in YouTube videos in real time by moving their own face in front of a standard webcam.
The new system appears to be an upgrade from a similar project by Nießner that used Microsoft Kinect cameras to measure not only RGB data, but also depth. The Face2Face software produces better results while only using an ordinary webcam. The system also improved the rendering of mouth interiors for improved realism in the final result.
For comparison, this is the team’s video about their previous project.
image via Matthias Nießner