Face Transfer with Multilinear Models
This research extracts a 3D model of a person's face from a video. The person's facial expressions and mouth movements can then be transferred to the 3D face model of another person. In this video, the two input videos are shown in the top corners. The facial movements of the person on the left are transferred to the person on the right, creating the main video shown. On each side of the output video are two perspectives of the new 3D face model. Published in SIGGRAPH 2005.
For more info (PDF):