Face2Face: Real-time Face Capture and Reenactment of RGB Videos (CVPR 2016 Oral)

This is quite impressive in that you don’t need any markers or depth information of the source actor’s face.

How difficult would it be to do this if you were using a Kinect or PrimeSense camera and create a 3D mask of the target actor manually in order to perform realtime face manipulation?

Leave a Reply