Real-Time Facial Expression Transfer with Single Video Camera.

Abstract

Facial expression transfer is currently an active research field. However, 2D image wrapping based methods suffer from depth ambiguity and specific hardware is required for depth-based methods to work. We present a novel markerless, real time online facial transfer method that requires only a single video camera. Our method adapts a model to user specific facial data, computes expression variances in real time and rapidly transfers them to another target. Our method can be applied to videos without prior camera calibration and focal adjustment. It enables realistic online facial expression editing and performance transferring in many scenarios, such as: video conference; news broadcasting; lip-syncing for song performances; etc. With a low computational demand and hardware requirement, our method tracks a single user at an average of 38 fps. Our tracking method runs smoothly in web browsers despite their slower execution speed.