Rapid generation of a 3D face from a human subject using the Intel RealSense sensor.
The face consists of 8 blendshapes that were captured as 3D scans then processed in our near-automatic pipeline and transformed into blendshapes in about 20 minutes. The rendering is done by SmartBody which simultaneously blends geometry and textures in order to obtain the high resolution face details captured by the scans.
The face is tracked and animated in real time. The video shows the digital face being puppeteered by the original subject, and then by a different person.