Text-based Editing of Talking-head Video
Supplemental materials
[back to index]
Comparison: using less data
We show results using visemes from 50, 10 and 5 percent of the full input video.