Figure 1: Given a monocular portrait video sequence of a person, we reconstruct a dynamic neural radiance field representing a 4D facial avatar, which allows us to synthesize novel head poses as well as changes in facial expressions.
translated by 谷歌翻译