Learning Locally Editable Virtual Humans

Hsuan-I Ho, Lixin Xue, Jie Song +1

In this paper, we propose a novel hybrid representation and end-to-end trainable network architecture to model fully editable and customizable neural avatars. At the core of our wo…