公开(公告)号：US20240212252A1

公开(公告)日：2024-06-27

申请号：US18597750

申请日：2024-03-06

申请人： Tencent Technology (Shenzhen) Company Limited

发明人： Yang WU ,  Pengfei HU ,  Xiaojuan QI ,  Xiuzhe WU ,  Ying SHAN ,  Jing XU

IPC分类号： G06T13/40 ,  G06T7/70 ,  G06T7/90 ,  G06T13/20 ,  G06V10/774 ,  G06V10/776 ,  G06V10/82 ,  G06V20/40 ,  G06V40/16 ,  G10L25/69

CPC分类号： G06T13/40 ,  G06T7/70 ,  G06T7/90 ,  G06T13/205 ,  G06V10/774 ,  G06V10/776 ,  G06V10/82 ,  G06V20/46 ,  G06T2207/10024 ,  G06T2207/30201 ,  G06V40/174 ,  G10L25/69

摘要： This application discloses a method for training a video generation model performed by a computer device. A phonetic feature, an expression parameter, and a head parameter are extracted from a training video of a target user. Network training is performed on a neural radiance field based on the condition input, three-dimensional coordinates, and a viewing direction to obtain a video generation model. The video generation model is obtained through training based on an image reconstruction loss. By introducing the head pose information and the head position information in the training process, a consideration of a shoulder motion status can be introduced into the video generation model so that a motion between the head and the shoulder is more coordinated and stable.