search

Doubiiu / CodeTalker

[CVPR 2023] CodeTalker: Speech-Driven 3D Facial Animation with Discrete Motion Prior
MIT License
515 stars 57 forks source link

how can I implement mean and standard deviation? #38

Closed zhengshi119 closed 1 year ago

zhengshi119 commented 1 year ago

image

Doubiiu commented 1 year ago

Hi, as we described in Sec. 4.3 (the 2nd paragraph):

  1. calculate the L2 distance of adjacent-frame facial motions
  2. calculate the mean and std along the temporal axis for each vertex