how to decomposition facial expressions and head pose rather than output them together? - Githubissues

AliaksandrSiarohin / first-order-model

This repository contains the source code for the paper First Order Motion Model for Image Animation

https://aliaksandrsiarohin.github.io/first-order-model-website/

MIT License

14.57k stars 3.23k forks source link

how to decomposition facial expressions and head pose rather than output them together? #475

Closed haiewu closed 3 years ago

AliaksandrSiarohin commented 3 years ago

You can check how it is done in this paper: https://openaccess.thecvf.com/content/CVPR2021/html/Wang_One-Shot_Free-View_Neural_Talking-Head_Synthesis_for_Video_Conferencing_CVPR_2021_paper.html