search

zhanglonghao1992 / One-Shot_Free-View_Neural_Talking_Head_Synthesis

Pytorch implementation of paper "One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing"
Other
735 stars 141 forks source link

Questions regarding the best frame #80

Open Romatic-zbj opened 2 months ago

Romatic-zbj commented 2 months ago

Hello, I would like to ask why best_frame is chosen for inference, rather than directly generating the video frame by frame from start to end?