search

YuvalNirkin / fsgan

FSGAN - Official PyTorch Implementation
https://nirkin.com/fsgan
Creative Commons Zero v1.0 Universal
749 stars 147 forks source link

target and source image for training #119

Closed saksham20aggarwal closed 3 years ago

saksham20aggarwal commented 3 years ago

Hi. What source and target pairs were used for training the model? Is it that source and target pairs were both taken from a single video and the task was to recreate the same above mentioned video?

YuvalNirkin commented 3 years ago

Yes, for the reenactment and inpainting models, the source and target frames are taken from the same video and the goal is to reconstruct the target frame. For blending the source and target frames are taken from different videos.