Open wade0604 opened 5 months ago
https://github.com/Zhen-Dong/Magic-Me/blob/16bf220e7f6319a8a5456e879c077f51c987413e/train.py#L357
we apply the masks on the original pixels. The masks are used to only keep the subject of the images.
https://github.com/Zhen-Dong/Magic-Me/blob/16bf220e7f6319a8a5456e879c077f51c987413e/train.py#L357