Closed cbarne10 closed 2 years ago
I loaded in a checkpoint from my 512x512 training into part_swap using --supervised and --use_source_segmentation so that the source replaces the targets' parts. It seems to be reproducing the source image as the output. Is there something I may have done wrong?
I did make the changes to the sigma, blend_mask and the yaml as noted in previous threads.
I loaded in a checkpoint from my 512x512 training into part_swap using --supervised and --use_source_segmentation so that the source replaces the targets' parts. It seems to be reproducing the source image as the output. Is there something I may have done wrong?