Open orena1 opened 1 month ago
are you doing machine learning stuff? is your concern on the memory usage? i have been solving similiar problem in pytorch land rendering out frames to a new dataframe npz file https://github.com/johndpope/MegaPortrait-hack/blob/main/EmoDataset.py#L59
If I have a 1200x1200 video, can I load only crop (central 100x100) or maybe every second pixel (spatial subsampling)? Thanks