Training the TecoGAN model into REDS Dataset

I have been modifying the code to have this run to windows but loading REDS Dataset seems to be hard as the program does not seem to want to get png frames of videos. Is it possible to modify the program in order to be able to have the model trained using the REDS dataset and also have Windows compatibility?
