I played a bit with the SIREN network from the other post and found that it could be improved by a using a different sampler for batch generation. By feeding pixels across the entire video and not only a limited set of frames, we can a much more faithful reproduction of the video. The model is exact
