That reminds me of a thread here from a few weeks ago about how somebody was training an ML model to add noise to video, so that the noise-free version of the v

That reminds me of a thread here from a few weeks ago about how somebody was training an ML model to add noise to video, so that the noise-free version of the video can be compressed better but then the noise can be added at playback to maintain the authentic feel of old video.