What would happen if I fed a video to an AI that turns images into blobs of labeled objects, and then fed THAT video to another AI that attempts to reconstruct the original picture from those labeled blobs? I used runwayml [https://runwayml.com]’s new chaining feature to tie