I might argue that generating and decoding an actual NTSC signal, as the OP project does, would be true in ways that a generative model based on all of that would not be.
replyCan't you just cut out the AI from this pipeline by recording footage onto VHS and then digitizing that?
replyI think the idea would be allowing arbitrary videos to be converted without the need for a working tape drive
reply