>>52266204There are a lot of variables. The hypernetwork I trained of myself used like 12 - 15 selfies at different angles. It took about 3 hours to train on my 1080 Ti, so I just left it running over night. You get much better results training with Dreambooth, but you can't do that with anything less than a RTX 3090 or something IIRC.
Also, I was curious, this is what it think a Sylveon looks like without and additional training, kind of interesting.
