I was able to train and generate great photos using the flux-dev-lora-trainer (link below), however there’s one HUGE problem. I uploaded a photos of myself with very short trimmed hair. On the generated photos I have a long hair. I assume the problem is that the AI “plants” my face and not my entire head.
Any solution?
Thanks a lot
One primitive solution is to add like these to the prompt.
man with very short hair, short hair, …
This is particularly noticeable with anime characters, but if you are strongly teaching things like hair length and color, there is probably some established know-how, so if you want to remake it, I think it would be a good idea to search for that.