A roughly thrown together concept using some old training materials. More of a test to see what training was like on Flux. It's pretty spotty but with some tinkering it can make some interesting images. Trained purely on anime-style images, so it can be hard to work with other styles. Since this is rough and I don't have a ton of knowledge on Flux, I won't be able to provide any assistance with this.Sweet spot seems to be around: 0.6-0.8Sampler: Euler -Simpler (but I've been playing with Deis - DDIM, for some decent images)

Files with SHA256: 06189f11dbbc61bced98d408bd3cc9e535e8b02e215c45463c9947f5e6e492d5