A small and cleaned dataset compared to the experimental model, focusing exclusively on fantasy anime and a bit of cartoon style. The dataset contains 100% of DALL-E 3 images.
Works with AAM - AnyLoRA Anime Mix - Anime Screencap Style Model
Recommended:
CFG scale: 4-9
Strength: 0.5-0.8
Sampler: DPM++ 2M Karras, Euler