Trained from 100 images by illustrator Kitada Mo. Works best with Anything V5 but is flexible enough to work on other models and still produce good results. Better results with more sfw prompts but can produce simple nsfw results or more complex ones with help from other LoRA. The image used for the thumbnail was generated using this LoRA.
Recommended weight: 0.7 ~ 1 works best, I found that a 0.8 weight was the most consistent.
CFG scale: 6 seemed to produce the best results for what I tested but it works fine at 5 or 7 too.
