V3 instructions

• Syntax is important, ex.1: cum, a naked man, penis. ex.2: cum, a naked latino man with long hair and tattoos, penis

• I use DPM adaptive (important note) for best results with a CFG of 8 and a lora strength of very close to 0.7

V2 works much the same as V3 except cum isn't so optional and it was trained on less images

Please leave any suggestions on how this model can be improved

sample images upscaled X2 to fix issues with blurry faces

https://ko-fi.com/alightbulb if you're feeling generous