LoRA that tries to reproduce Sibaketsu/Kyokyota's artstyle. I trained this LoRA with NoobAI EPS for use with StableMondAI.

Trigger is:

s1baketsu

I'd also highly recommend adding the following tags:

watercolor \(artwork\), traditional media \(artwork\)

I'd recommend running it at around 0.7 - 1 Strength for best results, but your milage may very.

It seems to be pretty flexible at this step count, but prefers alternative fashion / skimpy clothing, athletic, and especially 4_fingers. It also seems to handle duo scenes quite well if you are using the right species.

There is also a fair amount of ???? in the dataset, but so far I have not seen anything from simple basegens that would lead me to believe there's any bleed over. Gremlins are fairly common as well if you're not using (solo, anthro) however.