This LoRA tries to imitate AllCharacters18's artstyle.

The trigger is by AllCharacters18.

There were a couple bar censors and stuff in this dataset, but a good chunk of it was uncensored, so putting that in positive might get better results than just negativing everything for NSFW prompts.

Illustrious version: trained on a dataset with a mixed caption of JoyCaption + Wd-Moat, on PlantMilk: Walnut, mostly as an experiment. Don't expect more IL models.