This LoRa aims to imitate Hattori Mitsuru's artstyle.

The trigger is by Hattori Mitsuru.

No censorship because this dataset was all ages.

This LoRa was quite ornery to train, and took some convincing to produce results that looked like the inputs. I recommend to dial up the strength to like 1.2 if it doesn't look similar enough. Unfortunately this might reduce compatibility with other LoRas. Might also depend a lot on the checkpoint mix. Another possibility you can try is lowering the quality scores in the prompt.

Trained mostly on the the artist's recent work.