LoRA for the eromangaka Katsurai Yoshiaki, trained on sets of manga panels.

On certain finetunes the model may colourise outputs, prompting

monochrome, greyscale

can fix these issues if they do arise. I'm inferencing on WAI v14 and haven't had to do this. It may help to add

no_pupils

to the prompt if you want the blank eyes that appear in a few of his doujinshi.

This LoRA was trained on uncleaned manga panels, so adding

text, signature, watermark, speech_bubble, sound_effects, (((english_text, chinese_text, japanese_text, engrish, engrish_text))), username, censor, mosaic_censor, bar_censor, blur_censor

or similar (tweak if necessary) will help reduce artefacts. Unless you want these things, in which case prompting them should work.