>>2892458No, there was no trigger words. so as long as you add the lora is active,
> I probably could've included three times as many images (just basically drop all the manga scans of his work I have on hand), would that have helped? Probably not, because I trained lora with less and worse. Its just some artists catch better than others.
I made a Nakayama tooru style with barely 50 pictures and it caught well, and it had less stepts than this lora.
Then again I'm just guessing here, it was because its very distintctive
