Anima model version of my gothic anime screencap lora<h3 id="trigger-word-g4ss">Trigger word - G4SS</h3>P.S.: This version was trained with poor hand recognition results due to a low-quality dataset. I tried retraining it with a modified dataset, but I couldn't achieve the desired style (lucky random). Therefore, you definitely need "bad anatomy, bad hands, missing finger" in your negative prompts, maybe this should helpThe trigger word also seems to have no effect, because I forgot about "Keep token" and "shuffle caption". You can use trigger word just to make sure it worksWhy this version? Because it's pleasing to my eyes, and I couldn't achieve the same result by retraining, even with a good dataset and tweaking training options

Files with SHA256: cc9fcf030e73d29cb0b9ffdf209342193b2317c99a35a0d8dcfe15b9cbbc40ce