LoRA created from... well this one has 3 versions and I'm still not very satisfied with results:v1 and v2 where trained in 246 images, while v3 used only 100. All used various art pieces with にゃもた's style:Recommended weight for v1 an v2 is 0.5 to 0.8 , while v3 is 0.7 to 0.8, as always, depending on your model and prompt.No trigger word is needed, but you can use 'nyamota' if you want.Important to mention that there is no 'better version' but:v1 and v2 have stronger traits creating a very characteristic smile and tending to pale skins overall;v2 due to the bigger size is also the stronger style of the 3 versions;v3 was trained in a entire different dataset and will create images that are a little darker and with more color, but will have a weaker style.Due to artist influence most imagens will have weaker colors and tend towards empty deep eyes.

Files with SHA256: 21ef512235e93de7032d092af66008389348c99d8be31def75c303adf0512e11