Recommended weights for txt2img : 1-1.4
Description
FAQ
Comments (3)
哇哦！
yo , can you share the training dataset if possible , trying to train a lora for evernight , as her fan arts are quite complex , it would be helpful to see what images and tags you used .
I used ten pictures. The pictures are portraits, full body and half body. I erased the background on each one. When labeling, I did not write the character image, costume features related, just describe the action, expression, and white background