It is recommended to use kashi123 kashi123, highly detailed as the prompt trigger word. The styles of clothing can be checked in the lora-info tag. Interested friends can try to mix and match with other clothing Lora. More usage scenarios are still expected. I hope you will send more pictures!
Description
This version uses 13 high-definition pictures as the data set and is trained at a resolution of 1024. After 2250 steps of training, the loss convergence value is relatively the lowest.
这个版本采用的是13张高清图片作为数据集按照1024分辨率训练，训练2250步loss收敛值相对最低。