I am a newcomer learning LoRA training, and this is my first practice piece. So I cannot guarantee the quality of the images generated by this LoRA.
It's a simple LoRA trained with 20 game screenshots (10 original dress and 10 qipao).
Recommended strength above 0.9, below 0.9 can cause clothing errors.
The performance of the original dress is not so good, but the qipao is okay.
And of course, there's a big problem with the hands.
Recommended big model: final-pruned, Anything v3
There may be major errors in other mix models.
Trigger words：Hades, yuanpi, wangchuandengying