badanmu V10

The prototype is the Tiktok platform: badanmu

After selecting 6 complete video frames, the key angle images were screened using the McOrange model and Lora MengX Girl_Mix V40. After re-drawing the facial selection, I trained it with fine marking.

The main use scene of this model is for face modeling in video transfer. I hope you like it~

Trigger word: danmu

The prompt words for the cover image are as follows, and the bottom mold used is McOrange Realist V7.

Recommended number of iteration steps: 30

Sample recommended: DPM++ SDE karras Others are free to try

Lora: 0.2-0.7 Best at 0.4, can be combined with lora for private face optimization for a sweeter face shape.

When starting ADetailer, it is recommended to use the beautiful face prompt”