首帧模型的lora基于noobai ep11，只能写上半身的服饰和样貌等tag，不然会出错

Only upper body tags will work in the first frame lora.

Trained from noobai ep11.

Suggest:

分辨率(reso)1024*1024

weights: 1.2+

精度FP32

TIPS:

权重要视情况调高