SDXL

Tricky action but I think I finally got something good enough to post.

Works best with simple prompts at 720x1024 of 1024x1024.

Definitely still produces a lot of grabled nonsense but it gives something usable about 40% of the time.

Trained on base SDXL but I am getting good results on Juggernaut and I think other models should work well too.

Keep your prompts real simple
I used the base prompt :
woman . <lora:2FingersSDXL_v03:1> ,
for most of my example images.

Happy ! Let me know if you figure out any tricks to get good results.