The model did not seem to grasp the concept of a ???????? very well. Hands and ??????? can be problematic at times. I use a strength of around 1 for both model and clip. If there is any interest, I'll work on a v1.0 to improve some of these issues.

I've uploaded v0.4 (4000 steps) and v0.6 (6000 steps). v0.4 seems to produce better results for simpler scenes, while v0.6 helps when the scenes are more complex. You can decide which one you like.

Also try out my SDXL version.

If you have enjoyed any of my LoRAs and want to see more, consider supporting me. Anything helps!