Notes on usage:

1. LoRA weights around 0.8~1.1 should be fine.

2. Generating resolution of 1k*1k gives simple picture style, using hires-fix will give a much more detailed painting.

3. Works well on both scenery and characters.

4. The model is trained on AnythingV5, but also works for 2D bases such as AnyLoRA, CounterFeit.

Purpose of this model:

A LoRA model on art style test, trained on AnythingV5 base model, with 88pics, 10reps.