This is a new FE 10 LoRA.
I retrained it after I noticed many issues with the original dataset.
I fixed up the dataset, re-tagged it, and retrained it.
I still need to prune some tags but it works well enough regardless.
It's trained at 768x768, use weight around 0.7-0.8
Upscale to get rid of pixilation.
Please add (Angel wings, feathers:1.3) to negatives to prevent unwanted wings!!