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!!