Improved dataset to increase resolution and avoid face transfer and cross contamination. I've come to learn a LoRA such as this is not a specialist, but a Jack of all Trades, master of none. It is meant to simplify prompting.
I created this out of a limited training set. It is meant to be used inside the Detailer/Adetailer prompt specifically. Showcase pictures also include positive/negative embeddings.
Running into mismatch issues is a real possibility, Adjusting LoRA strength and/or Denoise often resolves the issue.
