Read the damn description!!!

I recommend using the LoRA model mainly for inpaint.

Inpaint steps:

1. Mask the face (with or without hair)

2. Increase the batch count or batch size for a larger selection of variations

3. Play around a bit with the denoising strength

4. Done

Tip: You can also process a face twice through the inpaint procedure.
Example: Original Image (DenStr 0.3) → Result Image (DenStr 0.3) → Completed image

Since the model is trained on a resolution of 1280x1280, I recommend using it only in resolutions between 960x960 - 1536x1536.