Read Me!
This LoRA is GREAT for inpainting
Generate base image with LoRA strength at least 0.3-0.5
Inpaint or use ADetailer for ultra-accuracy at strength = 1
Denoising strength ≈ 0.4 (estimate)
If your inpainting results suck, try these tips:
increase LoRA strength on base image
increase denoising strength
set inpainting resolution to 1024x1024
use a higher/lower padding (around 72 for far, 100+ for closeup)
Trained on Pony Realism v2.1
CFG 3-4 works better than recommended CFG 6-7
Use "fingers" in negative if you want to avoid hands
Modified but similar dataset & captions as original InniePussy1
Recommended Settings
Model: Pony Realism v2.1
Steps: 20-30
CFG: 3-4
Height/Width: any SDXL values
LoRA strength: 0.2-1 (higher = more closeups)