Trained on 768*768 images instead of 512*512
This LoRA is trained on images from /r/LiminalSpace
The base model is Suzumehachi.