Inspired by the work of John Buscema.

This is a LoKR and will work in ComfyUI and other generators which support LoKR architecture.

Trained using a single image reference with depth maps in Perceptual LoRA Toolkit.