This monster LoRA was created from 70 high-resolution film monsters. Each image was then sharpened with Flux Context and upscaled with a LoRA separately created for that purpose. The LoRA doesn’t recognize monster labels like Vampire, Goblin, Werewolf, etc., but relies on precise descriptions. Currently, it still has weaknesses in certain areas, and I’d appreciate it if you could let me know if you find any. I’d also be happy to receive suggestions for cool film monsters that I could integrate, especially those whose base form and characteristics are currently not yet or only very rudimentarily present.

All my images were created with the following properties:

Steps: 20-35

Sampler: DPM++ 3M SDE

Scheduler: Beta

Kohya Deep Shrink