THIS IS A VPRED LORA. IT WILL NOT WORK IF YOU RANDOMLY SLAM IT INTO WHATEVER MODEL YOU LIKE. SEE THE SUGGESTED RESOURCES FOR MODELS THAT WORK. YOU WILL NEED THEIR YAML FILES AND CFG RESCALE SET UP.

Generates Mimiga from the game Cave Story. Sue Sakamoto pollutes the dataset heavily, so expect to see a lot of her face x thing in the results. I could use the LoRA output to train another version that is less predisposed to the bias, but I don't really feel like it right now, sorry.

The sample images don't have gen data but the prompts were very short, I think the first one was like "mimiga, rainbow shirt, kneeling, outside, grass" and that was it.

Note: old, bad model. Use at your own risk.