This is my first attempt at creating a LoRA. It is experimental, expect wonky stuff.

In case you get errors, install the webui locon extension. That said, the LoRA still works even without it.

Based on BADCOMPZERO's artworks

Haven't done many tests, though seems to best work with 0.6-0.8 weight. Showcase pictures generated on YesMix 1.5. Using an upscaler seems to make the eyes lose its style.

• Trained on zakps' YesMix 1.5

• ~300 image set

• 5000 steps

• 10 epochs

• Used WaifuDiffusion 1.4 vae