This is a LoRa model I trained using the same materials used for my SDXL version. I decided to use Civitai's trainer as I could use a lot more steps than my previous setup, even though I couldn't use the LoCon format I've been using in my older model. The facial markings and the rest of Belldandy's facial features are all included in the "Belldandy" tag, so no extra description is necessary. Adding them on top of that tag is going to mess them up, either their shape or color. Her outfits aren't included of course. While I could've trained the model to just make her official goddess dress, I find this very restrictive when it comes to generating other outfits.

Files with SHA256: e61d2cc0ef5b4b45450e4447c7ae189991e085af55b00bce0a98b49110b5d1c4