So, this is an experiment regarding single image datasets.
God help me, I don't have the buzz for this, but the curiosity is too strong.

Trained on this image on the Hades II wiki.

It... works?
Not the best model by any means, but you get what you put in - Hera.

Needs all the tags to consistently get her, and hands go funny, but it works.

Cool tags: milf, peacock feathers

Might use this to train a better Lora at some point.