Trained on 14 images (art, cosplay, and movie stills). She's got the textured skin of the live action version. It can get a little over-zealous sometimes, but is manageable. Doesn't need clothes, but occasionally decides to wear a belt, and likes white dresses.
Examples are cherry picked from a batch x/y/z plot of various styles and checkpoints using a single random seed. No post processing of any kind (eg - inpainting, photoshop, restore face..) Xformers was enabled, so your image may differ slightly.
Description
Trained on SD1.5 for two epochs with 10 repeats per image, using Hollowstrawberry's Lora Trainer.
Works best with semi-realistic models due to skin color.