That's right! This is a 2-in-1 LoRA! I'm surprised I even got it working...

Dataset:

• 26 images of Artibun from Artifyber's Art Dump, all from 2024;

• 10 images of Eros from Artifyber's Art Dump, also from 2024.

Trigger Words:

• Style: artifyber_style;

• Artibun: artibun_(artifyber);

• Eros: eros_(artifyber).

Caveats:

• Generating Eros is a bit difficult. You have to be a bit more descriptive.

• Generating both of the characters in the same image is also very difficult.

• All of the sample images have been generated with the "Anima v1.0 Base int8 ConvRot" model, with the "Anima Turbo" lora, at a really low CFG of 2, 14 steps, LCM sampler, sgm_uniform scheduler, at 512x768, 4x upscaled. So, the final result highly depends on your settings.