Ver. 1.0B

Improved training

More compatible with many loras

This version was trained with the anime-full-final-pruned model, which makes it more flexible with models and loras.

I recommend using anylora for image generation.

The samplers: Euler a and DPM++ 2M Karras give better results

Use the weight from 0.9 to 1 to generate the style as well as its trigger word.

Have fun and share your generations

That's all!

Ver.1.0A trained with anylora, the recommended weight is 1, use the triggerword for better results.

you can generate mie using the following tags:

glasses, long hair, serafuku, sailor collar,yellow neckerchief, white shirt, purple eyes, school uniform, red skirt, red hair share your generations and enjoy it!