A style LoRA based on a 2231-image subset of card arts from Tokyo 7th Sisters. Dataset provided by Civitai user fizzballs.

Images were not tagged by character.

Had some faint hopes of the characters just naturally getting memorized by tags like some other style models do, but sadly that didn't happen.

The next step would be adding character tags either manually or programmatically, but I'm not seeing myself doing it anytime soon.