This LoRA focuses on creating a crossed-eyes effect and was trained on a photographic dataset.

It is designed to work with moderately wide compositions as well, but the eye details tend to be more consistent and stable when the subject is framed closer, preferably around the upper body.

Tips & Compatibility

Flexible Expressions: It supports a wide range of facial expressions, including smiles, surprised looks, tongue-out faces, and intense focus! Creative Ideas: Try pairing it with fun prompts—like a sci-fi glitch effect, extreme concentration during delicate work (like soldering), or a tiny bug sitting right on the character's nose.

これは寄り目に焦点を当てた、写真のデータセットでトレーニングしたloraです。ある程度引きの構図もできるよう設計しましたが、なるべく上半身に近い方が目の描きこみが安定すると思います。