Captioning Images for LoRA Training

Captions should name the intended trigger and describe visible details. Mention pose, outfit, object features, scene, and style when they matter.

Avoid private identity cues, real-person names, protected character names, and brands. Captions should help the model learn lawful visual concepts, not impersonation.