Talent · Character Library
Upload 3–8 reference photos of any person or character from different angles. Then place them in any scene with a single prompt. No LoRA training, no fine-tuning — just multi-reference inference via Nano Banana 2.