让图1的人物提着图2的包，保持人物其余特征不变
Want a different result? Change just a few parts
AI finds the swappable parts (subject/style/scene) — fill in what you want, keep the rest
