移除图像中的所有路人，保留画面中【前面两个自拍的人】，保持画面背景其他物品不动。
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
