Use this reference-image prompt to recreate the outfits, poses, framing, and lighting of a multi-person source while preserving the exact number of physical people. It distinguishes people from shadows, reflections, and screen images, then keeps each person's head, limbs, clothing, props, and occlusion relationships correctly matched. It explicitly prevents cloned faces, extra bystanders, fused anatomy, and incorrect reflections. The source post, prompt, and media were checked item by item, and the page provides copy-ready Chinese and English versions.