1. 图生视频，一张照片
适合只能输入一张图片的工具（大多数“照片转视频”模型）。
Vertical 9:16 handheld phone video, 9 seconds, one continuous shot at an evening quinceañera party in a hall with white curtains, gold chairs and pink and green lights. The person in the image is the only dancer in the foreground and keeps their exact face, hair and clothes in every frame. 0–2.5 s: they dance toward the camera with loose shoulders and small bouncy steps, holding a round Mexican sweet bread (concha) with a pink sugar top. 2.5–3.5 s: they lift the concha and take one clear bite; a piece is missing afterward. 3.5–5.5 s: close-up, cheeks full, chewing slowly, staring straight into the lens with a completely straight face, no smile, no talking. 5.5–9 s: still chewing, they turn and dance back between the guests with one arm raised while the camera follows. Guests stay in the background. No cuts, no text.
反向提示词
smiling, laughing, talking, singing, lip-syncing, extra hands, extra fingers, bread melting into the hand or mouth, face changing, second dancer in the foreground, cuts, captions, watermark
2. 参考视频生成，照片加原片
适合能同时输入参考视频和照片的工具。在没有现成模板的情况下，这是最接近原片的做法。
Use the reference video only for motion, camera path, framing and timing. Replace the foreground dancer with the person in the image for the whole clip: face, hair, skin tone, build and clothes come only from the image, including the close-up and the back of the head. Keep the party hall, the lights and the background guests as in the reference, and do not put this face on anyone else. Keep the four moments on time: dance toward the camera (0–2.5 s), one bite of the pink concha (2.5–3.5 s), a deadpan chewing stare into the lens, no smile (3.5–5.5 s), turn and dance back into the guests (5.5–9 s). One continuous vertical shot, no cuts, no added text.
反向提示词
original dancer’s face, smiling, talking, skipped bite, extra hands, melting bread, face swaps between people, new scene
3. 同一个梗，换一身衣服
加在第 1 或第 2 条后面，只换衣服，这里以燕尾服为例。
Dress the dancer in a black tuxedo with a white shirt and a black bow tie. Change only the clothes: the face, the party hall, the pink concha, the bite and the straight-faced stare stay exactly the same and happen at the same seconds.
反向提示词
smile, open-mouth grin, skipped bite, bread swapped for another food, the outfit changing back mid-clip, logos or lettering on the clothes, extra people in the foreground
