4 条 Wan 3.0 提示词,都配真实成片:2 条文生视频、1 条参考图、1 条首帧。每条都是 480P、5 秒、带声音,$0.20 一条。万相 Wan 3.0 标准版(wan3.0-video)已在 HiAPI 上线,点任意一个「用这个提示词生成」,模型页会自动填好提示词。新用户注册送 200 积分。
1. 带声音的街头小吃
A steaming noodle stall at a night market, close on the cook's hands as she
lifts a tangle of noodles from a boiling pot and swirls them into a
blue-rimmed bowl; slow push-in, warm tungsten light against cool blue street
glow, steam drifting through the frame. Sound: rolling boil, the clink of a
ladle, crowd murmur and a scooter passing in the distance.
用这个提示词生成 →(480P、5 秒、16:9、开声音)
wan3.0-video,文生视频,480P,5 秒,$0.20。
手、面条、蒸汽、远处驶过的摩托车都出来了,声音也按「Sound:」那句来。声音单独写一句放在最后,列 2–4 个具体的声音。
2. 带人物的风景
Aerial shot rising slowly over layered rice terraces at sunrise, mist pooling
in the valleys, a single farmer in a straw hat walking along a narrow ridge,
golden light catching the water in each terrace; smooth continuous camera,
no cuts. Sound: soft wind, distant birdsong.
用这个提示词生成 →(480P、5 秒、16:9、开声音)
wan3.0-video,文生视频,480P,5 秒,$0.20。
农民沿田埂走,一镜到底,每层梯田都反着光。5 秒里镜头只升了一点;想要更明显的运镜,把动作写两遍(「rising steadily, climbing higher」),或者给 8–10 秒。
3. 用参考图让产品动起来
参考图(Image1):

Use Image1 as the product: the ceramic pour-over coffee set. Slow quarter
orbit around it on the wooden counter, a thin stream of hot water pours into
the dripper and steam rises, soft morning window light, shallow depth of
field, keep the shapes and glaze of the set exactly as in Image1. Sound:
gentle pouring water and a quiet kitchen.
带着这条提示词打开模型页 →(把你的产品图加为参考图,480P、5 秒)
wan3.0-video,1 张参考图,480P,5 秒,$0.20。参考图不加钱。
哑光黑滤杯、玻璃壶、杯子和胡桃木托盘都和 Image1 一致,暗色棚拍背景换成了提示词要的明亮厨房。要的「环绕」出来只是轻微移动。多张图按传入顺序叫 Image1、Image2。
4. 用首帧让静物活起来
首帧:

The candle inside the brass lantern catches and flares into a steady flame,
warm light blooms across the glass and the walnut table, the camera pushes
in slowly; dust motes float in the glow. Sound: a soft match strike and a
faint crackle.
带着这条提示词打开模型页 →(加上你的首帧图,480P、5 秒、画幅 adaptive)
wan3.0-video,首帧,480P,5 秒,竖屏,$0.20。
声音里写了「划火柴」,模型就在画面里加了一只拿火柴的手。想让火苗自己亮起来,就把火柴从声音那句删掉。aspect_ratio 设成 adaptive,视频跟着竖版首帧出了竖屏。
三行写法要点
- 场景、动作、运镜、光线,最后单独一句「Sound:」。声音默认开、不加钱,用上。
- 声音会往画面里加东西。写了「划火柴」就多出一只手,只写你想看到声源的声音。
- 先用 480P 出草稿($0.20),定稿后用同一条提示词出 720P($0.40)或 1080P($0.80)。
三种模式的请求怎么写
{"model": "wan3.0-video", "input": {"prompt": "...", "resolution": "480P", "duration": 5, "aspect_ratio": "16:9"}}
{"model": "wan3.0-video", "input": {"prompt": "Use Image1 as the product ...", "reference_image_urls": ["https://your-cdn.example.com/product.jpg"], "resolution": "480P", "duration": 5}}
{"model": "wan3.0-video", "input": {"prompt": "...", "first_frame_url": "https://your-cdn.example.com/frame.jpg", "resolution": "480P", "duration": 5, "aspect_ratio": "adaptive"}}
分辨率要大写,默认 1080P,每次都要写。首帧和参考素材不能在同一个请求里混用。
完整价格见 Wan 3.0 API 价格。复制面摊那条提示词,花 $0.20 出第一条。





