30 秒视频,50+ 参考图。 立即体验
Seedance 2.0 提示词写作

Seedance 2.0 提示词指南

学习如何为电影感视频、产品广告、UGC、动漫、短剧和音乐视频写出更好的 Seedance 2.0 提示词。

利用提示词公式、时间轴、参考素材和实用案例,获得更可控的 AI 视频效果。

提示词基础

什么是好的 Seedance 2.0 提示词?

好的 Seedance 2.0 提示词应该清楚说明视频中出现谁、发生什么、镜头如何运动、遵循什么风格,以及要避免什么。对视频生成来说,主体和动作是最重要的部分。其他细节,如灯光、场景、运镜、风格和限制条件,可以根据你的需求补充。

原始提示词

A man walking in a market.

优化后的提示词

A young man walking through a lively outdoor street market at golden hour. He holds a small bouquet of flowers and looks surprised when he notices someone off-camera. The camera starts with a close-up of the bouquet, then slowly pulls back to reveal the crowd. Warm sunlight, shallow depth of field, natural motion, stable face, clean hand details, no text, no watermark.

优化提示词的效果

框架

Seedance 2.0 提示词框架

一个强有力的 Seedance 2.0 提示词从明确的主体和动作开始,这两点是必需的。其他细节,如场景、灯光、运镜、视觉风格、画质和限制条件,也建议加入,以让结果更可控、更一致。

01

精确的主体

主体是谁,或是什么?

02

动作细节

主体在做什么?随时间发生了什么变化?

03

场景与环境

场景发生在哪里?包含哪些背景细节?

04

光影基调

时间、光照风格、色调和氛围是怎样的?

05

运镜

镜头该如何运动或取景?

06

视觉风格

视频应遵循什么视觉风格?

07

画质参数

视频应该有多清晰、多细腻、多精致?

08

限制条件

模型应避免什么?

基础提示词公式

主体 + 动作 + 场景 + 灯光 + 运镜 + 视觉风格 + 画质 + 限制条件

@ 引用

如何用 @ 引用参考素材

当你向 Seedance 2.0 上传参考素材时,可以在提示词中使用 @ 引用,告诉模型每个文件应如何使用。这对多参考 / Omni Reference 工作流尤其有用。有了 @ 引用,Seedance 2.0 就能明确知道哪个参考素材控制角色、服装、场景、动作、音频或编辑目标,而不用靠模型猜测。

如何在 Seedance 2.0 中使用参考素材

角色参考

Use @image1 as the main character reference.

服装参考

Use @image2 as the outfit reference.

场景参考

Use @image3 as the background and environment reference.

动作参考

Use @video1 as the motion reference.

音频参考

Use @audio1 as the rhythm or voice reference.

首尾帧

Use @image1 as the first frame and @image2 as the final frame.

视频延展

Use @video1 as the previous clip and continue the same scene.

视频编辑

Use @video1 as the original video reference and change the background, outfit, or style.

应用场景

按场景分类的 Prompt 示例

浏览 Seedance 2.0 的 Prompt 示例,涵盖电影感视频、产品广告、社媒短片、UGC、动漫、短剧和音乐视频。

电影感

A dramatic horseback chase through a canyon at golden hour. A rider on a black stallion gallops at full speed along a narrow ledge, red dust billowing behind them. The canyon walls tower on both sides, glowing amber in the low sun. The horse leaps over a gap in the trail, all four hooves off the ground, mane and tail streaming. The camera tracks alongside from a parallel ridge. Rocks crumble from the ledge edge. The rider looks back over their shoulder. Hyper-realistic, epic Western cinematography, 8k, thundering hooves echoing off canyon walls.

广告与产品

(0-3s) Macro shot of a luxury perfume bottle among scattered pink peonies, shallow depth of field, petals floating in warm afternoon light, soft ambient music. (3-7s) Camera glides closer, a feminine hand enters frame from the right, fingers gently touch the glass bottle, the sound of silk rustling. (7-12s) Hard cut to slow-motion spray, golden mist diffuses through the air, particles catching rim light against a dark background, the hiss of the atomizer. (12-15s) Seamless pull-out to hero frame, product centered, volumetric lighting, minimal cream background, elegant silence. Hyper-realistic, 8k, fashion commercial cinematography.

电影感

A tight medium shot of two eccentric adults in typical, everyday clothing, one in a slightly oversized trench coat, the other in a weathered denim jacket, deep in a heated, animated conversation on a rainy West Village street corner. The one in the trench coat gestures wildly, the words “SOURDOUGH PRETZEL” appearing in pulsating electric blue: “It’s not just a pretzel, Arthur! It’s a sourdough pretzel!” The second guy says ‘Who cares. A pretzel’s a pretzel!’

UGC

A realistic young woman sitting casually in a softly lit bedroom during late afternoon. She is holding her phone very close to her face as if recording a private video or voice note. Framing is tight and slightly imperfect. Expression: thoughtful, slightly shy, natural. Minimal makeup, natural skin texture, relaxed clothing. Lighting: warm natural light fading from a window, soft shadows. Environment: simple bedroom, calm and lived-in. Style: ultra-realistic, looks like a real phone recording, slightly grainy, not cinematic.

UGC

Create a 10-second UGC-style fashion video. Use the girl in @Image1 as the main character. Keep her face, hairstyle, body shape, and overall vibe consistent throughout. Use @Image2 as the green top reference, and @Image3 as the handbag reference. [0s-2s] The girl faces the camera and holds up the green top from @Image2, showing it naturally with a cheerful expression. Then quickly transitioned to her wearing the green top. [3s-5s] She starts doing her makeup in front of the camera. [6s-7s] She fixes her hair while looking confident and stylish. [8s-10s] She picks up the handbag from @Image3 and presents it like a model. Keep the character, green top, and handbag consistent. Avoid deformation, outfit errors, hand issues, identity drift, and messy background.

社交媒体

Replace the man in black who is dribbling in @video1 with the girl from @image1. Keep the original scene, camera movement, dribbling motion, body posture, and basketball trajectory from @video1. The girl should closely follow the same movement and timing. Keep her face, hairstyle, and outfit consistent with @image1. No background change, no camera change, no body deformation, no hand errors, no ball clipping.

动漫

Use @image1 as the main reference image. Keep the character design, outfit, expression, and cute anime livestream style consistent throughout. Create a 15-second Japanese anime-style livestream video in a cozy bakery kitchen. The main character is a lively young female baker and livestream host, wearing the same outfit and headset style as @image1. Use warm lighting, soft pastel colors, and simple livestream UI elements such as a “LIVE” badge, floating hearts, comment bubbles, and product tags. Keep the UI clean and do not block the character. Include smiling, waving, kneading dough, presenting baked bread, natural Japanese lip sync, smooth camera movement, warm cozy lighting, and cute anime style. Avoid character inconsistency, outfit changes, dark lighting, cluttered UI, and oversized text.

短剧

[0-3s] Sweet Wedding: Mr. Mango and Miss Banana stand under a floral arch, smile at each other, and exchange rings in a bright warm wedding hall. [3-6s] The Intruder: Miss Strawberry bursts through the doors and shouts, “You can’t marry her! She’s a rotten banana!” [6-9s] The Reveal: Strawberry pulls down Banana’s dress hem, revealing a rotten blackened banana peel. Banana screams, “Wait! Let me explain!” [9-12s] Mango Runs Away: Mango runs toward the exit while Banana cries out. [12-15s] Final Twist: Sprinkler water pours over Mango, the golden paint melts away, and he is revealed as a potato. Cinematic, absurd, dramatic, clear subtitles only for dialogue.

音乐与对口型

Use the audio from @Video 1 to drive the protagonist in @Image 1 to generate an MV with multiple shots, change the outfit and background, and create an overall light, lively, and sunny atmosphere. Lip-sync strictly: "Waking up to golden light. Coffee warm and feeling right."

常见错误

Seedance 2.0 Prompt 常见错误

即使创意不错,如果 Prompt 过于笼统,或参考素材说明不清楚,效果也可能大打折扣。

多参考未使用 @ 引用

使用多参考 / Omni Reference 时,@ 引用相当于给上传素材下达指令。如果上传了图片或视频却没有说明其作用,Seedance 2.0 可能无法判断该如何使用它们。

较弱的 Prompt

Use this cup to make a video.

更好的 Prompt

Use @image1 as the main subject. Keep its vintage floral pattern and warm ivory color unchanged.

较弱的 Prompt

Uploading an image and a video without using @mentions.

更好的 Prompt

Use @image1 as the main subject and @video1 only as the motion reference. Follow the wrist movement and stirring rhythm from @video1.

情绪描述过于抽象

当情绪通过可见的动作、表情和肢体语言呈现时,Seedance 2.0 的效果会更好。

较弱的 Prompt

She is sad.

更好的 Prompt

Her lower eyelids are red and swollen, tears gather in her eyes but do not fall.

较弱的 Prompt

He is angry.

更好的 Prompt

He throws the phone onto the sofa, paces back and forth, breathes heavily, and clenches his jaw.

较弱的 Prompt

He is nervous.

更好的 Prompt

His fingers keep rubbing the corner of his shirt, his eyes quickly scan the room, and his breathing becomes shallow and fast.

核心原则:情绪要“被看见”,而不只是“被说出”。

未给出运镜指示

如果不描述镜头,Seedance 2.0 会自行决定画面构成,结果可能更随机。

较弱的 Prompt

A beautiful shot.

更好的 Prompt

Low-angle wide shot with slight lens distortion, blurred foreground, and backlit silhouette.

较弱的 Prompt

Move closer.

更好的 Prompt

Close-up shot focused on the eyes, with clear eyelashes and window reflections in the pupils.

较弱的 Prompt

No camera direction.

更好的 Prompt

Medium shot, slow push-in, focus shifts from the character’s face to the object in their hand.

核心原则:在需要时加入景别、镜头运动和对焦点。

动作设计过于复杂

Seedance 2.0 通常在流畅、连贯、自然的动作下表现更好。过于复杂或快速的动作会增加变形或结果不稳定的概率。

较弱的 Prompt

Intense fight scene, fast dancing, multiple people jumping and running.

更好的 Prompt

Slow turn, gentle hand movement, hair moving softly in the wind.

较弱的 Prompt

Complex interaction between many people.

更好的 Prompt

Two people stand face to face, make eye contact, and lean forward slightly.

较弱的 Prompt

Fast chase scene.

更好的 Prompt

Slow-motion walking with the coat gently moving in the wind.

核心原则:追求流畅、稳定、自然的动作,而非过度复杂的动作设计。

忘记添加约束条件

约束条件有助于减少常见的 AI 视频问题,比如面部扭曲、手部不稳定、画面闪烁或身份漂移。

较弱的 Prompt

No constraints.

更好的 Prompt

Stable face, natural body structure, clean hands, complete fingers, smooth motion, no flickering, no identity drift.

常见问题

关于 Seedance 2.0 Prompt 的常见问题

Seedance 2.0 Prompt 支持多少字符?+
在 SeeGen AI 上,Seedance 2.0 支持最长 4000 字符的提示词。这足够用来描述主体、动作、场景、镜头运动、风格、画质、约束条件,以及需要时的时间线。
更长的 Seedance 2.0 提示词一定更好吗?+
并不总是。更长的提示词不代表更好的结果。好的提示词应该清晰、聚焦、易于理解。对大多数短视频来说,一段详细的描述或一条简单的时间线就足够了。
Seedance 2.0 提示词中最重要的部分是什么?+
最重要的部分是主体和动作。务必明确说明视频中出现的是谁或什么,以及随时间发生了什么。
我应该在提示词中使用时间线吗?+
是的,尤其是 8-15 秒的视频。时间线能帮助 Seedance 2.0 理解动作和镜头运动应如何随时间变化。
如何让 Seedance 2.0 更好地遵循参考文件?+
使用 @ 引用来说明每个参考文件的作用。例如,写 @image1 作为主角参考,或 @video1 作为运动参考。
如何减少面部扭曲或手部问题?+
在提示词末尾添加明确的约束条件,例如:面部稳定、身体结构自然、双手干净、手指完整、无面部扭曲、无身体变形、无闪烁、无身份漂移。
写 Seedance 2.0 提示词时应避免什么?+
避免模糊的指令、在一个短片段中塞入过多动作、参考用法不明确,以及缺少约束条件。让提示词保持聚焦、具象,并便于模型理解。

写出更好的提示词,生成更好的视频。

从清晰的主体、动作、参考、镜头方向和约束条件开始,在 SeeGen AI 上创作可控的 Seedance 2.0 视频。

免费试用 Seedance 2