文本、图像、视频、音频四模态输入，精准复刻运镜，支持视频延长与音效生成。
三步创建 AI 视频
选择文生视频或图生视频，然后输入提示词或上传参考图片。
选择可用设置（如模型、宽高比、时长）并开始生成。
调整提示词并迭代，导出选项以套餐与设置为准。
Seedance 2.0 是一款多模态 AI 视频生成器，支持图像、视频、音频和文本四种模态输入。以下内容说明每项核心能力的工作方式及其在专业视频创作流程中的应用。
Seedance 2.0 accepts up to 9 images, 3 video clips (max 15 s total), and 3 MP3 audio files (max 15 s total) alongside a text prompt. You can freely combine these inputs in a single generation. Use the @ symbol in your prompt to assign each uploaded file a specific role — for example, referencing an image for character appearance and a video for camera movement.
This multimodal approach lets you control composition, motion, and tone simultaneously. Instead of relying on text alone, you can anchor the visual style with reference images, guide the rhythm with audio, and define the action with video clips.
Upload a reference video and Seedance 2.0 can replicate its camera trajectory, action pacing, and visual effects in a new generation. This is useful when you have a shot you like — a dolly push, a tracking pan, or a handheld shake — and want to apply the same movement to different subjects or scenes.
Combined with image references for characters and environments, motion replication gives you precise directorial control without describing every frame in text.
Seedance 2.0 can extend existing videos forward or backward. Upload a clip and describe what happens next (or what came before), and the model generates a seamless continuation. This is particularly effective for building longer sequences from short initial generations.
Extension preserves visual continuity — characters, lighting, and camera perspective stay consistent across the boundary between the original footage and the new frames.
You can replace characters or objects within an existing video while keeping the rest of the scene intact. Upload the source video alongside a reference image of the new character, and Seedance 2.0 handles the swap while maintaining motion and background consistency.
This capability is valuable for adapting content across different campaigns, localizing video assets, or experimenting with different character designs without regenerating the entire scene.
Seedance 2.0 generates sound effects and background music automatically. The audio output matches the on-screen action — footsteps, impacts, ambient sound, and musical cues are timed to the visual content. You can also upload reference audio to guide the tone and rhythm.
For music-driven content, audio inputs can serve as beat references. The generated video aligns its cuts, motion, and transitions to the rhythm of the supplied track.
Seedance 2.0 supports one-take long-shot generation that maintains consistent characters, environments, and camera flow across extended durations. You can also generate multi-segment videos and merge them with continuity between cuts.
These features are designed for narrative-driven projects where visual consistency across scenes is essential — product demos, short films, and brand storytelling.
Each generation can be configured with aspect ratio, duration (4–15 seconds), and generation mode. The output is rendered at 720p. You can iterate by adjusting one or two prompt elements at a time, comparing results, and converging on a final direction.
Export options and usage rights depend on your plan. Costs are displayed before generation so you can manage your credit budget across projects.
Seedance 2.0 支持同时上传最多 9 张图像、3 个视频片段（总时长≤15秒）和 3 个 MP3 音频文件（总时长≤15秒），配合文本提示词生成视频。你可以在提示词中用 @ 符号为每个上传文件指定参考用途——例如用图像指定角色外貌，用视频指定镜头运动。
多模态输入让你同时控制构图、运动和氛围。除了文本描述外，还可以通过参考图像锁定视觉风格，用音频引导节奏，用视频片段定义动作。
上传参考视频后，Seedance 2.0 能精准复刻其镜头轨迹、动作节奏和视觉特效。当你拥有满意的镜头——推轨、横摇或手持抖动——可以将同样的运镜应用到不同主体或场景。
结合角色与场景的图像参考，运镜复刻赋予你精确的导演级控制力，无需用文字逐帧描述。
Seedance 2.0 支持对已有视频进行前向或后向延长。上传片段并描述接下来（或之前）发生的内容，模型会生成无缝衔接的续写。这在从短片段构建更长叙事时尤为有效。
延长功能确保视觉连续性——角色、光线和镜头视角在原始素材与新生成帧之间保持一致。
你可以在保留场景其余部分不变的前提下替换视频中的角色或物体。上传源视频和新角色的参考图像，Seedance 2.0 会在保持运动和背景一致性的同时完成替换。
该功能适合跨活动内容适配、视频素材本地化，或在不重新生成整个场景的情况下试验不同角色设计。
Seedance 2.0 能自动生成音效和背景音乐。音频输出与画面动作匹配——脚步声、碰撞声、环境音和音乐节奏都与视觉内容同步。你也可以上传参考音频来引导音色与节奏。
对于音乐驱动的内容，音频输入可作为节拍参考。生成的视频会将剪辑、动作和转场与所提供音轨的节奏对齐。
Seedance 2.0 支持一镜到底的长镜头生成，在较长时长内保持角色、环境和镜头风格的一致性。你也可以生成多段视频并将其拼接，确保段落间的连贯衔接。
这些功能专为叙事驱动的项目设计——产品演示、短片和品牌故事讲述中，跨场景的视觉一致性至关重要。
每次生成可配置宽高比、时长（4–15秒）和生成模式，输出分辨率为 720p。你可以逐步调整提示词中的一两个元素进行迭代、对比结果，最终收敛到满意的方向。
导出选项与使用权限取决于你的套餐。生成前会显示积分费用，便于你在不同项目间管理预算。
以中性角度对比功能与流程适配度。
对比内容应基于可验证信息。
面向不同团队的 AI 视频草稿
用 Seedance 2.0 快速探索创意并迭代视频概念。
Draft short-form video concepts for TikTok, Instagram Reels, YouTube Shorts, and more.
Turn product images into short video drafts to explore presentation styles.
Storyboard ideas and create concept visualizations for pre-production.
Create video drafts for decks, training materials, and internal updates.
Explore video art, music visualizations, and experimental concepts.
Draft educational and explainer videos for learning content.
Seedance 2.0 定价方案
选择适合您工作流程的方案。功能与使用权限以套餐说明为准。
了解更多关于 Seedance 2.0
具体功能以页面说明为准。