面向文本、图片与多模型工作流的 AI 视频生成器
使用当前可用的 Seedance、Veo、Kling 与 Wan 工作流创建文生视频、图生视频和参考式片段。
Meet Seedance 2.0
Step into the new era of cinematic AI video creation.
Seedance 2.0 does not accept real human faces as reference inputs, and cannot reliably generate copyrighted characters (celebrities, anime, or film characters).
Cinematic action scene, a compact futuristic speeder with glowing twin thrusters races through narrow stone streets carved into a cliffside city. The camera follows from behind, weaving between sunlit walls as dust and motion blur trail the vehicle through the canyon-like passage.
Examples
1
Seedance 2.5 is live
Generate Seedance 2.5 videos online
Open the dedicated Seedance 2.5 video generator to create 4–30 second videos from prompts, frames, images, video, or audio references at 480p, 720p, or 1080p.
创作者展台 SEEDANCE
看看我们的用户用 AI 视频生成创造了什么

“A dense industrial sci-fi fly-through with towering structures, wet surfaces, and heavy atmospheric haze.”

“A close study of exposed cybernetic anatomy with braided red cables, glowing details, and anime-grade linework.”

“A downhill longboard run on a tropical mountain road, captured with fast FPV energy and clean speed blur.”

“An anime street sequence in a lantern-lit market, with flowing fabric, wet reflections, and cinematic motion.”

“A stylized anime action shot with aggressive forward motion, sharp contrast, and impact-driven composition.”

“A handcrafted macro setup focused on small tools, precise hand motion, and premium tabletop lighting.”

“A natural lifestyle portrait in a bright dessert shop, with casual pose, clean skin tones, and social-video framing.”

“A top-down Formula racing shot on a wet track with strong reflections, graphic framing, and controlled motion.”

“A luxury interior scene with warm practical lighting, balanced decor, and calm editorial composition.”

“A split-frame transformation close-up comparing live-action eyes and stylized animated eyes with strong facial continuity.”
本工作区当前可用的输入、控制项和限制
文生视频控制
从提示词开始,并在提交前选择生成器实际展示的宽高比、时长、分辨率和输出数量。
以图像引导运动
使用源图像作为图生视频请求的视觉起点,再描述所需的运动和镜头行为。
首尾帧引导
上传首帧,或同时上传首帧和尾帧,用于约束片段的开场构图与目标结尾。
多模态参考模式
参考模式接受受支持的图像、视频和音频组合;上传面板会执行当前文件与槽位上限。
当前输出限制
已注册的 Seedance 2.0 能力范围为 4–15 秒、480p 或 720p;生成器会按所选模式展示精确选项。
将音频作为参考输入
受支持的音频文件可用于多模态参考请求。本页面不承诺自动对白、完美口型同步或特定输出音轨。
套餐与积分检查
Seedance 2.0 需要 Creator 或 Studio 套餐;界面会在提交前按模式、时长、分辨率和数量计算积分。
预览与历史记录
在预览和历史面板中跟踪已提交任务;完成的媒体与终态错误都会关联到你的账户。
常见 问题
Seedance 2.0 当前的输入、限制、访问条件与账户条款。