What is Seedance 2.0 and how is it different from Seedance 1.5 Pro?

Seedance 2.0 is ByteDance's latest AI video generation model with significant upgrades: multimodal reference inputs (up to 7 images, 3 videos, 3 audio tracks), first/last frame control, web search enhancement, and real-person content support via asset management. Seedance 1.5 Pro supports basic text-to-video and image-to-video with up to 2 images.

What is the difference between Seedance 2.0 and Seedance 2.0 Fast?

Seedance 2.0 produces higher quality videos at a higher cost (720p: 46 credits/s without video input, 28 credits/s with video input). Seedance 2.0 Fast generates videos faster at lower cost (720p: 37 credits/s without video input, 23 credits/s with video input). Both share the same features and API parameters.

What multimodal references does Seedance 2.0 support?

Seedance 2.0 supports two types of multimodal references: reference images (up to 9) and reference videos (up to 5). First/last frame control, reference images, and reference videos are mutually exclusive — you can use one mode at a time.

How does the pricing work for Seedance 2.0?

Seedance 2.0 uses per-second billing. For 720p: 46 credits/s (no video input) or 28 credits/s (with video input). For 480p: 21 credits/s (no video input) or 13 credits/s (with video input). When using video input, total cost = (input video duration + output video duration) × rate per second.

What is the ByteDance Asset Upload feature?

When generating real-person content, you need to upload your reference materials (images, videos, audio) to ByteDance's asset system first to get an asset ID. This asset ID is then used in the generation process instead of raw URLs. The Asset Upload button in the Seedance 2.0 form handles this workflow automatically.

What resolutions and durations does Seedance 2.0 support?

Seedance 2.0 supports 480p and 720p resolutions with durations of 4, 8, or 12 seconds. It offers multiple aspect ratios: 1:1, 4:3, 3:4, 16:9, 9:16, and 21:9.

Does Seedance 2.0 generate audio?

Yes, Seedance 2.0 generates native audio by default, including speech, ambient sounds, and sound effects synchronized with the video. You can toggle audio generation on/off in the advanced settings.

How does Seedance 2.0 compare to Kling 3.0 and Veo 3.1?

Seedance 2.0 excels at multimodal reference-driven generation with rich reference inputs (7 images, 3 videos, 3 audio). Kling 3.0 focuses on multi-shot storytelling with multilingual dialogue. Veo 3.1 emphasizes cinematic realism. All three are available on Sora2 Hub for comparison.

Seedance 2.0 AI 视频生成器 — 多模态参考视频与原生音频

Seedance 2.0 是字节跳动新一代 AI 视频生成模型，支持多模态参考输入。使用文本提示、参考图片（最多 9 张）和参考视频（最多 5 个）生成高质量视频。具备原生音频生成、首尾帧控制、网络搜索增强和 480p/720p 分辨率。提供标准版和快速版两种模式。

文本生成视频

提示词

探索 Seedance 2.0 的模型

Seedance 2.0 Seedance 2.0 Fast Seedance 1.5 Pro

多模态参考生成

Seedance 2.0 支持多模态参考输入，包括最多 9 张参考图片和 5 个参考视频。将文本提示与视觉参考相结合，实现精准的创意控制，生成忠实匹配创意愿景的视频。

原生音频与声音生成

与视频同步原生生成音频，包括语音、音乐、环境音和音效。Seedance 2.0 在单次生成中产出丰富的视听内容，无需单独的音频后期制作。

首帧与尾帧控制

定义精确的起始和结束帧以实现精准运动控制。上传参考图片作为首帧或尾帧来引导视频生成轨迹——非常适合创建流畅的转场、角色动画和具有可预测结果的场景构图。

真人内容与资产管理

Seedance 2.0 通过字节跳动的资产管理系统支持真人视频生成。上传肖像图片、视频或音频以创建个性化的 AI 生成内容——完美适用于数字虚拟形象、虚拟主持人和个性化营销。

如何使用 Seedance 2.0？

简单几步即可创建多模态参考 AI 视频

输入提示词或上传参考素材

用文本提示词描述视频内容，或上传参考图片与视频进行多模态生成。如需生成真人内容，请先使用资产上传功能准备素材。

配置生成设置

选择 Seedance 2.0（更高质量）或 Seedance 2.0 Fast（更快速、更低成本）。设置分辨率（480p 或 720p）、宽高比、时长（4/8/12 秒），并开关原生音频生成。

生成并下载

Seedance 2.0 生成带有同步音频的视频。预览效果，高质量下载，尝试不同的参考组合以获得最佳效果。

Seedance 2.0 常见问题

Seedance 2.0 是字节跳动最新的 AI 视频生成模型，具有重大升级：多模态参考输入（最多 9 张图片和 5 个视频）、首尾帧控制、网络搜索增强和通过资产管理支持真人内容。Seedance 1.5 Pro 支持基础的文本生成视频和图片生成视频，最多 2 张图片。

免费在线试用 Seedance 2.0

使用多模态参考、原生音频和首尾帧控制创建 AI 视频。体验字节跳动最先进的视频生成模型。