1. Seedance-2.0视频
模力
  • 文档说明
  • 图像(Images)
    • Nano Banana生图
      • 官方格式(文生图)
      • 官方格式(图生图)
    • Gpt-image2图像
      • 生成图像
      • 图片编辑
  • 视频模型
    • Seedance-2.0视频
      • SeeDance2系列视频生成说明
      • 文生视频
        POST
      • 首尾帧生成视频
        POST
      • 多图参考生成
        POST
      • 多模态视频生成
        POST
      • 官方格式参数生成
        POST
      • 查询视频生成结果
        GET
  • 数据模型
    • Schemas
      • User
      • Log
      • Model
      • Token
      • Usage
      • PageInfo
      • Channel
      • Redemption
      • ApiResponse
      • ModelsResponse
      • ErrorResponse
      • Message
      • MessageContent
      • Tool
      • ToolCall
      • GeminiModelsResponse
      • ChatCompletionResponse
      • ChatCompletionRequest
      • ChatCompletionStreamResponse
      • CompletionRequest
      • CompletionResponse
      • ResponseFormat
      • ResponsesRequest
      • ResponsesResponse
      • ResponsesStreamResponse
      • ClaudeRequest
      • ClaudeMessage
      • ClaudeResponse
      • EmbeddingRequest
      • EmbeddingResponse
      • ImageGenerationRequest
      • ImageEditRequest
      • ImageResponse
      • AudioTranscriptionRequest
      • AudioTranslationRequest
      • AudioTranscriptionResponse
      • SpeechRequest
      • RerankRequest
      • RerankResponse
      • VideoRequest
      • ModerationRequest
      • VideoResponse
      • ModerationResponse
      • VideoTaskResponse
      • GeminiRequest
      • VideoTaskMetadata
      • VideoTaskError
      • GeminiResponse
      • OpenAIVideo
      • OpenAIVideoError
    • Schemas
      • ApiResponse
      • PageInfo
      • User
      • Channel
      • Token
      • Redemption
      • Log
  1. Seedance-2.0视频

SeeDance2系列视频生成说明

SeeDance2系列视频生成说明#

必看#

当前已兼容以下 Seedance 2.0 输入场景:#

多模态模式:图片、视频、音频输入
多图参考模式
首帧模式
首尾帧模式
纯文生视频
官方文档
视频生成 https://www.volcengine.com/docs/82379/1520757?lang=zh
视频查询 https://www.volcengine.com/docs/82379/1521309?lang=zh
视频生成教程 https://www.volcengine.com/docs/82379/2298881?redirect=1&lang=zh
Doubao Seedance 2.0 系列教程 https://www.volcengine.com/docs/82379/2291680?lang=zh

支持的请求方式#

当前支持两种传参方式:
1.
顶层直接传火山官方参数
2.
将火山原始参数放入 metadata
推荐优先使用方式 1,和火山官方文档更接近。
说明:中文提示词不超过500字,英文提示词不超过1000词。

方式 1:推荐写法#

1. 文生视频#

{
  "model": "doubao-seedance-2-0-fast-260128",
  "prompt": "一个女孩跳舞",
  "resolution": "480p",
  "ratio": "9:16",
  "duration": 5,
  "generate_audio": true,
  "watermark": false
}

2. 单图参考 / 首帧模式#

{
  "model": "doubao-seedance-2-0-fast-260128",
  "prompt": "让人物自然眨眼并轻微转头",
  "content": [
    {
      "type": "image_url",
      "role": "first_frame",
      "image_url": {
        "url": "https://example.com/start.png"
      }
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "watermark": false
}

3. 首尾帧模式#

{
  "model": "doubao-seedance-2-0-260128",
  "prompt": "镜头平稳推进,人物表情自然变化",
  "content": [
    {
      "type": "image_url",
      "role": "first_frame",
      "image_url": {
        "url": "https://example.com/start.png"
      }
    },
    {
      "type": "image_url",
      "role": "last_frame",
      "image_url": {
        "url": "https://example.com/end.png"
      }
    }
  ],
  "return_last_frame": true,
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "watermark": false
}

4. 多图参考模式#

{
  "model": "doubao-seedance-2-0-260128",
  "prompt": "融合多张参考图中的人物与场景气质,生成电影感短视频",
  "content": [
    {
      "type": "image_url",
      "role": "reference_image",
      "image_url": {
        "url": "https://example.com/ref1.png"
      }
    },
    {
      "type": "image_url",
      "role": "reference_image",
      "image_url": {
        "url": "https://example.com/ref2.png"
      }
    },
    {
      "type": "image_url",
      "role": "reference_image",
      "image_url": {
        "url": "https://example.com/ref3.png"
      }
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5
}

5. 多模态模式:图 + 视频 + 音频参考#

{
  "model": "doubao-seedance-2-0-260128",
  "prompt": "保持人物主体一致,参考视频运动节奏和参考音频氛围",
  "content": [
    {
      "type": "image_url",
      "role": "reference_image",
      "image_url": {
        "url": "https://example.com/portrait.png"
      }
    },
    {
      "type": "video_url",
      "role": "reference_video",
      "video_url": {
        "url": "https://example.com/reference.mp4"
      }
    },
    {
      "type": "audio_url",
      "role": "reference_audio",
      "audio_url": {
        "url": "https://example.com/reference.mp3"
      }
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "generate_audio": false,
  "watermark": false
}

方式 2:兼容写法#

如果你希望完全把火山原始参数包进 metadata,也支持:
{
  "model": "doubao-seedance-2-0-fast-260128",
  "prompt": "一个女孩跳舞",
  "metadata": {
    "content": [
      {
        "type": "image_url",
        "role": "reference_image",
        "image_url": {
          "url": "https://example.com/ref.png"
        }
      }
    ],
    "resolution": "480p",
    "generate_audio": true,
    "ratio": "9:16",
    "duration": 5,
    "watermark": false
  }
}

已兼容的主要参数#

当前已接入并会透传到火山上游的主要字段如下:
model
prompt
content
safety_identifier
return_last_frame
service_tier
execution_expires_after
generate_audio
resolution
ratio
duration
seed
camera_fixed
watermark
content 中已支持的素材类型:
text
image_url
video_url
audio_url
content[].role 当前不会被网关拦截,会按请求值透传。常见可用值包括:
reference_image
reference_video
reference_audio
first_frame
last_frame

3. 顶层参数与 metadata 同时传时#

当前兼容逻辑下:
metadata 中已有的同名字段优先
顶层字段用于补全 metadata 中缺失的同名字段
因此不建议同一个参数在顶层和 metadata 里传两份不同值。

当前推荐#

如果你在业务侧直接接 Seedance 2.0,建议统一采用:
顶层 content
顶层 resolution / ratio / duration / generate_audio / watermark
需要图片参考时使用标准 image_url
需要首尾帧时显式使用 first_frame / last_frame
修改于 2026-05-23 15:25:16
上一页
图片编辑
下一页
文生视频
Built with