创建视频任务¶

POSThttps://ai.shanhui.tv/v1/video/generations

提交文生视频、图生视频、首尾帧或多模态参考任务。接口为异步模式，成功后返回任务 ID。

请求头¶

Header	必填	值
`Authorization`	是	`Bearer sk-your-api-key`
`Content-Type`	是	`application/json`
`Accept`	否	`application/json`

请求体规则¶

扩展参数必须放在 metadata

model 与 prompt 放在顶层；content、duration、ratio、resolution、generate_audio、watermark 等扩展参数统一放入 metadata。把这些字段放错层级可能不会报错，但参数会被忽略。

{
  "model": "doubao-seedance-2-0-fast-260128",
  "prompt": "视频画面描述",
  "metadata": {
    "content": [],
    "duration": 5,
    "ratio": "16:9",
    "resolution": "720p",
    "generate_audio": true,
    "watermark": false
  }
}

文生视频示例¶

curl -X POST 'https://ai.shanhui.tv/v1/video/generations' \
  -H 'Authorization: Bearer sk-your-api-key' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "doubao-seedance-2-0-fast-260128",
    "prompt": "雨后的石板街，霓虹倒影，镜头贴近地面缓慢前移，电影感",
    "metadata": {
      "duration": 5,
      "ratio": "16:9",
      "resolution": "720p",
      "generate_audio": true,
      "watermark": false
    }
  }'

图生视频示例¶

curl -X POST 'https://ai.shanhui.tv/v1/video/generations' \
  -H 'Authorization: Bearer sk-your-api-key' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "doubao-seedance-2-0-260128",
    "prompt": "保持主体外观一致，人物轻轻转头，镜头缓慢推进",
    "metadata": {
      "content": [
        {
          "type": "image_url",
          "image_url": {"url": "https://example.com/first-frame.jpg"},
          "role": "first_frame"
        }
      ],
      "duration": 5,
      "ratio": "adaptive",
      "resolution": "720p",
      "generate_audio": false,
      "watermark": false
    }
  }'

成功响应¶

{
  "id": "task_xxxx",
  "task_id": "task_xxxx",
  "object": "video",
  "model": "doubao-seedance-2-0-fast-260128",
  "status": "queued",
  "progress": 0,
  "created_at": 1781001293
}

调用方应兼容 id 与 task_id，并优先保存非空值作为后续查询 ID。