Generation Mode Examples

Official Seedance 2.0 content combinations for text, image, video, and audio references.

Seedance 2.0 official requests use content[] to combine text, image, video, and audio inputs. Use reference_image, reference_video, and reference_audio for media roles.

{
  "model": "doubao-seedance-2-0-260128",
  "content": [
    {
      "type": "text",
      "text": "A golden Shiba Inu runs under cherry blossoms as the camera slowly rises"
    }
  ],
  "resolution": "480p",
  "ratio": "16:9",
  "duration": 5,
  "watermark": false
}

For image, video, and audio references, add image_url, video_url, and audio_url content items with the matching role.