Seedance 2 Mini
Commercial usebytedance/seedance-2-miniGeneration input
Text prompt (max 30000 characters). Use @image1, @video1, @audio1 to refer to reference_* media by array order.
Output video resolution (480p or 720p).
Output video duration from 4 to 15 seconds.
Output aspect ratio. Use adaptive for single-frame or first/last-frame I2V.
Generate synchronized output audio for the video. Disabled by default; enable only with an original sound description or authorized reference audio. Independent of reference_audio_urls, which only provides an audio cue.
Reference mediaOptional
First-frame image URL. The platform uploads it to the virtual asset library before generation. Mutually exclusive with reference_* fields.
Last-frame image URL. The platform uploads it to the virtual asset library before generation. Requires first_frame_url.
Public https reference image URLs (max 9). Every image is uploaded to the virtual asset library before generation. Mutually exclusive with strict first/last frame mode.
Public https reference video URLs (max 3). Motion/camera/rhythm cues only; videos that contain real people may be rejected upstream. Use portrait images marked virtual/real for person consistency.
Public https reference audio URLs (max 3). Used as BGM/voice/sfx cues and mapped to @audioN slots.
Creative direction
Smart Director Mode
Add an image, first frame, video, or audio reference first.
More technical options
Structured reference images: [{ url }]. Alternative to reference_image_urls; every image uses virtual asset ingest.
Portrait URLs always uploaded to the virtual portrait library before generation.
Portrait URLs (including real-person photos) uploaded to the virtual portrait library before generation.
Ask the result to include the generated video's final frame when available.
Output container format. mp4 is recommended for most integrations.
Enable model-side web/search context when available. Keep false for normal video generation.