Video

Seedance

ByteDance’s Seedance models: cinematic motion, native sound, and control from frames or reference images.

Which one to use

ModelPick it forPrices per secondLength
Seedance 2.5The best quality and longest clips, up to 1080p480p $0.135, 720p $0.305, 1080p $0.754 to 30 s
Seedance 2.0High quality at a lower price480p $0.097, 720p $0.199, 1080p $0.445, 4K $0.994 to 15 s
Seedance 2.0 FastQuicker turnarounds480p $0.074, 720p $0.1594 to 15 s
Seedance 2.0 MiniThe lowest price, for drafts and volume480p $0.047, 720p $0.104 to 15 s

All four take a prompt, a first and last frame, or up to 9 reference images, and make sound unless you set generate_audio to false. Seedance 2.5 and Seedance 2.0 also take reference videos and audio.

Prompting

  • Describe the motion, not just the scene: who moves, how, and what the camera does (“slow push in”, “handheld tracking shot”, “static wide shot”).
  • Name the light and mood: “overcast morning”, “neon at night”, “golden hour backlight”.
  • Keep one main action per clip. Long sequences come out better as several clips.
  • For speech, put the line in quotes and say who says it: the barista says "One flat white, coming up".
  • From a first frame, describe what happens next rather than what’s already in the picture.

Image-to-video

A first frame fixes the opening shot, and the video takes its shape. Set aspect_ratio to another shape and the frame is cut to that shape around its centre (a last frame too), so keep what matters near the middle, or crop the image yourself. Add a last frame to land on a specific image, for transitions and before-and-after shots.

Reference images

Up to 9 images of characters, products, places or a look. Refer to them in the prompt as @Image1, @Image2 and so on, in the order you sent them. The same references across several clips keep a character or product consistent.

Reference videos and audio

Seedance 2.5 and Seedance 2.0 also take reference_videos (a dance, a camera move, a product shot to match) and reference_audio (a voice, a song, sound effects), mentioned as @Video1 and @Audio1. Seedance 2.5 takes up to 10 of each (30 s of each in all) and audio on its own; Seedance 2.0 takes 3 of each (15 s each in all), with audio beside an image or video. Seedance 2.5 can also edit or extend a reference video: set reference_mode to edit or extend.

Request body: edit a video
{
"model": "seedance-2.5",
"reference_videos": ["https://example.com/street-scene.mp4"],
"reference_images": ["https://example.com/red-car.jpg"],
"reference_mode": "edit",
"prompt": "Replace the white car in @Video1 with the red car from @Image1.",
"resolution": "720p"
}

Reference video seconds are billed at part of a second’s price, and reference audio is free: see Reference videos and audio for the limits and prices.

Good to know

  • Most of the wait (2 to 6 minutes) is queueing; a 480p clip takes about as long as a 720p one.
  • Prompts that may show real people or protected characters can be declined (content_policy); declined videos are free.
  • Images that look like a photo of a real person, AI-made faces included, are usually declined as frames or references. A stylized or illustrated character works, or leave the image out and describe the person in the prompt. The error names the flagged image when the model says which.
  • Editing a video of a real person is usually declined too, though the same video works as a reference for a new one.
  • Videos are kept for 30 days; download the ones you want to keep.

Full API details: Video generation. Try it in the dashboard’s Playground, which shows the code for each video.

Questions, or something missing? Ask our support team and we’ll answer by email, usually within a minute.