Video model · ByteDance

Seedance 2.5

Cinematic image-to-video with fluid, natural motion

Seedance 2.5 turns a still image (or a text brief) into a short cinematic video clip with smooth, physically plausible motion. It is the strongest pick when you want quality motion from a reference frame.

This Dovoo workflow is not available yet

The model remains in the decision guide; use the Seedance 2.0 workflow for generation today.

What Seedance 2.5 is best for

Image to video

Animate a reference frame with camera and subject motion that respects the source composition.

Cinematic motion

Smooth, film-like movement with stable temporal coherence.

Camera control

Describe pans, zooms, and dolly moves to direct the shot.

Text to video

Generate short clips from a prompt when no reference frame exists.

Input modes

  • Image to video

    Provide a reference frame and describe the motion.

  • Text to video

    Describe the scene and motion directly.

Key parameters

Duration
5s / 10s(Longer durations cost more and need stable motion.)
Resolution
720p / 1080p(Match the platform export.)
Aspect ratio
16:9, 9:16, 1:1
Motion direction
camera + subject(Separate camera moves from subject action.)

Prompt guide for Seedance 2.5

  1. 1Separate camera motion (slow dolly-in) from subject motion (hair flowing) in the prompt.
  2. 2Keep the first frame strong; image-to-video quality depends heavily on the reference.
  3. 3Avoid requesting impossible physics; plausible motion stays stable longer.

Real outputs

Seedance 2.5 examples

Browse all prompts on the gallery →

Seedance 2.5 FAQ

Seedance 2.5 vs 2.0 — which should I use?

Seedance 2.5 produces smoother, more cinematic motion and better camera control. Use 2.0 for lighter, faster, lower-cost clips.

Image-to-video or text-to-video?

Image-to-video gives more stable, controllable results when you have a good reference frame. Use text-to-video to explore from scratch.

How should I prompt camera motion in Seedance 2.5?

Describe camera movement separately from one plausible subject action, then add only the secondary motion needed by the shot.

Canonical: https://modelpromptlab.com/models/seedance-2-5

Adults only

Want uncensored prompts, chat, and generation?

Everything on this site stays safe-for-work. For no-filter NSFW AI chat and image / video generation, continue on Stella AI.

No-filter AI chat, image, and video generation on Stella AI.