Models
Dolly·Aug 8, 2026
Seedance 2.5
Generate flagship multimodal videos with Seedance 2.5.
Dolly·Aug 8, 2026
5–30sduration range
50reference files
2resolution tiers
1generated video
What it does
Direct Seedance 2.5 with a complete multimodal brief
This pinned workflow prepares text and up to the supported image, video, and audio references for Seedance 2.5. Protected real-person and known-IP references are registered before generation when the model policy requires it.
- 1Generate clips from 5 to 30 seconds
- 2Use compatible image, video, and audio references
- 3Register protected references before generation
Seedance 2.5 black hole showcase
A ten-second cinematic deep-space sequence created with Seedance 2.5
How it works
Generate with Seedance 2.5
1
Build the multimodal brief
Write the shot and add only the image, video, or audio references that should influence it.
You bring
Prompt and optional references
You get
Seedance 2.5 brief
2
Prepare protected assets
Register real-person or known-IP references when required by the model policy.
You bring
Selected reference media
You get
Generation-ready asset IDs
3
Generate and review
Choose duration and resolution, generate the clip, then review motion, continuity, and reference adherence.
You bring
Prepared Seedance request
You get
Seedance 2.5 video
Good to know
Seedance 2.5 questions
What durations and resolutions are supported?+
Drama.Land exposes integer durations from 5 to 30 seconds at 480p or 720p.
Which reference types can I add?+
The workflow supports compatible image, video, and audio references, subject to the model limits.
Why might a reference need registration?+
Real-person and known-IP references may need to be registered and replaced with asset IDs before generation.