Seedance 2.5 green screen editing: what Sume can do
ByteDance says Seedance 2.5 improves green screen editing. Sume's docs document an edit mode only for Gemini Omni Flash 1.1; Seedance takes video references.

Sume's docs do not describe a green screen edit mode for seedance-2.5. What they do document is a video edit mode for gemini-omni-flash-1.1 through video_url, and video references that the Seedance 2.x models honor. Seed's own post is where the green screen claim comes from.
Seed's statements are from its July 31, 2026 Seedance 2.5 post; Sume's from Video generation and Video Router, read 2026-10-01.
What does Seed say green screen editing does?
The post says Seedance 2.5 enhances editing features such as green screen, camera perspective and reference-based editing, with timestamp-level control for targeted editing of audio and video. In its green screen example the model can replace backgrounds and tell different stories while keeping the main subject intact, and render how clothes, hair and lighting respond to the new environment. Its example prompt references a video as @Video 1 and assigns scenes to time ranges such as 0–4s.
Which Sume path matches an edit?
Two paths, with different guarantees.
| Path | Model | What the docs say |
|---|---|---|
| Edit mode | gemini-omni-flash-1.1 | Exposed through the Video Router video_url field; 3–10 seconds |
| Video references | Seedance 2.x, Wan 3.0, MiniMax H3 | Audio and video references are honored |
| Long clips | seedance-2.5 | 4–30 seconds at 480p/720p/1080p |
How does an edit request look on Sume?
For Omni, the prompt describes the edit and video_url is the source. The docs state that video_url is the edit source, not a reference: it cannot be combined with image_url, end_image_url or reference_*_urls. A background swap is therefore a prompt-level request to Omni, with no separate green screen parameter in the docs.
For Seedance, you would pass your footage as a video reference and describe the new scene in the prompt. Whether the result matches Seed's green screen behavior is not something Sume's docs promise, so test on your own footage before building on it.
What should I do for a background replacement job?
Try an Omni edit when you want an edit-in-place request; the docs list Omni clips at 3–10 seconds. Use seedance-2.5 with a video reference when you need up to 30 seconds. In both cases send an Idempotency-Key, and compare against Seedance 2.5 clay render and camera references for how reference inputs behave.
Sources
Related posts
More in Use cases
- AI workout video generator: Seedance 2.5, 4 to 30 seconds
For a workout demo, one Seedance 2.5 request runs 4 to 30 seconds on Sume, with video and audio references; most other catalog models stop at 15 seconds.
- Synthesia Avatar Builder credits: 14 per option, and Sume jobs
Synthesia Avatar Builder charges 14 credits per generated option. Sume creates an avatar with one job per request via POST /v1/avatar-1.0/generate.
- Synthesia bulk download of 20 videos vs Sume result URLs
Synthesia can bulk download up to 20 videos as .mp4 files. On Sume you list avatar-video resources and fetch each job's result for its media.sume.com URL.
- Synthesia CSV bulk personalization vs Sume bulk runs (1-100)
Synthesia maps CSV or XLSX columns to template variables. Sume bulk runs take a JSON array of 1 to 100 Format runs with a concurrency window of 1 to 16.
Written by Sume