One Seedance 2.5 clip in 9:16 and 16:9: Timeline fit blur or cover
Re-generate for each ratio, or reframe a finished clip? Timeline 1.0 fit modes cover, contain, stretch and blur, with output sizes, on Sume.

If one Seedance 2.5 clip has to ship as both 9:16 and 16:9, you have two options on Sume: generate it twice with a different aspect_ratio, or reframe the finished clip through Timeline 1.0. Reframing is a fraction of the cost of generating again, but it crops or pads; a fresh generation composes for the new frame. For a quick cutdown where the subject sits near the center, reframing usually works. For wide establishing shots, regenerate.
The two routes
The ratios in the first row of the last line come from the video docs' Seedance 2.0 catalog example; check supported_aspect_ratios for seedance-2.5 itself.
| Question | Generate each ratio | Reframe with Timeline |
|---|---|---|
| Composition fits the new frame | Yes, the model composes for it | Only what the crop or fit leaves |
| Same take in both | No, separate generations | Yes, same pixels |
| Cost | A generation per ratio | $0.10 per output minute, rounded up |
| Time | Model queue and render | Worker ffmpeg |
| Seedance 2.5 ratios | Six listed in the docs for Seedance models, such as 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | Output width and height are your choice, even integers 256 to 2160 |
Fit modes
Each video[] slot takes a fit value: cover (default) fills the frame and crops the overflow; contain keeps the whole picture and pads; stretch distorts to fill; blur fills the padding with a blurred copy of the clip. For a 16:9 clip going to 9:16, blur keeps the whole picture visible over a soft background, and cover crops to the center.
- Subject is centered and the sides do not matter:
cover. - Whole composition matters, such as a product beside text:
blurorcontain. - Never use
stretchfor people or products.
A vertical reframe
output.width and output.height must be even integers between 256 and 2160. The default output is 1080 by 1920. If you omit output.fps, the render follows the source rate, which avoids the judder the docs warn about when a rate is resampled.
curl -X POST https://api.sume.com/v1/timeline-1.0/render \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: reframe-9x16-001" \
-d '{
"audio": { "mode": "silence", "duration_seconds": 30 },
"video": [
{ "source_url": "https://media.sume.com/artifacts/artf_demo/wide.mp4", "start": 0, "duration": 30, "fit": "blur" }
],
"output": { "width": 1080, "height": 1920 }
}'Safe areas and text
Reframing breaks anything that sits near the edges. If your original has captions, logos or a price badge in a corner, a cover fit may crop it off. Burn text in after reframing, not before, or keep key elements near the center during generation. The compose route can overlay a still plate on a clip, so a logo or date plate can be added after the reframe.
Check the result at the final size: a clip that looks fine at 1920 wide may be unreadable at 1080 by 1920 on a phone.
- Add text after the reframe.
- Keep subjects near the center when you know a cutdown is coming.
- Review at the final size.
A decision rule
If the ratios are close, such as 16:9 to 4:3, reframe. If the ratios are opposite, 16:9 to 9:16, reframe with blur for a social cutdown, but regenerate for a hero placement where the subject must fill the frame. Reframing a wide shot to vertical with cover shows only the middle third, which is rarely the best third.
For a 16:9 output from a vertical clip, set width 1920 and height 1080 and use blur the same way. Use the unbilled plan call first to confirm the compile. Seed's launch page (read 2026-10-05) is the vendor reference for the model itself.
Sources
Related posts
More in Use cases
- Release notes to video: three features, three shots, one Wan 3.0 clip
Turn a changelog into a 30 s Wan 3.0 video on Sume: one shot per feature, UI screenshots as references, exact labels burned as captions. Prompt and request.
- Relight a product shot with Ideogram 4.5: 3 lighting passes, cost
Three lighting variants of one product photo through POST /v1/images with Ideogram 4.5 cost $0.1125 at low and $0.225 at medium on Sume. Prompts and checks.
- One 24-second Seedance 2.5 clip for a TikTok App Bundle ad
TikTok recommends 21 to 30 seconds for App Bundle video. One seedance-2.5 call can cover it at 9:16. Request body, checks, and where captions fit.
- Renovation reveal: Omni first frame before photo, last frame after
Turn a before photo and an after photo of a room into a 10-second Gemini Omni 1.1 Flash reveal on Sume with image_url and end_image_url. Request and cost math.
Written by Sume