LTX retake vs Sume: fix one section of a video
LTX's POST /v1/retake regenerates one time region of a video. Sume's video edit takes a whole clip, so trim the bad section, edit it, and rejoin it.

LTX's POST /v1/retake regenerates a specific time region of an existing video and leaves the rest alone. Sume has no region parameter: its video edit takes one whole source clip as video_url, and the prompt describes the change. To fix one section on Sume, cut that section out, edit the short clip, and join it back.
LTX details are from its longer AI videos post; Sume details from the Video Router docs, read 2026-10-01.
What does LTX retake do?
The LTX post calls retake the repair tool rather than a lengthening tool. On the API it is POST /v1/retake and runs on ltx-2-3-pro only. Its stated benefit: when segment four of six is wrong, rerunning from there changes everything downstream, while retaking the bad region leaves the surrounding material intact.
What does Sume's edit take instead?
Sume's Video Router routes gemini-omni-flash-1.1 by request shape. A request with video_url is an edit (video_to_video): the prompt describes the edit, resolution is optional, and aspect_ratio and duration are not sent. The docs say the output follows the source clip. There is no start/end field, so the whole clip you pass is the thing that gets edited.
| Question | LTX retake | Sume edit |
|---|---|---|
| Endpoint | POST /v1/retake | POST /v1/video-router/generate |
| Scope of the change | A specific time region | The whole video_url clip |
| Model | ltx-2-3-pro on the API | gemini-omni-flash-1.1 |
| Describe the change | Per the LTX workflow | prompt |
How do I fix one section on Sume?
Use three steps, each a documented Sume call. First, import the video and cut the bad range with Video trim: POST /v1/video-trim takes video_url, start, and exactly one of end or duration, and returns a new MP4 with only that range. Second, send that short clip to the edit route with a prompt describing the fix. Third, assemble the pieces in order with Timeline 1.0. Trim the good ranges before and after the same way.
Because the join is a normal timeline, the seam is yours to check. For the assembly side see assemble long-form video with the Timeline API.
curl -X POST https://api.sume.com/v1/video-router/generate \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: fix-section-001" \
-d '{
"model": "gemini-omni-flash-1.1",
"prompt": "Replace the bottle with an apple. Keep everything else the same.",
"video_url": "https://example.com/clip.mp4",
"resolution": "720p",
"mode": "async"
}'What should I watch for?
LTX says retake leaves the surrounding material intact. On Sume the edited piece and its neighbors are separate clips, so check the cut points after the join. The edit applies to the whole short clip you send, so keep the trimmed piece tight around the problem. For a similar whole-clip edit comparison, see Luma Ray 3.2 modify video.
Sources
Related posts
More in Developers
- LTX frame counts must be 8k+1; Sume durations are whole seconds
LTX frame counts must satisfy (F-1) % 8 == 0, so 30 frames is invalid. On Sume you send whole seconds and read the allowed set from the catalog, not frames.
- Luma API 429 requests per minute: sliding window vs Sume
Luma counts requests in a sliding 60-second window and returns 429 if RPM or concurrent jobs fails. Sume returns 429 rate_limited: back off, reuse the key.
- Luma API concurrent jobs limit vs Sume plan concurrency
Luma caps active generations per API client and answers 429 when full. Sume ties concurrency to your plan and queues extra jobs until queue_full.
- Luma Agents API presigned URL, generation id and Sume job ids
Luma presigned video URLs expire after 1 hour but the generation id does not. For Sume, store the job id and re-read the job instead of keeping a video URL.
Written by Sume