Recipe reel from 6 step clips: video trim exact or keyframe, $0.12
Cut six cooking steps from one kitchen take with Sume video trim: exact re-encode or keyframe copy, $0.02 per cut, then one $0.10 Timeline render. Total $0.22.

To cut a recipe reel from one long kitchen take, run one Sume video trim per step at $0.02 each, then assemble the six cuts in one Timeline render at $0.10. Six cuts cost $0.12, the render $0.10, so the reel costs $0.22 before any voice-over or captions. The one choice that matters is precision: exact is frame-accurate, and keyframe is faster but can start early.
Exact or keyframe
Video trim takes one media.sume.com clip, a start, and exactly one of end or duration. It returns a new MP4 with only [start, end). The source does not change.
With precision: "exact" (the default), the cut is a frame-accurate re-encode (libx264, yuv420p). With keyframe, the job copies the stream, and the cut can start a GOP early. The result then reports actual_start_seconds, and you re-base your times against it.
| Mode | What the job does | Use for a recipe step when |
|---|---|---|
exact (default) | Frame-accurate re-encode, AAC audio kept as remuxed | The cut must begin on the exact frame, such as the knife hitting the board |
keyframe | Stream copy; may start a GOP early | A little extra lead-in is fine and you want no re-encode |
output conform | Width, height, fps (24, 25, 30, 60); exact only | All six clips must share one frame rate before assembly |
Six cuts, one request body each
Each step is its own job with its own Idempotency-Key. Use a key that names the recipe and the step, such as soup-v1-step3, so a retry does not queue a second cut. A step is at least 0.2 seconds and at most 900 seconds. The source can be up to 1800 seconds, so a half-hour cooking take fits.
{
"video_url": "https://media.sume.com/artifacts/artf_take/soup.mp4",
"start": 41.5,
"duration": 6,
"precision": "exact",
"audio": "keep",
"output": { "width": 1080, "height": 1920, "fps": 30 }
}Why conform the frame rate
Timeline renders at the rate of the sources when you omit output.fps. If sources differ from the output rate, the job repeats or drops a frame every few frames, which shows as judder, and the job reports output_fps_resamples_sources. Conforming each trim to 30 fps with the output block avoids that warning when the steps come from different cameras.
The cost of the whole reel
Prices are the public rates in the docs: $0.02 per trim job and $0.10 per ceil(output minute) for Timeline. Confirm live rates in GET /v1/catalog.
| Item | Count | Rate | Subtotal |
|---|---|---|---|
| Video trim | 6 | $0.02 | $0.12 |
| Timeline render (45 s reel, 1 minute) | 1 | $0.10 | $0.10 |
| Total | $0.22 |
Picking the cut points
Find the step boundaries before you spend anything on cuts. Video inspect probes one clip, returns sampled stills (up to 24 per call, for sources up to 1800 seconds), and can transcribe on request. A spoken method with phrases like "now add the stock" gives you timestamps for each step from the transcript, and the stills let you confirm the frame. Then write the six start and duration pairs yourself.
Keep each step short. A six-second cut per step gives a 36-second reel, which fits one Timeline minute with room for a title card. If a step needs more time, trim it longer and shorten another, because the render bills by the ceiling of the output minute, not by the number of cuts.
Where the six cuts go next
Each result has a new artf_ video_url. Put those into video[] with source_in 0. If the reel runs past 60 seconds, the render bills the next minute, so 61 seconds is $0.20. Add a spoken method with a separate voice job, and add captions only on the final cut, since a standalone caption job is $0.20 per clip up to 60 seconds.
Sources
Related posts
More in Media tools
- Remove backgrounds from 500 product photos: RMBG API cost and code
Sume RMBG 1.0 removes a background for $0.0225 per image regardless of size. Batch cost for 50 to 5,000 photos, the request body and a Python loop.
- Restyle captions four ways for 80 cents without transcribing twice
Each Sume caption render is $0.20, so four styles of one clip cost $0.80. A restyle with source_caption_id reuses the word timings and skips re-transcribing.
- Restyle captions three times with source_caption_id: $0.60 total
To test three caption looks on one video, send source_caption_id: no second transcription, but each restyle is a render at the same job price.
- script_alignment_mismatch: fixing a caption job with script_text
When script_text does not match the speech, a Sume caption job fails with script_alignment_mismatch or script_alignment_failed. Simplify the script or omit it.
Written by Sume