Draft a Short at 720x1280 in Timeline, then render 1080x1920

To review a vertical cut before the final render, send the same Timeline 1.0 document with output: {"width": 720, "height": 1280}, then drop the output block for the final. Timeline's default is 1080x1920. The docs price the render at $0.10 per output minute, rounded up, with no size term, so a smaller frame does not change the rate; what you save is review time and file size. Confirm the live rate in GET /v1/catalog.

What the output block accepts

output.width and output.height are even integers from 256 to 2160. output.fps is 24, 25, 30 or 60 and should be left out unless you need to force a rate, because a rate that differs from the sources makes the job repeat or drop frames and report output_fps_resamples_sources.

Plan first, render second

POST /v1/timeline-1.0/plan compiles the document and returns duration_seconds, segment_count, billable_minutes and an estimate. It creates no job, reserves nothing and downloads nothing, and it does not need an Idempotency-Key. Use it to confirm that a 58-second cut bills as one minute while a 61-second cut bills as two.

Import the clip first with POST /v1/media-imports and send an Idempotency-Key on writes.

curl -X POST https://api.sume.com/v1/timeline-1.0/plan \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "audio": {"url": "https://media.sume.com/artifacts/artf_demo/voice.wav", "duration_seconds": 58},
    "video": [{"source_url": "https://media.sume.com/artifacts/artf_demo/a.mp4", "start": 0, "duration": 58}],
    "output": {"width": 720, "height": 1280}
  }'

Keep the limits in view

YouTube's help page says a Short can be up to 3 minutes and is limited to 1080p (YouTube Help, read 2026-10-05), so 1080x1920 is already the ceiling; there is nothing to gain from rendering larger. Timeline's own cap is 1800 seconds of audio per render, far above a Short.

What a draft tells you, and what it does not

A 720x1280 frame has 921,600 pixels, which is 44% of the 2,073,600 in 1080x1920. The cut rhythm, the order of clips, the transitions and the fit mode all look the same, so judge them on the draft. Sharpness, caption legibility and fine edges only show at full size, so judge those on the final.

Because the draft and the final are different documents (the output block differs), send the final with its own Idempotency-Key. The draft's files are separate artifacts; neither replaces the other, and the source clips are unchanged. Remember that the draft is a real render, billed at the same $0.10 per started output minute, so use POST /v1/timeline-1.0/plan when you only need to confirm that the document compiles.

Using the draft in a review loop

One practical loop is: plan, draft, review, adjust the document, plan again, and render the final once. The plan call is unbilled and tells you the duration, the segment count and the billable minutes. If a Short is 61 seconds, the plan shows two billable minutes, and trimming a single second from the last slot brings it back to one. That is a small saving per clip, but it adds up over a weekly batch, and it is visible before you spend anything.

Keep the draft's video[].fit the same as the final. A draft that uses cover and a final that uses blur would look different in ways the review cannot predict.

One more check on size: both 720 and 1280 are even integers inside the 256 to 2160 range the docs allow, so the draft is a valid output block. A size such as 721x1281 would be refused because Timeline only accepts even values, which keeps the yuv420p encode safe. If you ever need a different draft shape, such as 540x960, the same 9:16 ratio holds and the numbers are also even.