Cloudinary video trimming: percent or seconds vs Sume video trim

Cloudinary lets you trim by setting a start time, an end time, or a duration, in seconds or as a percentage of the asset's length. Sume's video trim takes seconds only: start plus exactly one of end or duration. If you think in percentages, read the clip length first and convert before you call.

The Cloudinary facts here come from its trimming and concatenating overview, read 2026-10-02. That page links to a separate Video trimming reference for the exact parameter syntax, which this post did not read, so it does not quote Cloudinary parameter names. Sume's side comes from the video trim docs.

What does the Cloudinary overview say?

It says a trim keeps the section you specify, set by a start time, an end time or a duration, and that each can be given in seconds or as a percentage of the asset length. It treats concatenation separately: the splice flag joins media assets into one instead of placing the added asset as an overlay, and transition effects such as a cross fade are a third topic.

How do I trim by percentage on Sume?

There is no percentage field. Probe the clip with video inspect (frames: false skips the stills), take its length in seconds, and multiply. For a 60-second clip, the middle 50% is start: 15 and duration: 30. Pass exactly one of end or duration; both is video_trim_range_conflict.

duration runs from 0.2 to 900 seconds. An end past the source clamps, with a trim_clamped_to_source warning in the result.

curl -X POST https://api.sume.com/v1/video-trim \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: middle-half-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4",
    "start": 15,
    "duration": 30
  }'

Does Sume re-encode the cut?

By default yes: precision: "exact" is a frame-accurate re-encode. precision: "keyframe" copies the stream, and the cut may begin up to one GOP early, so read actual_start_seconds from the result. output (width, height, fps) works only with exact. Dropping audio is audio: "drop".