Product drop teasers at T-7, T-3 and T-0 from one master: $0.70

Three teasers cut from one master for a Nov 6 drop: trim, dim the first two, burn authored cues. Sume media jobs total $0.70, with no speech-to-text.

5 min readSume
All posts

To tease a product drop three times from one master video, cut three different lengths with video trim, dim the two early teasers so they read as dark and unfinished, and burn the date lines as authored caption cues. For a drop on Friday, Nov 6, 2026 that means teasers on Oct 30, Nov 3 and Nov 6, and the Sume media jobs total $0.70.

This post assumes you already have the master clip. It prices only the edit, using the Sume trim, filter and caption pages read on 2026-10-03. The dim amounts and cue wording are design choices, not documented values.

The three cuts

Oct 30 is a Friday, Nov 3 is a Tuesday and Nov 6 is a Friday, so the offsets are seven days, three days and zero. Each teaser reveals more of the master than the last: six seconds of the opening, ten seconds from a third of the way in, and fifteen seconds of the full reveal.

Trim is a range cut with start and exactly one of end or duration, for example start: 0, duration: 6. The default precision is exact, a frame-accurate re-encode, which matters when a caption cue must land on a specific beat. The dim op multiplies luma by an amount in (0, 1], where 0.45 is darker and 1 is unchanged, and the docs name it as the step before captions on bright footage.

Teaser pack for a Nov 6 drop, with Sume media-job rates (read 2026-10-03)
TeaserPost dateTrimDim (amount, my choice)CuesJob cost
T-7Fri Oct 30start 0, duration 60.55Drop in 7 days$0.02 + $0.02 + $0.20 = $0.24
T-3Tue Nov 3start 4, duration 100.753 days to go$0.02 + $0.02 + $0.20 = $0.24
T-0Fri Nov 6start 0, duration 15noneLive now$0.02 + $0.20 = $0.22
Pack$0.70

Captions without transcription

A teaser with no spoken line has nothing for speech-to-text to hear, and the caption page says a silent clip fails as caption_no_speech, with next_action: use_overlay_captions. Passing cues (or segments) with text, start and end skips ASR and burns exactly that copy at those times.

A standalone caption job is $0.20 for videos up to 60 seconds, and all three teasers are well under that. script_text, words, cues and segments are mutually exclusive, so send one of them. The input must be a fetchable public HTTPS video URL; the docs show a media.sume.com artifact URL as a valid example.

{
  "video_url": "https://media.sume.com/artifacts/artf_demo/teaser-t7.mp4",
  "style": "slam",
  "cues": [
    { "text": "DROP IN 7 DAYS", "start": 0.5, "end": 3.0 },
    { "text": "NOV 6", "start": 3.2, "end": 5.8 }
  ]
}

Order and checks

Run trim, then the dim filter, then captions, so the dim is applied to the footage and not to the burned text. Use the free filter check before paying for the encode. Each job needs its own Idempotency-Key; derive it from the drop and teaser, such as drop-nov6-t7-trim, so a retry replays instead of billing twice.

If the drop slips a day, you re-run only the caption jobs: the trims and dims are reusable artifacts, and a caption restyle that supplies a source_caption_id reuses the source video and timings. A new date line means new cues, which is a new $0.20 job per teaser, so a one-day slip costs at most $0.60.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume