One captioned 15-second clip for Bluesky, Threads and WhatsApp status

Make one 15-second muted-friendly clip with burned-in captions and post the same file on Bluesky, Threads and a WhatsApp status. Model options and cost.

3 min readSume
All posts

Burn the captions into the picture, and one 15-second file stays readable for viewers who watch with the sound off. A Wan 3.0 clip at 720p plus a cue-based caption job costs $2.075 on Sume. Because the text is in the pixels, you do not depend on any platform's caption sidecar support; check each platform's own page for size and length rules before you upload.

Why cues, not speech

A generated clip may be silent, and speech captions fail on a silent clip with caption_no_speech. Sending cues (text, start, end in seconds) burns authored text without speech-to-text, which is the documented path for silent clips.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: muted-15-cap" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/clip15.mp4",
    "style": "black-outline",
    "cues": [
      {"text": "New roast, Friday", "start": 0.5, "end": 4},
      {"text": "Small batch, 40 bags", "start": 5, "end": 10}
    ]
  }'

Model choice for the 15 seconds

Wan 3.0 and Seedance 2.5 reach 15 seconds, and the other catalog models have a 15-second ceiling, but Gemini Omni Flash 1.1 stops at 10.

Sume list price, 9:16, read 2026-10-08
ModelResolutionLengthClipWith caption job
wan-3.0480p15 s$0.9375$1.1375
wan-3.0720p15 s$1.875$2.075
seedance-2-mini720p15 s$2.835$3.035
seedance-2.5720p15 s$8.667$8.867
gemini-omni-flash-1.1720p10 s$1.25$1.45

One master, three posts

Export once, keep the original, and let each post reuse the same file. If a platform needs a trim, POST /v1/video-trim returns a new MP4 for $0.02 without touching the master.

Keep the text readable on a phone

Burned captions need room. Keep each cue under about six words, hold it for at least a second and a half, and leave the lower third free of other graphics so the text is not covered by a platform's own buttons. If the clip has speech, skip cues and let the caption job time the words from the audio; if it is silent, your cues are the only text, so proofread them. Because the caption is a separate render, you can produce an English and a second-language version of the same master by running two caption jobs against the same source.

Related posts

More in Use cases

All Use cases posts

Written by Sume