LinkedIn CTV ad video specs: 1080p, 6-90 s, -23 LUFS audio

A LinkedIn Connected TV (CTV) ad is a 16:9 MP4 at 1920 x 1080 (1280 x 720 is also accepted), H.264, 6 to 90 seconds long, up to 500 MB, with 2-channel audio at -23 integrated LUFS. That comes from LinkedIn's CTV Ads specifications help page, read on 2026-10-03. It is a separate spec from LinkedIn's regular video ads, so a clip that passes in the feed can still fail here.

Two numbers on that page are easy to miss. The first is the loudness target: -23 integrated LUFS is a broadcast-style level, so a clip mixed for the feed may need its level changed. The second is the length list: 6, 15, 30, 45, 60, 75 and 90 seconds, with 6, 15, 30 and 60 recommended and a note that some inventory might not accept the 75 and 90 second formats. The page names no other lengths, so plan to cut to one of those.

The spec, row by row

Every row below is from the CTV Ads specifications page. The only change is that the page lists a general value and a recommended value for some rows; both are shown.

What Sume can make for this spec

Sume covers the picture and the cut. For generation, seedance-2.5 accepts 4 to 30 second clips at 480p, 720p or 1080p, according to the video generation docs. Ask for 1080p and 16:9, and read supported_aspect_ratios for the model in GET /v1/video-router/models before you submit, because limits are set per model. A 6, 15 or 30 second CTV spot fits inside one generation; a 45 or 60 second spot needs two or more clips joined.

Joining is Timeline 1.0's job. Its output accepts even width and height between 256 and 2160 and an fps of 24, 25, 30 or 60, so 1920 x 1080 at 24, 25 or 30 fps lands inside LinkedIn's frame-rate list. Timeline needs an audio spine (audio.url, audio.parts[] or audio.mode: "silence") with duration_seconds up to 1800, and it is billed at $0.10 per output minute, rounded up. Trim a single clip to a CTV length with video trim (duration between 0.2 and 900 seconds, $0.02 per job, precision: "exact" by default).

curl -X POST https://api.sume.com/v1/timeline-1.0/render \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: ctv-30s-001" \
  -d '{
    "audio": { "url": "https://media.sume.com/artifacts/artf_demo/vo.wav", "duration_seconds": 30 },
    "video": [
      { "source_url": "https://media.sume.com/artifacts/artf_demo/a.mp4", "start": 0, "duration": 15 },
      { "source_url": "https://media.sume.com/artifacts/artf_demo/b.mp4", "start": 15, "duration": 15,
        "transition": { "type": "fade", "duration": 0.25 } }
    ],
    "output": { "width": 1920, "height": 1080, "fps": 30 }
  }'

What Sume does not do for CTV

Sume does not expose a loudness measurement or a normalizer, and it does not expose bit rate or codec settings: the video endpoints reject codec, crf and similar ffmpeg fields with ffmpeg_fields_rejected. Timeline offers audio.gain_db from -60 to 12 for a flat level change, but a gain move is not a loudness target. The earlier post on podcast loudness covers the same gap.

So treat -23 LUFS, the 48 kHz sample rate and the bit-rate floor as checks you run on the finished file. Pull the audio with audio detach (format: "wav", sample_rate: 48000), measure it in a loudness meter, adjust audio.gain_db, and render again. Use video inspect with frames: false to read probe facts and confirm the clip has audio before you spend anything on a detach.

A short checklist before upload

A spec page changes without notice, so re-read the LinkedIn page the day you upload.

• Length is exactly 6, 15, 30, 45 or 60 seconds (75 and 90 only if your inventory takes them).
• The audio track is as long as the video, 2-channel, and measured near -23 LUFS.
• The frame rate is one of the listed rates, and the export is MP4 with H.264.
• The file is under 500 MB and 1920 x 1080 or 1280 x 720.
• The campaign meets the setup rules in LinkedIn CTV campaign setup.