LinkedIn video thumbnail under 2 MB: pull the frame with Sume

LinkedIn's help page says a video ad thumbnail must be a JPG or PNG up to 2 MB. To make one from your own clip, call Sume's video frames endpoint with a timestamp, ask for jpeg, and check the file size before you upload.

The LinkedIn rule

The LinkedIn help page states the thumbnail limit.

Pull a frame

POST /v1/video-frames takes one media.sume.com clip from your workspace plus exactly one of at[] or fps. Import first with POST /v1/media-imports, because there is no open-internet fetch. Optional format is jpeg (the default) or png, and max_edge is 16 to 2160 and clamps the long edge; omit it to keep the source size. Submit always returns 202, so poll the job. Frames come back as durable image artifacts at frames[{t,url,width,height}].

JPEG is usually much smaller than PNG for photographic frames, and a smaller max_edge shrinks the file further.

curl -X POST https://api.sume.com/v1/video-frames \
  -H "x-api-key: $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: li-thumb-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/ad.mp4",
    "at": [1.5],
    "format": "jpeg",
    "max_edge": 1080
  }'

Pick the right moment

POST /v1/video-inspect with the default frames program returns 8 mid-bin stills, which is a quick way to scan a clip for a clean face or product shot. Then request an exact timestamp from video frames for the final thumbnail. Avoid frames with motion blur or a half-closed eye.

Check before you upload

Sume does not set a byte target, so check the downloaded file against the 2 MB limit. If it is over, lower max_edge and extract again. Frame extraction is billed by its Modal compute, so keep the number of tries small.