HeyGen get video scenes API vs Sume video inspect stills

HeyGen's GET /v3/videos/{video_id}/scenes returns a video's scene-by-scene composition: background, placed elements and script. Sume has no equivalent typed-scene tree. POST /v1/video-inspect reads one hosted clip and returns probe facts, sampled stills and an optional transcript.

HeyGen text is from its developer changelog (August 2026 entry); Sume text from Video inspect. Both read 2026-10-01.

What does HeyGen's endpoint return?

Per scene, a background, a set of elements placed within the frame, and a script for the audio, plus video-level context such as title, aspect ratio, resolution and captions. It works for any video in your HeyGen workspace and describes the video as it currently stands.

What does Sume's video inspect return?

The docs say it plainly: "Not typed scenes". The clip must already be a media.sume.com file owned by your workspace, and inspect never re-encodes it or makes a new MP4.

How do I call it?

A minimal request, as in the docs:

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

Which should I use to understand a finished video?

If you need to know what a cut contains, inspect gives you stills you can look at and an audio transcript to read; you or a model do the interpretation. If you need the structure of a video built on HeyGen, only HeyGen's endpoint has it. Cost framing for ad breakdowns is in UGC ad analysis versus inspect cost. The MCP tool is video_inspect, listed in MCP tools and gates.