C2PA regions of interest: mark only the AI segment of a video

The C2PA implementation guide covers regions of interest for audio and video segments, and the 2.3 spec allows time ranges. What that enables for mixed footage.

4 min readSume
All posts

Yes, C2PA can point at a part of a video instead of the whole file. The implementation guide lists Regions of Interest for audio and video segments as one of its five areas, and the 2.3 specification page says regions can use RFC 3339 time ranges. That suits a cut with real footage plus a few generated shots.

Why it matters for mixed cuts

Most real edits are not all-AI. A product demo may be camera footage with two generated B-roll shots. A single label for the whole file overstates or understates. A region lets a claim apply to the seconds where it is true.

Regions of interest, read 2026-10-02
ItemSource says
Applies toAudio and video segments
Time formatRFC 3339 time ranges (2.3 spec)
Listed inC2PA implementation guide, five areas

Keep the timeline you need

To mark a region later you need to know where each generated shot sits. Sume Timeline 1.0 takes video[] slots with a declared start and duration, and the docs say declared starts are authoritative. Keep that timeline JSON, and you can list which seconds were generated.

Timeline 1.0 is $0.10 per ceil(output minute). A 95-second render bills 2 minutes, so $0.20. See Timeline 1.0.

Caveat

The Sume docs do not say Sume writes a manifest or regions. This is a record-keeping method on your side, and any signer you use must support the fields.

Sources

Related posts

More in Developers

All Developers posts

Written by Sume