Skip to content
Lorca Labs
Live demo
Integrity

Media Provenance

Attribution that survives re-encoding — the model, session, and prompt behind any generated image, verified from the pixels.

The problem

Synthetic media is proliferating, and consumers — internal and external — need to know what generated it. Metadata is stripped on upload; pixel-level provenance survives.

What it does

01

Embeds a model + session + prompt-hash strip into an image carrier.

02

Verifies the strip from the pixels with an integrity check.

03

Lists and retrieves provenance records.

04

Revokes a record with a reason — verification then reports it.

Who it’s for

  • Teams shipping AI-generated media
  • Platforms that need to attribute generated content
  • Investigators verifying media origin

In context

The pixel-level sibling of IP Watermarking — attribution for generated content, watermarking for proprietary artifacts.

Spec & limits

Interface Input limits Output artifact Failure mode Non-claim
Studio ephemeral /demo; pilot durable /v1 on your infrastructure. TBD — verified in pilot acceptance TBD — verified in pilot acceptance TBD — verified in pilot acceptance TBD — verified in pilot acceptance

Related guide

See the media provenance guide for a deeper problem walkthrough and verification notes — evaluation notes for the live demo and for a private pilot. Also: try the live experiment and read the product overview.

Frequently asked questions

/ 05
01 Where does the provenance data actually live — in the file or on a server?

The provenance strip is embedded in the media itself (pixels/metadata) so it travels with the file — verification needs no external lookup.

02 Does this embed C2PA-standard metadata, or your own format?

It is an embedded provenance strip verified on your infrastructure — not a certified standard or badge.

03 Can provenance be verified offline / on my own infrastructure?

Yes — the strip is designed to be verified without contacting a service; the studio demo runs the same verifier on your browser or your host.

04 Does it work for video, or only images?

The embedding works on raster images and per-frame keyframes for video; the verification report covers the embedded origin metadata.

05 What happens if the media is heavily re-encoded before verification?

Mild re-encoding preserves the strip; heavy recompression may reduce recoverability — the verifier reports integrity rather than silently failing.

Try it live in the studio

A no-signup, in-browser experiment against an isolated demo host. Load sample data, run the flow, and see the result — nothing is stored.