feat: Register LumaTrace (ID 26)

[Cyrah2R]

Registering LumaTrace algorithm (ID 26).

Description: Java-native Adaptive Spatial Watermarking robust against cropping. License: MIT Maintainer: @Cyrah2R

[jcollomosse]

Thanks for this contribution which will be reviewed at the next meeting of the watermarking task force (Feb 4). To expedite merging of this contribution please can you take the following actions:

To conform to the entry schema please provide an email address within the 'contact' field rather than a URL.

Please can you confirm your relationship to the lumatrace.org namespace?

[Cyrah2R]

@jcollomosse Thank you. I have updated the submission as requested:

1. Contact: Updated to contact@lumatrace.es (official maintainer email).
2. Namespace: Updated to es.lumatrace to match our official domain lumatrace.es.
   • I confirm that I am the owner of the lumatrace.es domain (corporate website currently under construction).

Ready for the Feb 4 meeting.

[domguinard]

Thanks for the entry and the changes @Cyrah2R, I added a review and will bring this to the group for our next meeting on Feb 4.

[Cyrah2R]

@jcollomosse @domguinard

Thank you for the approvals. I have just pushed a fix to address the description_trailing_period linting error by removing the periods from the description fields, ensuring full compliance with the repository's style guidelines.

[domguinard]

Thanks for fixing the description @Cyrah2R. The WM TF session was postponed to January 18, the team will review your entry then. Thanks for your patience.

[Cyrah2R]

Thanks for the update! I appreciate the heads-up.

I am standing by for the review. Just a note regarding the failing check: it seems to be flagging an issue in softbinding-algorithm-list.schema.json (specifically empty schemas at lines 140/153), which is a file I haven't touched. My submission only modifies the .json list.

Best regards.

[domguinard]

Thanks for the update! I appreciate the heads-up.

I am standing by for the review. Just a note regarding the failing check: it seems to be flagging an issue in softbinding-algorithm-list.schema.json (specifically empty schemas at lines 140/153), which is a file I haven't touched. My submission only modifies the .json list.

Best regards.

No worries about this one @Cyrah2R this is an issue unrelated to your submission and there is a PR to address it already. We will be reviewing your entry today, if you have an opportunity to join the meeting that would be great.

[domguinard]

One remaining question we have @Cyrah2R is whether the submitted watermarking technology is capable of embedding an identifier? This was not obvious looking at the documentation but is a requirement to be added to the list of supported algorithm as this is key to support C2PA manifest recovery.

[Cyrah2R]

Hi @domguinard,

Thank you for the feedback and for raising this important question.

To clarify: Yes, the LumaTrace technology is specifically designed to embed a unique 64-bit identifier, which is the cornerstone of our support for C2PA manifest recovery through "Soft Binding."

As seen in our submitted lumatrace-core and lumatrace-cloud modules, the mechanism works as follows:

1. Identifier as the Seed: In our Spread-Spectrum implementation, the identifier is not a static tag but the governing parameter of the signal itself. We use a 64-bit cryptographic seed (derived from the Master Key and asset metadata) to modulate a deterministic pseudo-random noise (PRNG) signature.

2. Payload Recovery: The WatermarkDetector performs a multi-scale spatial synchronization scan. By testing the asset against the derived seed, the engine extracts the identity via 2D cross-correlation. This allows us to re-identify an "orphaned" asset (where Jumbf metadata has been stripped) and link it back to its original C2PA manifest stored in our PostgreSQL persistence layer.

3. Forensic Robustness: Our benchmarks confirm that this identifier remains recoverable (with a statistical confidence $\sigma \geq 4.0$) even after aggressive post-processing, such as JPEG Q=50 compression and 50% cropping.

4. HVS Optimization: To maintain invisibility, we apply an adaptive gain based on local image entropy, primarily injecting the signal into the blue chrominance channel where the Human Visual System (HVS) has lower sensitivity.

LumaTrace was built to be platform-agnostic (Pure Java/Kotlin) to ensure it can serve as a reliable reference implementation for C2PA provenance in both cloud and mobile environments.

I look forward to discussing the payload bit-depth and the recovery workflow in today's meeting.

Best regards,

Cyrah

[domguinard]

Accepted by the WM TF during TF meeting.

[domguinard]

@Cyrah2R your algorithm has been approved and added to the list, welcome and congratulations.

[Cyrah2R]

Thank you so much, @domguinard! It’s a huge honor to have the algorithm approved and to officially contribute to the list. I'm thrilled to be part of this ecosystem and looking forward to continuing to build and improve digital provenance solutions. Cheers!!