Skip to content

Releases: veritaschain/cpp-spec

CPP v1.4 - Depth Analysis Extension & Multi-Platform Support

29 Jan 06:44
26a8f67

Choose a tag to compare

CPP v1.4 Release Notes

Release Date: January 29, 2026

Overview

CPP v1.4 adds the Depth Analysis Extension for screen detection and expands platform support to include physical cameras, drones, surveillance systems, and embedded devices.

What's New

🔬 Depth Analysis Extension (OPTIONAL)

  • LiDAR/ToF-based screen detection to identify photo-of-screen attacks
  • Statistical depth analysis (min, max, mean, standard deviation)
  • Plane analysis for surface flatness detection
  • Screen detection verdict with confidence scoring
  • Privacy-preserving: raw depth data is NOT stored, only statistical summaries

📷 Multi-Platform Support

New DeviceClass field supporting 13 device categories:

  • Smartphone, Tablet - Mobile devices
  • DigitalCamera - DSLR, mirrorless, compact cameras
  • ActionCamera - GoPro, DJI Action
  • Drone - DJI Mavic, Autel
  • Surveillance - IP cameras, CCTV
  • BodyCamera - Axon, law enforcement cameras
  • Dashcam - Vehicle-mounted cameras
  • IndustrialCamera - Machine vision (Basler, FLIR)
  • MedicalImaging - Endoscopes, dermascopes
  • Webcam, Embedded, Other

🔧 New Data Model Fields

  • DeviceInfo.DeviceClass - Platform identification
  • DeviceInfo.CaptureDevice - Physical camera metadata (lens, aperture, ISO, etc.)
  • DeviceInfo.SecurityModule - TPM/HSM/SecureElement support
  • SensorData.DepthAnalysis - Complete depth analysis object

📊 16 Depth Sensor Types

Platform-independent sensor identification:

  • Mobile: LiDAR, TrueDepth, ToF, StructuredLight, Stereo
  • Camera: DualPixelAF, PhaseDifferenceAF
  • Industrial: IndustrialLiDAR, IndustrialToF, RGBD, Radar, Ultrasonic
  • External: ExternalLiDAR, ExternalToF, StructuredLightScanner

Embedded Systems Support

Minimum requirements for resource-constrained implementations:

  • CPU: ARM Cortex-M4+
  • RAM: 256KB
  • Flash: 512KB
  • Optional: SHA-256 hardware acceleration

Backward Compatibility

  • All v1.3 events remain valid
  • DepthAnalysis is OPTIONAL - existing implementations unaffected
  • Merkle tree construction unchanged from v1.3

Migration Guide

No migration required. To adopt v1.4 features:

  1. Add DeviceClass to DeviceInfo
  2. Optionally add DepthAnalysis to SensorData if depth sensor available
  3. Optionally add CaptureDevice for physical camera metadata

Reference Implementation

Links

CPP v1.0 - Capture Provenance Profile Initial Release

18 Jan 06:26
5383b4a

Choose a tag to compare

CPP v1.0 - Capture Provenance Profile

Release Date: 2026-01-18
Status: Release Candidate

Overview

CPP (Capture Provenance Profile) is an open specification for cryptographically verifiable media capture provenance. Part of the VAP (Verifiable AI Provenance) Framework.

Key Features

🔐 Completeness Invariant

Mathematical deletion detection using XOR hash sums - detects ANY missing events in the capture chain.

⏱️ External Third-Party Verification

RFC 3161 TSA mandatory (Silver+) - eliminates self-attestation vulnerabilities found in existing solutions.

🔒 Privacy by Design

  • Location OFF by default
  • Zero-knowledge biometric attestation (ACE)
  • GDPR-compliant crypto-shredding

🔗 C2PA Interoperability

Designed to complement C2PA, not compete. Export format included.

What's Included

  • docs/CPP-Specification-v1.0.md - Main specification
  • docs/C2PA-Interoperability.md - C2PA integration guide
  • schemas/cpp/ - JSON schemas for core events
  • schemas/ace/ - ACE extension schemas
  • examples/ - Sample events and verification packs
  • test-vectors/ - Conformance test data
  • tools/ - Reference implementation (Python)
  • regulatory-mapping/ - EU AI Act, GDPR mapping

Conformance Levels

Level TSA Anchor ACE Target
Bronze Optional Optional Hobbyists
Silver Daily Optional Families
Gold Per-capture Required Legal/Journalism

Links

License

  • Specification: CC BY 4.0
  • Code/Schemas: Apache 2.0

VeritasChain Standards Organization
standards@veritaschain.org