V
VERDAT

Comprehensive image analysis

VERDAT combines multiple detection methodologies to provide thorough image authenticity assessment. Every analysis produces explainable, audit-ready results.

Analysis Capabilities

Three complementary detection layers work together to identify manipulation and synthetic generation.

Metadata Analysis

Deep inspection of EXIF data, file headers, and encoding structures to identify tampering artifacts and inconsistencies.

  • EXIF data extraction and validation
  • File structure integrity checks
  • Encoding pattern analysis
  • Timestamp consistency verification
  • Device signature matching

Visual Forensics

Pixel-level analysis to detect clone regions, splicing artifacts, and lighting inconsistencies that indicate manipulation.

  • Clone detection and localization
  • Splicing boundary detection
  • Lighting direction analysis
  • Compression artifact mapping
  • Noise pattern analysis

AI Generation Detection

Specialized models trained to identify synthetic images from major generative AI architectures.

  • Diffusion model fingerprinting
  • GAN artifact detection
  • Cross-generator coverage
  • Version-aware detection
  • Emerging model adaptation

Output & Integration

Results designed for both automated workflows and human review processes.

Probabilistic Scoring

Calibrated confidence scores that support risk-based decision thresholds rather than binary verdicts.

Structured Reports

Machine-readable JSON responses with human-friendly audit reports for compliance documentation.

Signal Explainability

Every finding includes specific evidence—locations, patterns, and technical details—for review.

Model Versioning

Each analysis records the exact detection model version used for reproducibility and audit trails.

Async Processing

Batch endpoints for high-volume workloads with webhook notifications on completion.

Data Isolation

Images processed in isolated environments with no retention unless explicitly requested.

Built with v0