AI Watermark & Provenance Inspector Gemini & Flow Spec

Multi-tier forensic verification: Visible Glyphs vs. SynthID Frequency Carriers vs. Cryptographic C2PA

Spatial Domain (Visual Viewport) Clean / Watermark Off
Overlay: None
Frequency Domain (2D-FFT Spectrum) Carrier Harmonic Active
SynthID Ring: 14.8 kHz Peak
Distribution Perturbation Stress Testing
Quality 85%
0% Cropped
0.0 σ
Forensic Reality: Google's toggle removes the visible corner badge to avoid visual clutter in creative workflows. However, DeepMind's SynthID embeds subtle mathematical phase shifts into high-frequency DCT/Fourier coefficients that withstand lossy recompression and cropping. Cryptographic C2PA manifests bind the model lineage (e.g. Gemini 2.5) via SHA-256 asset hashes.
SynthID Detection 98.4%
Imperceptible frequency carrier detected with high confidence (>90%).
C2PA Cryptographic Tree VALID
ECDSA P-256 signature chain matches Google Trust Root CA.
Visible Glyph Presence ABSENT (OFF)
Cosmetic watermark stripped via Gemini/Flow toggle.
C2PA Manifest Claims

Forensic Watermarking & Provenance Architecture

Read the explanation

AI watermarking operates across three distinct forensic tiers: visible corner badges, imperceptible frequency embeddings, and cryptographic manifests. Toggling off the visible watermark simply strips the cosmetic corner overlay. The imperceptible SynthID harmonic ring persists untouched in high frequency bands. When lossy compression or cropping is introduced, the mathematical confidence score degrades continuously, while cryptographic C2PA invalidates if pixels diverge past the digest threshold.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.