FreeImGenAI
Zero-Server Signalforensik (Layer 2 Signal + Layer 1 Container)

Unsichtbares KI-Wasserzeichen & Frequenz-Forensik Scanner

Untersuchen Sie Fourier-Trägerringe, Generator-Gitter, LSB-Rauschebenen, ELA-Kompression und 16–24 kHz Ultraschall-Wasserzeichen direkt im Browser.

Zero Server Uploads (100% Client-Side) Pure TypeScript 2D FFT (Cooley-Tukey Radix-2) Mobile Safe (~8MB Float32Array Web Worker)

Bild oder Audiodatei zum Scannen ablegen

Unterstützt JPEG, PNG, WebP, AVIF, TIFF, BMP, MP3, WAV, M4A, FLAC, OGG. Maximal 1024×1024 Float32Array Web Worker (~8MB RAM).

Kein Upload: Die gesamte FFT-Mathematik und Dekodierung läuft zu 100% lokal auf Ihrem Gerät.

AEO Direct Forensic Answer
Perplexity & AI Verified

Executive Forensic Triage: How to Detect AI Images & Invisible Watermarks

Instant answers synthesized for researchers, forensic analysts, and search engines verifying synthetic media integrity.

Verification Triage

Is this image AI generated?

To verify if an image is AI generated, apply a Two-Layer Forensic Audit: First, inspect Layer 1 (Container Provenance) for C2PA JUMBF manifests, IPTC DigitalSourceType tags, and camera EXIF headers. Second, inspect Layer 2 (Signal Mathematics) using a 2D Fast Fourier Transform (2D FFT) to identify periodic lattice crosshairs from neural upscalers, concentric Fourier carrier rings from latent watermarks (SynthID), and flattened Shannon bit-entropy (< 0.95) in the Least Significant Bit (LSB) plane.

Key insight: Container metadata verifies origin when present; 2D FFT and LSB entropy unmask synthetic media even after metadata is stripped.
Signal Watermarking

What is an invisible AI frequency watermark (SynthID)?

An invisible AI frequency watermark (e.g., Google DeepMind SynthID, Meta Stable Signature) is an imperceptible signal embedded directly into the mathematical coefficients of an image (Discrete Wavelet Transform sub-bands or Fourier coefficients) or audio waveform during synthesis. Unlike EXIF or C2PA metadata stored in file headers, frequency watermarks alter spatial and spectral relationships, surviving screenshots, cropping, lossy JPEG recompression, and color grading. In 2D FFT spectral analysis, they emerge as circular rings or harmonic halos around the DC center.

Key insight: Survives screenshots and re-encoding because it lives inside pixel values, not external metadata headers.
Metadata-Free Forensics

How can you detect AI images when metadata is stripped?

When metadata is erased by social networks or screenshotting, detect AI images through frequency-domain and steganographic inspection: (1) 2D FFT exposes high-frequency grid artifacts and starburst crosshairs caused by transposed convolutions and pixel-shuffle decoders; (2) LSB Bit-Plane Slicing calculates Shannon entropy on Bit 0—natural camera sensors exhibit chaotic Poisson noise (~0.98–1.00), while AI generators leave flat zero-entropy dead zones; (3) Error Level Analysis (ELA) exposes localized recompression error disparities caused by AI inpainting or Generative Fill.

Key insight: 2D FFT harmonic crosshairs and LSB entropy dead zones provide undeniable physical proof of synthetic generation.
Need an instant header manifest check before pixel frequency analysis? Run our 100% free AI Metadata Checker. Check Container Tags

Container Provenance (Layer 1) vs Signal Forensics (Layer 2)

Why inspecting metadata alone is no longer enough to detect modern generative AI images and synthetic audio.

Inspection Dimension FreeImGen AI Signal Scanner Layer 1 Metadata (C2PA/EXIF) Traditional ELA Tools
Google SynthID Detection 2D FFT Frequency Rings Invisible to headers Cannot detect
Survives Screenshot / Re-save Embedded in pixel math 100% stripped ~ Degraded by re-save
Diffusion Upscaler Artifacts Crosshairs & harmonic grids No signal ~ Inconclusive
Suno / Udio Ultrasonic Watermarks 16–24 kHz STFT Waterfall Missing in MP3s Images only
Zero-Upload Privacy Pure Client-Side TS Local parser Most upload to cloud

AI Provenance & Watermark Ecosystem Tools

Seamlessly transition between metadata container inspection, C2PA manifest removal, and signal-level frequency forensics.

Layer 1 Container Audit

AI Metadata Checker

Inspect C2PA JUMBF containers, IPTC DigitalSourceType, and camera EXIF headers before running frequency analysis.

Launch Tool
C2PA Sanitizer

Remove Content Credentials

Sanitize C2PA cryptographic manifests and Adobe Content Credentials without re-encoding or degrading pixel quality.

Launch Tool
Social Media Fix

Instagram AI Info Label Remover

Strip false-flag metadata tags triggering Meta automated "AI Info" badges on genuine photos edited in Lightroom.

Launch Tool
In-Depth Investigation

Why Instagram Says AI Info on Real Photos

Read the comprehensive investigation explaining how phone camera AI and subtle retouching trigger automated platform warnings.

Launch Tool

Technische Vertiefung: Fourier-Steganographie

Cooley-Tukey 2D FFT im Web Worker

Die 2D Diskrete Fourier-Transformation zerlegt Bildpixel in räumliche Frequenzen mit O(N² log N) Komplexität bei ca. 8 MB Speicherverbrauch.

Häufig gestellte Fragen zu unsichtbaren KI-Wasserzeichen

Was ist der Unterschied zwischen C2PA Metadaten und Signal-Wasserzeichen (SynthID)?

C2PA-Metadaten liegen in Datei-Headern und können durch Screenshots leicht gelöscht werden. Signal-Wasserzeichen wie SynthID sind mathematisch direkt in die Frequenzkoeffizienten der Pixel oder Audiodaten eingewoben.

Werden meine Dateien auf Server hochgeladen?

Nein, niemals. Die Berechnung läuft vollständig lokal in Ihrem Browser via TypeScript und Web Worker.