Invisible.
Traceable.
Forensic.
SpectraSeal embeds imperceptible forensic watermarks into images — encoding your identity, timestamp, and IP address directly into the DCT frequency domain. Even after compression, cropping, or screenshots, the watermark remains recoverable.
How It Works
Three layers of signal engineering combine to make SpectraSeal watermarks both invisible and robust.
DCT Frequency Embedding
Your payload is encoded into the mid-frequency DCT coefficients of the image — the same domain used by JPEG compression. Modifications here are perceptually invisible to the human eye while remaining machine-detectable.
Reed-Solomon ECC
The payload is protected with Reed-Solomon error-correcting codes before embedding. This means up to 30% of the encoded signal can be corrupted or lost — through cropping or noise — and the full payload is still recoverable on decode.
Survives Attacks
SpectraSeal watermarks are designed to survive real-world image transformations: JPEG re-compression at quality ≥70, moderate cropping, screenshots on typical displays, and downscaling to half resolution.
What Gets Embedded
Every encoded image carries a structured JSON payload that ties the image back to the moment and machine that created it. This metadata is invisible — until you decode it.
- ✓ ID & Label — your custom label prefixed with "DEMO"
- ✓ IP Address — the public IP of the encoding request
- ✓ Geolocation — city and country resolved from the IP
- ✓ Datetime — UTC timestamp of encoding
- ✓ Image Hash — SHA-256 fingerprint of the original image
// Decoded payload example
{
"id": "DEMO - your label",
"ip": "203.0.113.42",
"geo": "Seattle, US",
"datetime": "2026-03-25T22:15:00Z",
"image_hash": "a1b2c3d4e5f6"
}Demo Usage Limits
This is a public demo with rate limiting to prevent abuse. Each IP address is permitted 5 uses per calendar month. Both encode and decode operations consume one credit from the same pool. Credits reset automatically at the start of each month.