Structural Content Identity & Forensic License Tracing. Re-links modified content to registered originals through structural fingerprinting — surviving hard visual modifications.
IFP reads cryptographically signed C2PA manifests at registration. When a dispute arises, a valid manifest helps read more signals — signature date, device identity, and provenance chain. Cameras: Leica, Nikon, Sony. Software: Photoshop, Lightroom. IFP is read-only — we do not sign or issue C2PA manifests and are not a conformant product.
Read more
3D depth sensor data embedded by camera hardware is extracted and stored at registration. A copy or screenshot always loses depth data — its presence is strong evidence of an original capture. Pixel, Samsung, Huawei portrait mode.
Read morePhoto by Abhi Verma
IFP detects when your original image was used — not when it was recreated. If someone takes your photo and crops it, flips it, rotates it, changes colors, adds overlays, or blurs it — the image structure stays the same, and IFP finds the link back to your registered original.
This is not AI content tracing. IFP does not track images that were regenerated, restyled, or created from scratch using AI tools like img2img or style transfer. The original pixels must still be present — just transformed.
Upload your source image to IFP before you publish it anywhere. IFP creates a structural fingerprint — a unique identity based on the image itself, not metadata. If your original includes C2PA provenance data, upload it as-is — we store the cryptographic manifest and it will protect your authorship in disputes.
Spot your image on someone’s website, in an ad, or on social media — cropped, flipped, recolored, blurred, it doesn’t matter.
Just a regular screenshot of where it’s being used. Even with UI elements, browser frames, or overlays around it.
Upload the screenshot to “Check Image” in IFP. The system will match it back to your registered original — through the modifications, through the UI noise, through everything.
IFP shows the match with confidence score, your registration timestamp, and license data. You now have evidence to request removal from ads, websites, or platforms.
Detects originals rotated at any arbitrary angle, including non-standard like 17° or 243°
Finds originals under heavy interface elements, watermarks, text boxes and graphic overlays
Survives significant brightness shifts, contrast adjustments and exposure changes
Identifies originals even when heavily cropped — only a portion of the image remains
Detects horizontally flipped images used to avoid reverse image search
Works through grayscale conversion, color grading, hue shifts and desaturation
Survives gaussian blur up to level 45 out of 100 — far beyond recognizable quality
Handles perspective distortions from photos of screens, projections and angled captures
Resists noise overlay, JPEG compression, re-encoding and graphical element additions
Distinguishes between genuinely identical content and visually similar but different images
Forensic license tracing with unique IDs — tracks specific copies through any modification chain
Multiple heavy modifications at once — crop + flip + contrast + rotation + UI overlay combined
| Google TinEye YandexGoogle / TinEye / Yandex | IFP | |
|---|---|---|
| Hard modifications | Reverse image search breaks after rotation, crop, flip, or overlay — modified copies are not found | Finds originals after rotation + crop + blur + overlay combined |
| B/W & color shift | Grayscale conversion or color grading makes the image unrecognizable to perceptual hashing | The algorithm uses brightness-independent analysis — B/W, color grading, and hue shifts are handled |
| Crop & flip | Cropping changes the image hash entirely; flipping evades all perceptual matching | The algorithm recovers geometry and handles mirrored copies automatically |
| License forensics | No mechanism to trace specific licensed copies through modification chains | Each image gets a unique forensic license ID traced through any chain of modifications |
| False positive control | Returns visually similar but different images as matches — no identity verification | Structural identity, not visual similarity — zero false positives on similar-looking images |
Register your first image and see how IFP identifies it through any modification.
Get started