Why Lumethic
A check, not a guess
Most tools that judge whether an image is real read the finished picture and estimate the odds. Lumethic takes a different route. It compares a photograph with the RAW file the camera recorded, so the answer rests on sensor data rather than on a probability.
The core idea
Deterministic verification against probabilistic detection
The two approaches sound similar but work in opposite ways, and the difference decides how much you can rely on the result.
Lumethic checks physical evidence
A camera sensor leaves traces that a generated image does not have, including noise patterns specific to the sensor and the interpolation a colour filter array produces. Lumethic looks for these in the RAW file and compares them with the published photo. The result follows from the data, so the same inputs give the same answer.
Detectors estimate from pixels
An AI detector studies the finished image and reports how likely it is to be generated. It has no access to the original sensor data, so it works from statistics that generators keep learning to imitate. The result is a probability that shifts as models improve, and it flags genuine photos when they have been heavily processed.
What the check covers
Four comparisons behind every result
Verification runs several checks against the RAW file and the export. Each one looks at a different way a photograph can fail to be a genuine, unaltered capture.
Sensor authenticity
The noise profile and sensor characteristics are checked against the stated camera and settings, which a generated image cannot reproduce.
Visual consistency
The export is compared with a reference built from the RAW. Normal edits pass, while added, removed, or swapped content does not.
Recapture detection
The system looks for the patterns left when an image on a screen is rephotographed, a route used to fake a real capture.
Metadata validation
Embedded data such as timestamps and camera details is reviewed for internal consistency rather than taken at face value.
How it relates to other methods
Where Lumethic sits next to the alternatives
Several other methods aim to establish trust in an image, and they solve different parts of the problem. Here is how Lumethic relates to each.
Capture-time signing
Some systems record authenticity at the moment of capture through a controlled app or signing hardware, an approach taken by tools such as Truepic. It is strong at the source but needs a specific app or new camera, so it does not reach existing equipment or older photos. Lumethic verifies after capture from the RAW file an ordinary camera already records, which covers that larger body of work.
Signature infrastructure
Services that sign and timestamp files, such as CertNode, prove a file has not changed since signing. They do not inspect the image, so they sign a generated picture as readily as a real one. Lumethic adds the missing step by checking the photo against its RAW source before it signs.
AI detectors
Detectors estimate from pixel statistics how likely an image is to be generated. They lose ground as generators improve and flag real photos that have been heavily processed. Lumethic compares against the camera original instead, which is a check rather than a guess.
Watermarking at the generator
Watermarks like Google SynthID mark content that cooperating AI models produce, which helps label known synthetic output. A missing watermark does not prove a photo is real, since it could come from a model that does not participate. Lumethic answers that side directly, by showing a photograph is a genuine capture.
Verify, then sign
The credential reflects a checked image
A C2PA credential records that a file has not changed since it was signed. On its own it says nothing about whether the original was a real photograph, because a tool that signs whatever it receives will sign a synthetic image just as readily.
Lumethic runs its checks first and signs only when they pass. The credential therefore stands for an image that was verified against its source, which is what makes the signature worth trusting.
Read about verify then signYour files stay yours
Verification without keeping your originals
The RAW file is the most valuable thing a photographer holds, so Lumethic does not store it. Uploads are processed in memory on infrastructure based in the EU, and the RAW is removed once the check is done. Verification does not require handing over your originals for storage.
Read about data handlingSee the difference for yourself
Run a photo through the check and read what the credential records.