Reference set
Known-source images establish the candidate sensor residual.
For you / Content checks
Which cameracould this be from?
Compare a questioned image with original photos from a known camera.
See how it worksPRNU Attribution
In plain English.
Camera sensors leave tiny patterns in their images. Comparing those patterns is called PRNU attribution—a kind of camera fingerprint, not a fingerprint of a person.
A situation
worth a second look.
A claim includes a photograph said to come from one particular camera.
Defeated / content checkGuided example
shared-photo.jpg
See what a useful answer looks like.
Can you trust this photo?
A signal, the reason behind it, and a practical next step. This guided example uses fixed sample findings, not a live detection model.
A clearer next step.
Use suitable original reference photos to test that source. A weak or missing match is inconclusive, not a fake verdict.
Inspect your own contentFor technical teams: explore the method
Illustrative product views and fixed sample findings. Advanced analysis requires approved services and suitable reference data.
The guided sample supports a same-camera hypothesis within stated thresholds.
PRNU tests a same-camera hypothesis. Failure is inconclusive, and the method does not determine whether content is real, fake, or AI-generated.
How the review works
1Enroll
Collect 3–20 original images from one known camera.
2Extract
Estimate sensor residuals while masking unreliable regions.
3Correlate
Compare the questioned residual to the enrolled reference.
4Conclude
Support same-camera or return inconclusive—never real/fake.