Photoshop Detector
Find out whether a photograph has been edited — objects removed or added, regions cloned, a face retouched, two pictures spliced together. This is a different question from “was it made by AI”, and it needs different evidence: not generation traces, but places where one part of the file has a different history from the rest.
How editing leaves a trace
An edited photo is a file with an inconsistent past. The camera wrote it once; an editor rewrote parts of it. Every method below hunts for that inconsistency rather than for anything that looks wrong to the eye.
Uneven compression history
Each save re-compresses the image. An area pasted in later has been through a different number of saves and returns a different error level — the core idea behind Error Level Analysis, and the strongest single signal here.
Repeated pixel neighbourhoods
The clone stamp copies a region. Statistically identical patches appearing in two places almost never happen in a real photograph.
Noise that stops matching
Sensor noise is uniform across a genuine frame. Retouched skin, a removed object or a pasted sky breaks that uniformity.
Edges that are too clean
A cut-out object has a boundary sharper than the lens could produce, often with a one-pixel halo where it was masked.
Editing software in the metadata
EXIF frequently keeps the editor's name and the modification timestamp — and a missing capture block on a supposedly original photo is a signal in itself.
Lighting and perspective that don't agree
A spliced object brings its own shadow direction and vanishing point. Our edge and colour analyzers pick up the mismatch.
How the analysis works
Upload the original file
Full resolution, straight from the source. Screenshots and re-saved copies destroy most of the evidence we measure.
12 independent methods run in parallel
Three neural networks look at the image as a whole. Nine forensic methods examine it as a signal: frequency (FFT) analysis, noise pattern, pixel-level statistics, EXIF metadata, Error Level Analysis, texture coherence, colour distribution, Benford's Law and edge detection.
You see the breakdown, not just a number
Every method reports separately, so you can tell whether the verdict rests on one weak signal or on nine that agree.
How much each method counts
Most detectors give you one number from one model. We run twelve independent analyzers and weight them — three neural networks carry the majority, but nine forensic methods examine the file as a signal and can overrule them. Here is the exact weighting we use in production:
| Method | Weight in final score |
|---|---|
| Neural network #2 | 25% |
| Neural network #1 | 15% |
| Neural network #3 | 15% |
| EXIF metadata | 10% |
| Frequency (FFT) | 6% |
| Benford's Law | 6% |
| Noise pattern | 5% |
| Error Level Analysis | 4% |
| Texture coherence | 4% |
| Edge detection | 4% |
| Pixel statistics | 3% |
| Colour distribution | 3% |
What people actually upload
These are our own numbers, not an industry estimate — every check run through FakeSec since launch. The split is closer than most people expect: roughly four in ten uploads really do turn out to be generated.
When people check for editing
- Marketplace listings — A product photo with defects edited out.
- Insurance claims — Damage added, or pre-existing damage removed.
- Dating profiles — How heavily a photo was retouched.
- Journalism and evidence — An image presented as an unaltered record.
Accuracy — and where it breaks down
On our internal benchmark of 240,000 images (120,000 AI-generated, 120,000 real) the combined score reaches up to 98% on unedited, full-resolution files. That number drops, sometimes sharply, in four situations — and you should know them before trusting any result:
- Heavy compression. Social platforms re-encode everything. A photo pulled from Instagram has already lost much of its noise signature.
- Screenshots. A screenshot is a new image of an image. Original metadata and compression history are gone.
- Stylised content. Flat colour and clean lines remove the texture and noise evidence forensic methods depend on.
- Post-processing. Added grain, filters, upscaling and manual retouching all mask generation traces.
No detector on the market is 100% accurate — ours included. Treat a high score as strong evidence, not as proof. When a decision matters, look at which of the 12 methods agreed.
One honest caveat specific to this check: every photo that has been resized, filtered or exported by a phone gallery is technically “edited”. What matters is whether editing is localised to a region — that is what indicates manipulation rather than routine processing. The per-method breakdown shows you which it is.
Frequently asked questions
Related tools
Check your first image now
Create a free account and get 3 checks. Full breakdown across all 12 methods, no card required.