Photo Forensics: Error Level Analysis & Noise Map
Upload a photo to see an error level analysis (ELA) map and a noise map, the two classic ways to look for a pasted, retouched or re-saved region. Each map comes with plain-language notes on what it can and cannot tell you, and you can run the AI detector on the same file with one click.
Drop or paste a photo to analyse it
Error level analysis, noise map and JPEG quality estimate in seconds.
Choose a photo →Works best on the original JPEG. Screenshots and social downloads flatten both maps.
How to read the two maps
Error level analysis re-saves the picture once and shows how much each area changed. In a photo that was saved once, the whole frame responds about the same, with brighter edges where there is detail. A region that was pasted in from another file, or re-saved separately, can respond differently from its surroundings: brighter, or noticeably darker.
The noise map shows the fine grain left after a small smoothing pass. A camera leaves a fairly even grain; a region with a different texture can point to a paste, a heavy retouch, or generated content.
Three honest limits. Sharp edges, text and flat skies stand out on both maps by themselves. An image re-saved as a whole after editing looks uniform, so a clean map is not proof of anything. And neither map is an AI detector; for that, run the pixel detector on the same file below the maps.
What this photo forensics tool does
- ELA map, auto-scaled so the differences are visible, with a note when one region stands apart from the rest.
- Noise map, the amplified fine grain of the image.
- JPEG quality estimate, read from the file's own compression tables, which tells you whether the maps are likely to mean anything.
- One-click AI check on the same upload, and a link to the edited photo detector, which reads the file's editing history from its metadata instead of its pixels.
Frequently asked questions
What is error level analysis (ELA)?
ELA re-saves the image once as a JPEG and shows how much each part changed. Areas that were saved a different number of times, or at a different quality, can respond differently from their surroundings. It is a lens for looking, not a verdict.
Can ELA prove a photo was photoshopped?
No. It can point at a region worth a closer look, but sharp edges, text and flat areas also stand out naturally, and an image that was re-saved as a whole after editing looks uniform. Treat an odd region as a question to investigate, not an answer.
Why does my ELA map look uniform or blank?
Because the file has been through heavy or repeated compression. Screenshots, social-media downloads and messaging-app forwards all flatten the map. Ask for the original file if you can.
Does this detect AI-generated images?
Not by itself. Neither map is a detector. Run the AI detector on the same file, one click below the maps, for that question.
What does the noise map show?
The fine grain left after a small smoothing filter. Camera sensors leave a fairly even grain across a photo; a region with a different texture can indicate a paste or a heavily processed area, though blur and flat surfaces also change it.
Is the photo forensics tool free and private?
Yes. It is free with no sign-up, the maps are made on the fly and sent back to you only, and the uploaded image is deleted right after analysis.
Related: edited photo detector · metadata checker · fake image detector · remove metadata from a photo