Free tool · Fully offline
Image Comparison
Load two photographs and find out whether they are the same file, the same picture saved twice, the same scene shot twice, or nothing to do with each other.
The three questions this answers
Is it literally the same file? That is the SHA-256 hash. Two files with the same hash are byte-for-byte identical, full stop. Two files with different hashes may still be the same picture — re-saving a JPEG changes every byte while leaving the image looking untouched.
Is it the same picture? That is the perceptual hash. It reduces each image to a small fingerprint of its light and dark structure, so it survives resizing, re-compression, minor colour shifts, and watermarks. Two versions of the same photograph will score very close even when their file hashes are completely different.
Was it taken by the same camera? That is in the metadata comparison. Matching body serial numbers are close to conclusive. Matching make and model are suggestive and no more — a great many people own the same phone.
Reading the similarity score
- 100% — the same image. Possibly re-saved, cropped very slightly, or renamed.
- 90–99% — almost certainly the same photograph, edited, compressed or resized.
- 75–89% — same scene or same subject, different frame. Two shots seconds apart score here.
- Below 70% — different images.
One caution worth knowing: perceptual hashing works on structure, so two very flat images — a pale wall, a blank document scan, a night shot that is mostly black — will score high against each other simply because neither has much structure to compare. Look at the pictures before you trust the number.
Where this gets used
Confirming a profile photo on one platform is the same image as one on another. Checking whether a photograph submitted as new is actually a recycled older one. Establishing that two accounts posted the identical file, which is a much stronger link than two accounts posting similar pictures. And proving a file has not changed since you received it, which is what the hash is for.