Duplicate Image Finder

About this tool

Select a batch of images and this groups the ones that are the same or look nearly the same — for clearing out a folder of screenshots taken twice, burst-mode frames, or the same photo re-saved at different sizes. All the work happens in your browser; no image is uploaded.

How it matches. Each image is reduced to a small greyscale thumbnail and turned into a perceptual hash (an "average hash"): a 64-bit fingerprint describing whether each cell of an 8×8 grid is lighter or darker than the image's average brightness. Two images are compared by counting how many of those 64 bits differ (the Hamming distance). Because the fingerprint is based on overall light and dark structure, not exact bytes, a resized or re-compressed copy of a photo still produces a near-identical hash and is matched.

The sensitivity slider is that bit-difference threshold. At "Exact matches only" (0), images must have an identical fingerprint. Raising it groups images whose fingerprints are within a few bits — visually similar composition, colours, and brightness — which catches more near-duplicates but also starts grouping photos that merely resemble each other. Start strict and loosen it only if obvious duplicates are being missed.

What it catches well: exact copies, the same image at different resolutions or JPEG quality, minor edits (a small crop of a border, a watermark added, brightness nudged), and re-saves through different tools.

What it misses: a photo that's been rotated, flipped, or significantly cropped — those change the brightness grid enough that even the "same" picture looks different to the hash. It also can't tell which copy is the higher quality one; it only groups them for you to choose.

Practical use. Load a folder, keep sensitivity low, and review each group — the tool shows every image in a group side by side so you can pick which to keep and delete the rest in your file manager. For a large folder, do it in batches to keep the browser responsive.

Building the batch. Selecting images doesn't have to happen in one go — pick a batch, then open "Add images to compare" again (even for a different folder) and the new files are added to the existing selection rather than replacing it, so you can check across folders in one pass. Use "Clear selection" to start over.

To compare two specific images in detail, use the image diff tool. To shrink the copies you keep, the image compressor; to rename a kept set consistently, the batch file renamer.

Frequently asked questions

What does "similarity sensitivity" do?

It's how many bits of the 64-bit perceptual hash are allowed to differ. At 0, only identical fingerprints group together. Higher values group images that merely look alike, at the cost of false positives.

Will it catch a photo that's been cropped or rotated?

Not reliably. The hash is built from a fixed brightness grid, so cropping, rotation, or flipping changes the fingerprint even for the same picture. It's best at resized, re-compressed, or lightly edited copies.

Does it tell me which copy is best?

No. It groups matches and shows them together; you choose which to keep based on resolution or file size.

Is any image uploaded?

No. Thumbnailing, hashing, and comparison all happen in your browser.

Can I add images from more than one folder?

Yes — select a batch, then open "Add images to compare" again and pick more, from the same folder or a different one. Every selection adds to the current batch instead of replacing it; use "Clear selection" to start fresh.

How many images can I check at once?

There's no hard cap, but comparison is pairwise, so very large sets get slow. Work in batches of a few hundred.

Can it delete the duplicates for me?

No — a web page can't modify your files. It identifies the groups; you remove copies in your file manager.