Account

What new tool do you want?

Image Tools

Duplicate & Similar Photo Finder

Find identical, near-identical and similar photos in a folder: exact copies by file fingerprint, resized or re-saved copies and burst shots by image similarity. Pick which to keep (the largest is suggested), then download a CSV list or a ZIP without the duplicates. Nothing is uploaded.

How to use

  1. 1Drop photos or choose a whole folder. Each one is fingerprinted and a tiny thumbnail made in your browser.
  2. 2Choose identical files only, near-identical (resized or re-saved copies), or similar (also burst shots and light edits). Duplicates are grouped, with the best copy marked Keep.
  3. 3Change which copy to keep if you like, then download a CSV of the duplicates or a ZIP of the photos to keep.

Frequently asked questions

How are similar photos found?

Identical files are matched by their SHA-256 fingerprint. For similarity, each photo is shrunk to 9 × 8 grey pixels and turned into a 64-bit 'difference hash' recording whether each pixel is brighter than its neighbour. Copies that were resized, recompressed or slightly brightened keep almost the same hash; the near-identical setting allows up to 4 of 64 bits to differ, the similar setting up to 10. Photos of clearly different shapes are never grouped.

Which copy is suggested to keep?

The one with the most pixels, then the largest file (usually the least compressed), then the oldest date taken, then the shortest folder path. Click another photo in the group to keep that instead.

Can it delete the duplicates for me?

No. A web page can't delete files on your computer. Download the CSV (it lists every file's path and whether it's a keeper or a duplicate) and delete them yourself, or download the ZIP of photos to keep, check it, and replace the folder.

What about HEIC photos?

Exact duplicates are found for any file. Browsers other than Safari can't decode HEIC, so in Chrome or Firefox HEIC photos are only matched when the files are identical and show no preview.

Are my photos uploaded?

No. Fingerprints, hashes and thumbnails are all computed in your browser.

Sources

Example: cleaning up a phone backup

Choose the folder where your phone backups and WhatsApp downloads have piled up, say 3,000 photos. Near-identical finds the same photo saved at full size, as a WhatsApp copy (resized and recompressed) and as an emailed copy, and suggests keeping the full-size one. Switch to Similar to also catch bursts of nearly the same shot, then go through the groups and choose the best frame of each burst.

Why some similar photos aren't grouped

The hash describes the overall pattern of light and dark, so it's robust to resizing and recompression but not to cropping, rotation or large edits: a portrait and the same photo cropped to a square won't match. Photos taken a moment apart where the subject moved a lot also differ. The Similar setting catches more of these at the cost of occasionally grouping two different shots of the same plain scene, so look before you delete.

Related tools

More: Image Tools