Step by step

How to remove duplicate photos without deleting the originals

Short answer: don't delete anything. Move it, hold it, and look at your archive before you commit. Every irreversible photo loss involving a duplicate finder happens because deletion and identification were the same action.

The walkthrough below uses DuplicateFinder Pro, which is ours. The principle isn't ours and applies to whatever you use: nothing should leave your archive on the same click that found it.

Scanning and the full reports are free, so you can run everything on this page and see the result before deciding anything.

Why "delete duplicates" is the wrong verb

A duplicate finder makes two separate claims, and people usually only hear the first one.

The first claim is these two files have identical contents. That one can be proven, and good software proves it by reading both files rather than comparing their names.

The second claim is and therefore you don't need this one. That is not a fact about the files. It depends on which drive you're keeping, which folder your catalog points at, whether the second copy is a backup you meant to have, and whether the thing you're about to delete is genuinely duplicated or merely similar. Software can't know any of that.

So the safe shape of this job is: let the software make the first claim, keep the second one for yourself, and make sure the gap between them is reversible. In practice that means files get moved and held, not deleted, and you get to look at your archive before anything becomes permanent.

Before you run anything

  • Decide which location is the keeper. If you're comparing more than one folder or drive, work out in advance which one is the archive you intend to keep. This is the only decision the software genuinely cannot make for you.
  • Know where your catalog points. If you work in Lightroom or Capture One, quarantining a catalogued photo will make it show as missing until you restore it. What that actually looks like is worth reading first.
  • Don't start with your only copy of anything. This isn't specific to our software. Any operation that moves thousands of files is one you want a backup behind.

Installing it, and what Windows will say

DuplicateFinder Pro isn't code-signed. A signing certificate is an annual cost that hasn't been taken on yet, and without one Windows describes the publisher as unknown. That's a statement about paperwork, not about the file — but it produces warnings that deserve a straight explanation rather than a "just click through".

  1. Check the fingerprint before you run the installer The download page publishes the installer's SHA-256 hash. Open PowerShell in the folder you downloaded to and run Get-FileHash .\DuplicateFinderPro_Setup_v1.5.exe -Algorithm SHA256. It should match character for character. If it doesn't, don't run it — download it again, and if it still differs, email support.
  2. Your browser says the file isn't commonly downloaded That's a statement about download volume. A new installer from a small publisher has no history behind it. Choose Keep — in Edge it's under the arrow beside Delete.
  3. SmartScreen shows a blue "Windows protected your PC" screen Click More info, then Run anyway. This is exactly what the fingerprint check was for: you verified the file yourself, so you don't need a certificate to tell you it's the file that was built.
  4. The installer asks for administrator permission It installs to Program Files and registers an uninstaller. Nothing outside its own folder is touched.
  5. Activate, if you bought a licence Help → Activate, then paste the key from your receipt. You need to be online once. After that it works offline indefinitely, with no repeated checks and no account. If you haven't bought one, skip this — everything up to moving files works without it.

If any of that leaves you uncomfortable, that's a reasonable place to land. Run the fingerprint check, and if you're still not comfortable, don't install it. Clicking through warnings you don't understand is a worse habit than skipping one piece of software.

Scanning one drive

If you're cleaning up a single archive — one drive, one folder tree — this is the whole flow.

  1. Set the scan target Top left, under Scan Target. Use Browse to pick a folder, or point it at a whole drive root.
  2. Choose what gets scanned RAW photos, exported JPGs, raw video, rendered video and GoPro audio are separate switches in the left rail. On a multi-terabyte drive, running one category per pass finishes sooner and lets you act on the first result while the second runs.
  3. Set the RAW/JPG similarity threshold if you need to The slider defaults to 90. It only affects the pairing of a RAW with its exported JPG, which uses a perceptual comparison rather than an exact one. Lower catches more heavily cropped exports; higher restricts it to near-exact matches. Leave it alone unless your exports are cropped hard.
  4. Start the scan Progress is checkpointed at phase boundaries, so pausing or an interrupted run resumes rather than starting over. If the window greys out, Windows is busy, not crashed.

Comparing two or more locations

Comparing separate folders or drives is a different tab, and it's easy to miss. Scan Target in the left rail takes one location. Cross-location comparison lives in Folder Duplicates, which has its own list and its own scan button.

Add locations with Add Folder / Drive — folders or whole drive roots, two or more, on any connected drives. No naming convention or folder structure is assumed: every file under each location is checked by content, at any depth, against every file under the others.

The rule that matters is stated on screen above the scan button: the copy in the first folder listed is the one kept. Matching copies in the folders below it are the ones flagged. There's no drag to reorder — to change which copy wins, select a folder, Remove Selected, and add it back in the order you want. Then click Scan for Duplicate Files, not Start Scan.

If you're consolidating two external drives specifically, there's a longer piece on how to compare two external drives — including the order of operations that keeps the drive you're retiring safe.

Reading the results

Results are split by kind across the tabs: RAW and JPG pairs in the same folder, RAW and JPG pairs across folders, duplicate JPGs, duplicate RAWs, duplicate videos, GoPro audio, and folder comparisons.

Every row carries a Role — whether it's the copy being kept or a copy being flagged — plus the file, its size, its folder, and a Match column naming which test found it. Select any row to preview the file. A running Recoverable figure, group count and flagged count sit along the bottom.

Three tests run in order, cheapest first. Size, which eliminates almost everything without reading a byte. A partial fingerprint of the first and last megabyte of whatever survived. Then a full read of whatever's still standing. Only a full-content match is reported as a duplicate — never a matching name, never a matching timestamp.

Export the CSV before you move anything. It costs nothing and it's your record of what the drive looked like on the day you did this.

Rows marked Review

RAW files sharing a filename but not byte-identical are grouped anyway, so you can see them side by side. Those rows read stem+size in the Match column rather than md5, and if the sizes differ by more than 5% they're marked Review instead of being flagged. The software is telling you it found something related and is declining to decide. Look at those yourself.

Flag, quarantine, delete — three different verbs

These get confused, and the confusion is the dangerous part.

What each one does

  • Flag for Deletion marks a row. It does not touch the file. Despite the name, nothing is deleted and nothing has moved.
  • Move Flagged to Quarantine moves the marked files into a quarantine folder on their own drive and records where each came from.
  • Move Selected to Quarantine does the same for rows you've highlighted, skipping the flagging step.
  • Emptying quarantine is the only action that permanently removes anything, and it's yours alone to take.

What quarantine costs

  • It moves rather than copies, so holding files uses no extra space.
  • The honest trade: the space isn't reclaimed either, until you empty it.
  • Files are re-checked against the copy being kept at the moment they move, not just when the scan ran. Anything changed in between is skipped and named.
  • Identical filenames from different drives get unique names in quarantine, each keeping its own recorded origin.

The 24 hours, and putting things back

Quarantined files carry a 24-hour hold. Use it. Open your archive in whatever you actually work in — Lightroom, Bridge, Explorer — and look at it properly before emptying anything. The moment you regret a cleanup is almost always the moment after it finishes.

Restore is one click and returns each file to its exact original location. If a file's origin is on a drive that isn't connected, the app names that drive, leaves the file safely in quarantine, and offers to continue with the rest — it won't redirect anything to a path that merely looks close enough. Each file is written in full before its quarantine record clears, so an interrupted restore can't leave a half-written file in your archive.

Restore is never locked. Not without a licence, and not after a refund. Files this software moved are always files you can move back.

Questions

Can I do this without buying anything?

Scanning, the Storage Map and the full CSV and HTML reports are free, with no account. You can run every scan on this page and see exactly what was found and how much space it would recover. The $34 licence is only needed to move flagged files into quarantine.

Is anything deleted automatically?

No. Nothing is deleted at any point by the software. Flagged files are moved to a quarantine folder on the same drive and recorded. Emptying quarantine is a separate action you take yourself, and restore and permanent delete are never locked behind the licence.

What happens if I quarantine something Lightroom has catalogued?

Lightroom will show it as missing until you restore it, which takes one click within the 24-hour hold or after it. The app excludes Lightroom's own working folders from scanning but has no knowledge of which photos are catalogued — catalog awareness is planned, not shipped.

Why does it say "Flag for Deletion" if nothing gets deleted?

Flagging marks rows so they can be moved in bulk afterwards; the file is untouched until you click Move Flagged to Quarantine, and quarantine is a folder rather than a bin. The label is more alarming than the behaviour.

Does it find duplicates of documents or music?

No. It looks at RAW photos, exported JPGs, MOV and MP4 video and GoPro WAV audio, and ignores everything else on the drive. It is not a general-purpose duplicate file cleaner — it is built for the one archive where a wrong deletion is unrecoverable.

How long does a scan take?

It depends on the archive, but the staged tests mean most files are eliminated on size alone without ever being read. A multi-terabyte pass can run for hours; it's checkpointed, so it can be paused and resumed rather than restarted.

Try it on your own archive

The free scan tells you how much of your drive is genuinely redundant before you commit to anything. On a 931 GB drive here, it found 78.8 GB of second copies — worth knowing either way.

Download and verify → What it does