How to review duplicate photos without losing the best copy

Two similar photos are not necessarily interchangeable. The best frame can differ in expression, sharpness, crop or the edit you intended to keep.

By IndepApps · Updated

Separate three decisions

Exact duplicates, near-duplicates and unwanted photos are different categories. Exact files can often be checked mechanically. Near-duplicates need visual judgment. A blurred photo can still be the only record of an important moment, so a blur score should not decide for you.

Before cleaning a large library, confirm that you have a backup and understand whether deletion syncs to other devices. Start with a small set of photos whose history you know.

Try the built-in option first

Apple Photos has a Duplicates collection with a merge operation. Read Apple’s duplicate-merging instructions for the behavior of your macOS version. If that resolves your problem, an additional tool may be unnecessary.

Sift adds perceptual grouping, sharpness analysis and a review workflow. Its direct edition requires macOS 14 or later on Apple silicon or Intel. Its suggestions still need your judgment.

Check each candidate group

  1. Compare full-size images, not only thumbnails.
  2. Check focus, expression, framing and intentional background blur.
  3. Inspect resolution, date, location and any edits you want to retain.
  4. Keep both files when they serve different purposes, such as an original and a finished edit.
  5. Review the selected removals again before confirming.

For a first test, include an exact copy, an edited version, two similar frames and a unique image. Record which groups the tool suggests. Do not turn a suggested group into an automatic deletion rule.

Confirm recovery before scaling up

Sift documents removals through Recently Deleted. Test recovery with a disposable photo first and follow the recovery period shown by Photos. Do not permanently empty that collection just to prove a cleanup worked.

Use the check sheet to record the expected groups and your observed results. This is a review protocol, not a measured accuracy claim. See Sift versus Photos duplicate merging for the decision between those workflows.

A keeper decision you can explain

Imagine two portraits from a burst: the first is sharper, but the second has the expression you want. They are similar files with different value, so keep both until you decide their purpose. In a second pair, one image is a lower-resolution export of the same edited photo; check whether the full-resolution version also contains the edit before discarding the export.

For a blurred flag, inspect the subject rather than only the background. Deliberate motion or shallow depth of field can be intentional. Write a short rule for the current group—such as keep all original edited portraits—then apply it only to the images you have reviewed.

See Sift's review scenarios or compare photo and general-file cleanup tools. These examples are illustrative, not a customer case study.