Settings

Settings: Detection Settings

Control how thoroughly Duplicate Detection scans for duplicate and similar files

Overview

Detection Settings control how the Duplicate Detection process decides that two files are the same image. They set the accuracy level — which perceptual-hash algorithms run and how strict matching is — and whether videos and RAW+JPEG pairs are included.

These settings shape accuracy and speed; they never delete anything. Every match is still presented in a review group where you choose the copy to keep.

When to Use

Adjust Detection Settings before running Duplicate Detection to match your collection and how carefully you want it scanned. Keep Balanced accuracy for most libraries. Choose Fast when you have a very large collection and want a quick first pass, or Thorough when you are cleaning up an important archive and want to avoid mistaking similar-looking photos for duplicates. Turn on Include video files when your collection contains videos you want checked, or turn it off to scan images only and finish faster. Leave Detect RAW+JPEG pairs on if you shoot RAW and keep processed copies alongside your originals. Settings apply to the next scan, so you can fine-tune them and run again if the results are too strict or too loose.

Requirements

  • No special requirements - Detection Settings are available immediately

Settings

Every control on this tab, grouped as it appears in Settings.

Accuracy Level
Detection AccuracySegmented · Fast / Balanced / Thorough · default Balanced

Sets the accuracy level — which perceptual-hash algorithms run and how strict matching is. Fast uses one algorithm (aHash); Balanced uses two (aHash + dHash) and must have both agree; Thorough uses all three (aHash + dHash + pHash) and requires two of the three to agree.

When to useKeep Balanced for general use — a reliable mix of speed and accuracy. Choose Thorough for a critical cleanup where false matches must be minimised; choose Fast for a quick pass over a very large collection where occasional false matches are acceptable.
ExampleCleaning an irreplaceable archive, you switch to Thorough so resized or re-exported copies are caught and unrelated frames aren't wrongly paired.

Higher accuracy is slower — roughly 1× (Fast), 1.5× (Balanced), and 3× (Thorough). The card lists the active algorithms and the matching threshold for the chosen level.

File Types
Include video filesToggle · default On

Scans video files for duplicates by extracting and comparing frames.

When to useLeave on for mixed photo-and-video collections. Turn off to scan stills only and speed up a photo-heavy run.
ExampleA wildlife project with thousands of clips keeps this on; videos are compared using 3 frames on Fast, 5 on Balanced, and 7 on Thorough.
Related Files
Detect RAW+JPEG pairsToggle · default On

Flags a RAW file and its processed counterpart as “Related Files” rather than duplicates. Processed counterparts include JPEG, HEIC, TIFF, and PNG.

When to useKeep on for any RAW workflow so a CR3 and its JPEG export aren't treated as a duplicate to delete. Turn off only if you never keep RAW + processed pairs.
ExampleDSC_4821.CR3 and DSC_4821.jpg are grouped and labelled “Related Files” rather than duplicates, so the intentional pair is clearly distinguished. They stay selectable for deletion, and if you mark any related file, a “Delete Related Files?” confirmation warns you before it is removed.

Filenames need not match for a pair to be detected.

Tips & Important Notes

  • Detection Accuracy is a set choice of Fast, Balanced, or Thorough - there is no free similarity slider.
  • Thorough matches more strictly, so fewer look-alike photos are flagged as duplicates; Fast matches more leniently and runs quicker. Balanced suits most collections.
  • Duplicate Detection reports its findings in three groups: Exact Duplicate, Visual Duplicate, and Related Files.
  • Settings save automatically and apply to the next scan, so adjust them and run again if the results are too strict or too loose.