Duplicate video finder

Find duplicate and visually similar videos before deleting anything.

VideoCull finds exact copies and visually similar files, even when the filenames differ. Each group shows the paths, thumbnails, similarity, and quality details, so you choose the keeper before anything is marked.

  • Compare all loaded videos or the current filtered set
  • Exact-file checks plus visual or pHash comparison
  • Suggested keepers can always be changed
  • Duplicate review never deletes files automatically
Duplicate review · compare evidence before deleting
VideoCull duplicate review comparing open movie encodes with similarity scores, file metadata, playback controls, and suggested keepers

From scan to decision

A separate review pass for repeated files.

Duplicate results do not get mixed into an automatic cleanup action. You work through the groups first.

  1. 01

    Choose the comparison scope

    Run across every loaded folder or narrow the job with the current grid filters.

  2. 02

    Inspect each group

    Compare sampled frames, paths, duration, resolution, bitrate, frame rate, file size, and playback.

  3. 03

    Confirm or replace the keeper

    Use the suggestion, pick a different file, protect an existing Keep decision, or dismiss a false match.

  4. 04

    Mark the unwanted copies

    The selected files become Delete decisions. They remain on disk until the normal final delete action is confirmed.

How matching works

Filenames and file sizes are not enough.

The same clip can be saved twice under different names, and a re-encoded copy has a different size and checksum. VideoCull combines an exact file check with a comparison of what sampled frames actually look like.

What it will not catch

Frames are compared at matching points in each file, and durations must fall within a set tolerance. Trimmed versions, re-edits, or a clip cut from a longer video usually do not match.

  • Exact copies

    Files with the same size and a similar duration are hashed in full with SHA-256. Identical bytes form an exact group, whatever the filenames say.

  • pHash comparison

    The default method. Frames sampled across each video are reduced to a perceptual hash and compared. It is strict and produces few false positives.

  • Visual similarity

    Compares downscaled grayscale frames directly. It catches more alternate encodes, resized copies, and near-identical footage, at the cost of more false positives.

  • Mirrored copies

    An optional flipped comparison also checks horizontally mirrored versions of the same footage.

Keeper suggestions

The app supplies evidence, not a verdict.

By default, VideoCull ranks files by resolution, video bitrate, duration, frame rate, and file size. The first meaningful difference selects the suggested keeper.

No automatic deletion

Suggested keepers only help with selection. You still inspect the group and confirm the final Delete batch yourself.

  • Correct a false match

    Save two files as “not a match,” exclude one video from a group, or dismiss the complete group. Ignored pairs stay excluded from later runs.

  • Tune difficult libraries

    Adjust similarity, sample count, duration tolerance, sampling window, black-frame handling, and flipped comparison when the defaults do not fit the footage.

  • Play before choosing

    Open any result in Review mode with only that duplicate group as the current navigation scope.

VideoCull for Windows

Start with the folder you keep avoiding.

No account, no upload, and no separate FFmpeg installation.