Your video library probably has more duplicates than you think. Not because you copied the same file twice — but because the same footage exists in multiple formats, resolutions, and encodings.
A 4K original, a 1080p export for sharing, a compressed version you uploaded somewhere. Same video, three files, taking up 3x the storage. But when you run a duplicate finder, it often says: "No duplicates found."
Why? Because most tools use exact matching. And exact matching has a fundamental blind spot when it comes to video.
The traditional approach to finding duplicates is straightforward: compare the raw data of each file.
The tool computes a checksum (like MD5 or SHA-256) for every file. This checksum is a fingerprint of the file's raw bytes. If two files produce the same checksum, they're identical — bit for bit, byte for byte.
This works well for:
Exact matching is fast and reliable for what it does. If you only have actual file copies — same bytes, same format — it's the right tool.
Here's the problem: in the real world, most video "duplicates" aren't exact copies. They're the same footage processed differently.
Consider how a single vacation video ends up as multiple files:
Original recording: vacation_4k.mov — 4K ProRes, 2.1 GB
Exported for sharing: beach_edit.mp4 — 1080p H.264, 340 MB
Uploaded to social: clip_720p.mkv — 720p VP9, 85 MB
Same footage. Three files. Three different codecs, resolutions, containers, and file sizes. An exact-match tool computes three completely different checksums and says: "No duplicates."
This happens constantly with:
Perceptual hashing takes a fundamentally different approach. Instead of fingerprinting the file data, it fingerprints what the video looks like.
Here's the process:
Frame sampling. The tool decodes the video and extracts frames from multiple points throughout the footage — not just the first frame or a single thumbnail.
Visual fingerprinting. Each sampled frame is reduced to a compact visual hash — a representation of what the frame looks like at a structural level, ignoring pixel-level details like compression artifacts or resolution.
Similarity comparison. The tool compares visual fingerprints across all files. If two videos' fingerprints are close enough, they're flagged as duplicates — along with a similarity percentage.
This means a 4K ProRes file and a 720p H.264 file of the same footage produce nearly identical visual fingerprints — because they show the same thing.
Not all perceptual hashing is equal. A single-pass approach can produce false positives — two videos with similar intros but different content might match. Dup solves this with a two-pass system:
Pass 1: Primary hash. Samples a 30-second window from the video and generates a broad visual fingerprint. This is fast and catches candidates with a generous tolerance.
Pass 2: Duration + verification hash. For each candidate pair, Dup first checks whether the videos have similar duration (within 5% or 5 seconds). Then it samples a second section from the middle of the video — a completely different part — and generates a tighter fingerprint. Only videos that pass both checks are confirmed as duplicates.
Why two passes matter: A TV intro might look similar to many episodes, but the middle content will differ. Two-pass verification eliminates these false matches while keeping the genuine ones.
| Capability | Exact-match tools | Dup |
|---|---|---|
| Same file, same name | Yes | Yes |
| Same file, different name | Yes | Yes |
| Same video, different codec | No | Yes |
| Same video, different resolution | No | Yes |
| Same video, different container | No | Yes |
| Same video, different bitrate | No | Yes |
| Duration-aware verification | N/A | Yes |
| Similarity percentage | N/A | Yes |
To be fair: if all your duplicates are actual file copies — same bytes, same format, same everything — then exact matching is perfectly fine. It's faster, simpler, and there are many free tools that do it well.
Exact matching is likely sufficient if:
You need perceptual matching if any of these apply:
In other words: if you've ever done anything with your videos besides copy-paste them, you almost certainly have duplicates that exact matching will never find.
The average video library has 5-15x more duplicate footage than exact-match tools report.
Based on real-world tests — 47 vs. 3 duplicates in the same 200 GB folder
Dup is free on the Mac App Store. Point it at your video folder and compare the results to whatever tool you're currently using. The difference is usually significant.
All processing happens locally on your Mac — no uploads, no cloud, no tracking. Your files stay on your device.
Find the duplicates your current tools miss
Dup finds similar photos and videos — not just exact copies. Free on the Mac App Store.