Everything it finds, and everything it reads

The complete list — every file type Deduplo opens, which ones it compares by content and which ones it can only compare byte-for-byte, and what it deliberately leaves alone.

Deduplo 1.0.0 · Windows 10 (21H2+) & 11 · last updated 14 September 2026

What counts as a duplicate

Most tools compare bytes. Two files are duplicates only if they are identical down to the last byte — which means a photo you resized once, or a video your phone re-encoded on the way to a chat app, looks like a completely different file. Deduplo compares bytes too, and then compares what the file actually is.

One scan surfaces all of these relationships:

Photos and camera RAW

Fully compared by content — these are matched even after resizing, re-compression, rotation and cropping. 70 FORMATS

Camera RAW files are matched from the preview the camera embeds, and decoded from the sensor data when a file carries none. 34 FORMATS

Five image types are matched byte-for-byte only — identical copies are still found, but a resized version of one will not be:

Video

Matched by what the footage looks like over time, so the same clip is found after re-encoding, a resolution change or a trim. Clips of genuinely different length are kept apart rather than merged. 37 FORMATS

A container using a codec your machine cannot open is skipped for content matching and falls back to exact matching, exactly like any unreadable file.

Audio

Every audio file is matched byte-for-byte and by its tags — same artist, same title, same album. 38 FORMATS

On top of that, the common formats — MP3, FLAC, WAV, OGG, OPUS, AIFF, M4A and WMA — are also compared by how they sound. That is what finds the same song ripped at 128 kbps and at 320 kbps, or the same track with the tags rewritten, which tag matching alone would miss entirely.

Documents

Documents are compared by what they say, not by how they are formatted — so the same report saved as a Word file and as a PDF is recognised as a near-duplicate. Small edits do not break the match. 189 EXTENSIONS

* .ts is treated as video, not as source code — the same extension is used by both, and a transport-stream recording is far more likely in a photo library.

Inside ZIP, TAR and 7z files

Deduplo looks inside .zip / .tar / .7z archives and matches what it finds there against the loose files on your drive — both byte-for-byte and by content. That answers the question nobody else answers: "have I already backed these photos up?"

Other archive formats — .rar, .cab, .iso and the rest — are not opened. To the scan they are ordinary files: with Other files ticked, their exact duplicates are still found.

What it will not touch

Some folders have nothing to clean and a lot to break, so the scan walks straight past them:

Anything not in one of the lists on this page is simply not part of a scan — Deduplo will not report it, and cannot delete it — unless you tick Other files, which then matches any file byte-for-byte.

Checked twice

After a scan, a deep re-check re-examines the borderline sets and throws out the ones that only looked like duplicates — they are listed separately, so you can see what was rejected instead of wondering.

You decide how strict it is

• A match-strength slider sets how alike two files must be to count as copies. Move it after a scan and results regroup instantly — nothing is read from disk again, so trying 95 and dropping back to 90 costs seconds, not another scan.
• A separate Match badge filter hides sets below a percentage you pick. The slider decides what Deduplo finds; the filter decides what you are shown.
• "Why did these match?" — right-click any set and Deduplo explains, in plain words, what made it group those files. If the answer does not convince you, mark them Not duplicates and that pair is left alone in every future scan.

One scan, five tabs

The same scan can also be grouped by time, by place or by burst — photos that belong together but are not duplicates, and are never ticked for you.

Select any two photos and press Compare for a side-by-side, or Pixel diff to see exactly where they differ — usually the fastest way to settle a set of near-identical shots. You can also sort sets by size, count or date, hide files below a size, leave a note on a set, and mark sets as reviewed as you work through them.

Whole-folder answers

Some questions are not about single files:

• Folder copies — whole folders duplicated somewhere else, found in one click.
• Redundant folders — folders whose every file already exists elsewhere, so the folder itself can go.
• Folder breakdown — how much space each part of your library could give back.
• Consolidate — moves the best copy of every set plus every unique photo into one folder you choose, and sends the leftovers to the Recycle Bin. A library scattered over four drives ends up in one place. It is a single undo step that moves everything back.

Choosing what to keep

Finding duplicates is the easy half. Deciding which copy survives is the half that decides whether you trust the result.

• Every set arrives with a keeper already chosen, ranked by sharpness, resolution and file size — the copy you would probably have picked yourself.
• You can override it on any set with one click, or set a rule that applies to all of them — keep the best quality, the sharpest, the highest resolution, the largest, the newest or the oldest, or let it weigh them all.
• Sets of genuinely different photos — burst runs, RAW+JPEG pairs, places, dates — are labelled so that no bulk action can auto-tick them. They are for browsing, not for deleting.
• Not duplicates? Mark a set that way once and it is never proposed again, in this scan or any future one.
• Protected folders. Mark a folder protected and nothing inside it is ever ticked — not by any automatic rule, not in any bulk action. For your master library, your originals, or anything a mis-click must not reach.
• Bulk select across the whole result list by condition — in a folder, by file type, larger than, older than. The keep-one rule still applies to everything it does.

Deleting safely

Scale and speed

• Incremental re-scans. The first scan reads everything; after that only what changed is looked at again.
• All your cores. Files are processed in parallel, with the number of workers chosen from how much memory your machine actually has.
• Multiple drives at once. Each physical drive gets its own read budget — few concurrent reads on a spinning disk, many on an SSD — so a scan can span drives without thrashing any of them.
• It changes strategy as libraries grow. Above a hundred thousand files or so it changes how it groups, so the comparison fits the memory your computer has. You do not choose this; it happens on its own.
• Huge result sets stay responsive. Results are paginated, and far-off sets are unloaded until you scroll back to them.
• Pause and stop any scan without losing the work already done.
• Watch folders and scheduled scans. Deduplo can keep an eye on your folders and tell you when new duplicates appear, or run a scan every few hours or weekly while the app is open — a scan it missed while closed runs at the next launch.
• Save, load and export. Store a result list — the sets and their files — and reload it later without scanning again. Export to CSV for a spreadsheet, HTML for a readable report, or JSON for your own scripts.
• Insights — scan statistics, the folder breakdown, and a heatmap of when your duplicates were created.

Privacy

Deduplo does its work on your own computer — your photos, videos and documents are read locally and never uploaded (no file, thumbnail, filename or scan result ever leaves your machine). It connects to the internet for one thing only: to activate and check your licence, sending just your key, a salted hash of your machine ID, the app version and your OS over an encrypted connection. Not for updates, not for analytics, not for crash reports, and never a file.

The working data it keeps — the fingerprints that make second scans fast — stays in your own user folder and is encrypted on disk. You can wipe all of it from Settings ▸ Clear cached data. Read the full privacy policy.

The interface ships in 16 languages: English, 中文, हिन्दी, Español, العربية, Français, বাংলা, Português, Русский, 日本語, Deutsch, 한국어, Italiano, Bahasa Indonesia, Türkçe and اردو.

What the licence includes

There is one licence and one price — $29.99, paid once. There is no higher tier and no add-on. Finding duplicates is free for ever, at any depth; what the licence buys is acting on what was found.

Everything else on this page is free, with no time limit and no account: scanning at any depth, edited-copy search, the folder-copy finder, watch folders, insights, opening a saved scan and the advanced tuning.

The licence is activated online the first time you enter your key, never expires, and covers one person and up to three computers that person uses. Updates to version 1 are free — for as long as version 1 is maintained. A future major version is a separate product; the version you bought stays yours and keeps working. Buying for a team or an office? Talk to us about bulk licences.

Not here yet

Things people ask for that Deduplo 1.0 does not do. They are on the list, without dates attached — we would rather ship them late than promise them early.

Your licence is for what Deduplo does today, on Windows. Nothing on this list is part of what you are paying for, and nothing here is promised for a particular date. If one of them is what you actually need, tell us — it genuinely changes what gets built first.