Re-uploading video: what uniquifying actually does

Updated: August 25, 2026

The same clip rarely lives in one place. It gets re-uploaded to another platform, published again months later, prepared in several variants for different accounts. And nearly always the same question comes up: will the platform notice the video has been here before.

It will, if nothing changes. Let us go through how platforms look for duplicates, what can be changed in a file and what cannot, and how to use that without illusions.

How duplicates are detected

• File hash. The crudest method: byte-identical files produce the same checksum. Change one byte and the match is gone. That is why "just rename the file" does nothing while any rebuild does.
• Container metadata. Capture date, device model, encoder name and, for phone footage, sometimes coordinates. Not a fingerprint of the picture, but a reliable way to link uploads to each other and to your device.
• Perceptual fingerprints. Frames are shrunk to a tiny matrix and turned into a short code that survives brightness changes, noise and re-encoding. Hence an important conclusion: the popular advice to "add noise and nudge the brightness" achieves nothing — those filters are exactly what such fingerprints were designed to ignore.
• The audio track. It has a fingerprint of its own, often more robust than the video: music is recognised even in a heavily altered clip.

What our uniquifier does

The Video uniquifier turns one file into several different copies by changing what actually changes the picture, rather than what the algorithms shrug off:

• geometry — a small offset crop and a zoom back: the frame really shifts and its edges leave the view;
• tempo — a micro shift of a few percent, with video and audio stretched or compressed together;
• colour — slight changes in brightness, contrast and saturation;
• metadata — the container is rebuilt from scratch: capture date, device model, coordinates and encoder name are not carried over.

Parameters for each copy are chosen at random within the selected mode, so copies differ from the original and from each other. Two copies made in a single run are not identical — which matters when the clip goes to different accounts.

How to use it, step by step

1. Open the Video uniquifier.
2. Upload a clip in MP4 or MOV. The free plan allows up to 60 seconds and 50 MB; Premium allows 90 seconds and 100 MB.
3. Choose a mode: soft, hard or extreme.
4. Set the number of copies: up to 3 copys on the free plan, up to 5 copys on Premium.
5. Wait for processing. This takes longer than metadata cleaning: every copy is rebuilt from scratch and the time scales with clip length.
6. Download the copies one by one, or take them all at once with "Download all as ZIP".

Finished copies are deleted from the server after download, and unclaimed ones are removed automatically after two hours. Download them right away.

Which mode to choose

Soft changes the frame minimally: the shift and the speed change are barely visible and the clip still looks like the original. Good when the picture must stay intact — for example when it contains text or an interface that cannot be cropped.

Hard is the sensible default: the changes are visible in a side-by-side comparison but do not stand out on their own.

Extreme changes the frame noticeably: a larger crop, speed and colour pushed further. Reasonable when the edges of the frame are empty and expendable, and the result matters more than fidelity to the original.

There is no universal answer: the stronger the change, the less the copy resembles the source — and the more the viewer notices. Start with hard and judge the result.

What uniquifying does not do

We do not promise to defeat copy-detection systems such as Content ID, and we could not guarantee it: platform algorithms are closed, they also weigh the audio, and they change without notice. Any service promising guaranteed evasion is promising something it cannot verify.

The honest wording is this: the file stops being the same file, its metadata is gone, and the picture is changed geometrically and in tempo. Whether that is enough on a particular platform depends on the platform and on the clip itself.

A word on audio: we change the tempo of the track along with the video, but we do not replace the music. If a clip is recognised by its soundtrack, the work has to happen on the audio, not the picture.

If you only need the metadata gone

Sometimes uniquifying is not needed at all: the clip is being published for the first time and the only task is to remove the capture date, device model and coordinates. The Metadata remover does exactly that — the video is not re-encoded, the container is rebuilt with no quality loss and much faster, and the picture stays frame for frame identical.

A simple rule: publishing for the first time — metadata cleaning is enough. Publishing again and wanting the file to differ — that is what the uniquifier is for.