Checking video uniqueness means measuring how much a copy differs from the source video and from other copies. To do this, frames and digital frame fingerprints are compared. ClipClone runs this check automatically for every copy: it calculates SSIM, pHash, dHash, PDQ and PSNR and marks the copy as passed or failed.
Why you can't check by eye
Two versions of a video can look the same to a person but differ for an algorithm — and the other way round. Platforms compare videos not by title or file size but by the content of the frames and the sound. That is why the difference between copies has to be measured in the same way: by picture similarity and by digital frame fingerprints.
Which metrics ClipClone calculates
| Metric | What it measures | How to read it |
|---|---|---|
| SSIM | structural similarity of the frames of the copy and the source | from 0 to 1; 1.00 means the frames are identical, the lower the value, the stronger the difference |
| pHash | difference of the digital frame fingerprint (perceptual hash) | how many of the 64 bits of the fingerprint did not match; more means a stronger difference |
| dHash | difference of the fingerprint based on brightness gradients | also out of 64 bits; complements pHash |
| PDQ | difference of the fingerprint by the PDQ algorithm, open-sourced by Meta | how many of the 256 bits did not match; the strictest of the fingerprints |
| PSNR | picture quality after the changes | in decibels; a value that is too low means the picture is degraded |
Fingerprints are calculated on several frames of the video, and the median goes into the report. This way a single random scene does not distort the result. Detailed definitions are in the guide “SSIM, pHash, dHash, PDQ and PSNR in plain words”.
Check thresholds
A copy passes the check only if all conditions are met. Below are the thresholds of the “Optimal” preset, which is the default; in “Strong” and “Strong + vertical” they are the same:
| Condition | Threshold |
|---|---|
| Similarity to the source, SSIM | no higher than 0.88 |
| Similarity to the most similar neighboring copy, SSIM | no higher than 0.92 |
| Difference from the source by pHash | at least 8 of 64 |
| Difference from the source by dHash | at least 8 of 64 |
| Difference from the source by PDQ | at least 55 of 256 |
| Difference from neighboring copies by pHash / dHash / PDQ | at least 5 / 6 / 35 |
| Picture quality, PSNR | no lower than 12 dB |
The presets differ in the strength of the changes, not in the strictness of the check. The thresholds of all four presets are in the guide “Presets for unique video versions”.
Each copy is compared not only with the source but also with 100 neighboring copies from the same run. This protects against a situation where two copies turn out nearly identical to each other.
An example of a real check
A latte art video 5.7 seconds long, the “Optimal” preset. All four copies passed the check:
| Copy | SSIM | pHash, of 64 | PDQ, of 256 | PSNR, dB |
|---|---|---|---|---|
| V-01 | 0.64 | 8 | 66 | 16.4 |
| V-02 | 0.65 | 8 | 72 | 18.2 |
| V-03 | 0.63 | 8 | 66 | 16.3 |
| V-04 | 0.66 | 10 | 80 | 17.4 |
You can see these copies next to the source on the home page.
What to do if a copy fails the check
Usually you don't need to do anything: the service makes a replacement by itself, free of charge, with up to 5 attempts for each ordered copy. If you still don't have enough copies:
- Look at the reason in the report on the copy: which metric did not meet its threshold.
- If the copy is too similar to the source, choose a stronger preset, for example “Strong”.
- If the copy is too similar to a neighboring one, just make more copies: the changes are random, and the new copies will turn out different.
- If the quality is low (PSNR), choose a milder preset: the changes are too strong for this video.
Other questions about the check are collected on the page “FAQ”.
Frequently asked questions
Does the check guarantee views in the feed?
No. The check shows how much a copy differs from the source and from other copies by measurable features. The platform decides what is shown in the feed.
Can I check my own video without processing?
The check is built into processing: the metrics are calculated for each copy the service has made. There is no separate check for uploaded files yet.
Where can I see the check result?
In the Dashboard, in the project. Each copy has a mark — “Unique”, “Too similar to source”, “Too similar to another copy” or “Low quality” — as well as the metric values and the thresholds.
Other guides
- Unique video versions: what they are and how they workWhat making unique versions of a video means, what changes in each version, how the difference between copies is checked and how to make hundreds of versions of one video in the cloud.
- How to make many versions of one videoHow to make many different versions of one video: by hand in a video editor, with software on your computer or in a cloud service. A comparison of the methods and step-by-step instructions.
- Unique video versions for Reels, Shorts, TikTok and VK ClipsHow to prepare different versions of a vertical video for Reels, Shorts, TikTok and VK Clips: the “Strong + vertical” preset, how the copies differ and what the platforms say.
- Presets: Standard, Optimal, Strong, Strong + verticalClipClone presets “Standard”, “Optimal”, “Strong” and “Strong + vertical”: what changes in the copies, which check thresholds each one has and which preset to choose.
- SSIM, pHash, dHash, PDQ and PSNR in plain wordsWhat SSIM, pHash, dHash, PDQ and PSNR are: plain-language definitions, what each number in the report means and which check thresholds ClipClone uses.
- How much unique video versions cost and how output minutes are countedHow the cost of unique video versions is calculated in ClipClone: output minutes, calculation examples, plans from 2,990 ₽ per month, price per minute and the free trial.