Guide
Forensic watermarking explained
A forensic watermark is a hidden mark built into a video, different in each copy. If a copy leaks, the mark helps show which copy it was.
Forensic watermarking explained in short: you give each person their own copy of a video, and each copy carries a different hidden code. The code does not stop anyone copying. It helps you trace a leak afterwards.
What is the difference between a visible mark and a hidden mark?
Both are watermarks. They do different jobs.
| Type | What you see | What it is for |
|---|---|---|
| Visible mark | A name, a logo or a line of text on the picture | Deter. People think twice when their name is on screen. |
| Hidden mark | Nothing | Trace. It still carries the answer when the visible mark is covered. |
A visible mark is easy to read, and easy to crop out. A hidden mark is hard to see, so a person who wants to leak the clip cannot just cover it. Good practice is to use both.
How does a hidden mark work?
The simplest way to put a mark on a video is to lay it on top, like a sticker. A hidden mark is different. It is built into the picture itself.
Here is the plain idea. The system makes tiny changes to the picture, spread over many frames. The changes are so small that people do not notice them. But a program that knows what to look for can find them. Each copy gets a different pattern of changes, so each copy carries a different code. The code is linked to the person the copy was made for.
Because the code lives in the picture and not on top of it, covering a name or cutting a corner does not remove it.
How is a leaked clip read?
You take the leaked clip, run it through a reader, and compare the pattern it finds to the codes you issued. If 1 code matches clearly, you know which copy it came from. If the clip is poor, the reader may only be partly sure. A careful system says how sure it is instead of giving a flat answer.
Where is forensic watermarking used?
It is used wherever a valuable video goes to a small group before it is public.
- Film and TV. Early copies sent to reviewers and partners are often marked per person.
- Sport broadcast. Live and recorded feeds can carry marks that help trace restreams.
- Sensitive internal video. Board briefings, results updates and team videos can be marked per viewer.
The tools and the scale differ, but the idea is the same: every recipient gets something slightly different.
What can a hidden mark survive?
Designers build hidden marks to last through everyday changes. To different degrees, a good mark can stay readable after:
- re-compressing the video, as most platforms do when you upload
- resizing it to a smaller or larger picture
- cropping part of it
- someone filming the screen with a phone
"To different degrees" matters. A little re-compression is easy to survive. Filming a screen in a dark room is much harder.
What can weaken or break a mark?
Heavy re-processing can weaken a mark. Many rounds of re-encoding, strong filters, heavy cropping, slicing a clip into small pieces, or blending several copies together all make the code harder to read.
Research shows that with enough effort, a mark can be damaged. No honest vendor can promise a mark cannot be removed, or that it works every time. That is why a mark should be 1 layer among several, not the only defence.
The key point: a hidden mark shows whose copy leaked. It is strong evidence when it agrees with other clues. It does not, on its own, show who leaked it or why.
Why does the trail matter as much as the mark?
A mark points to a copy. The trail around that copy tells you more. It records who signed in, on what device, from which network, and at what time.
Why does this help? A person's account can be shared or hacked. If the mark says a copy leaked, and the trail shows it was only ever opened by 1 phone on 1 home network, the case is much stronger. If the trail shows 4 devices and an unusual country, you have a different story. Several clues that agree make the strongest case. For the steps, see how to find out who leaked a video.
What can a watermark not do?
It helps to be clear about the limits.
- It does not stop anyone from filming, saving or sharing a video.
- It does not show who leaked the clip. It shows which copy.
- It does not work on a clip that is too short, too dark or too damaged to read.
- It does not replace good habits, such as sharing with fewer people and switching off access that is no longer needed.
What should you ask before you rely on a watermark?
Whether you build or buy, a few plain questions tell you a lot.
- Is every copy really different, or only some of them?
- What happens when the clip is poor? Does the system say how sure it is?
- Is there a trail of sign-in, device, network and time around each copy?
- Can you switch off 1 person's access quickly?
- Can you export a report to give to your lawyer?
Be wary of any answer that sounds like "never" or "always".
How is forensic watermarking different from DRM?
DRM tries to stop copying or playing outside set rules. A watermark does not stop anything. It helps you find out afterwards. They answer different questions, and many teams use both. Our guide watermark vs DRM compares them in detail.
How does SealCopy use forensic watermarking?
SealCopy gives each person 1 private link. They sign in with a code, their name moves across the picture, and a hidden mark is built into the copy. Every view is logged. If a clip leaks, you upload it, we read the mark, and the case report says which copy it was, how sure we are, and the trail around it. A very poor recording can make the mark harder to read, and then we say so. See how SealCopy works for the full path.
What can you not afford to see on the news?
Tell us what you are trying to protect. We reply by email, and we show you a private demo.