Watermarking

A marker burned into the video itself. Unlike encryption it prevents nothing — its job is to make a leaked copy point back at the account it came from, which turns an anonymous act into an attributable one.

Updated: July 2026 5 min read Security
In short

Watermarking stamps an identifying mark into the video frames themselves — a fixed logo, or a per-viewer ID — so it prevents no copying but makes a leaked recording traceable to the account it came from.

What is video watermarking?

Watermarking embeds an identifying mark into the picture. It survives screen recording, re-encodingTranscodingRe-encoding a source video into multiple resolutions and bitrates so a player can switch between quality levels on the fly.Open definition → and re-uploading, because it is part of the frames rather than metadata around them — which is exactly the point. Every other protection layer fails at the camera; this one keeps working.

It is worth being blunt about what that means: a watermark does not stop anybody. It changes the arithmetic for the person considering a leak, because the copy they hand over carries their own name.

Static and dynamic watermarks

TypeWhat it showsWhat it is for
StaticYour logo or fixed text, identical for every viewerBranding, and a mild deterrent against casual reposting
Dynamic (forensic) The individual viewer's name, email or account IDAttribution — tracing a specific leaked copy back to one account

The distinction matters more than it sounds. A static mark tells the world whose content it is; a dynamic one tells you whose copy it is. Only the second changes anyone's behaviour, because only the second makes the leaker identifiable.

Dynamic marks are usually rendered by the player over the video, often drifting position so they cannot be reliably cropped out, independently of the resolutionResolutionThe pixel dimensions of a video frame — 480p, 720p, 1080p, 2160p (4K). Determines how sharp the picture looks on a given screen.Open definition → being served. Truly forensic schemes go further and encode the identifier into the pixel data itself, invisible to the eye and recoverable from a re-encoded copy.

How a per-viewer watermark works

Identity arrives with the session

When playback is authorised, your backend supplies whatever identifies the viewer — name, email, account ID, seat number.

The mark is composed

That string becomes an overlay: position, opacity and movement pattern configured so it is readable in a recording but not intrusive to a paying viewer.

It rides on the frames

Because it is drawn over the video during playback, it appears in any screen capture, phone recording or re-upload made from that session.

A leak becomes evidence

When a copy surfaces, the mark names the account. That is usually enough to close the hole rather than guess at it.

What watermarking does and does not do

ThreatDoes watermarking help?Why
Screen recordingYes — the main caseThe mark is in the frames, so it survives into the recording
A phone pointed at the monitorYesThe one attack no encryption touches; the mark still reads
Re-upload to another platformYesRe-encoding does not remove a burned-in overlay
Downloading the original fileNoNothing is prevented — that is what DRMDRMEncryption of the video file plus a licence server that issues a per-session key — so a downloaded copy stays unplayable.Open definition → is for
Sharing the playback linkNoLink control belongs to signed URLsSigned URLA playback link carrying an expiry and a cryptographic token — it stops working once the window closes, and a tampered link never works.Open definition → and domain rules

Read the table as a division of labour rather than a scorecard. DRM stops the file being taken, signed links stop the URL being passed around, and watermarking covers the gap neither can reach — the moment a camera or a capture tool is pointed at a legitimate session.

Making it work in practice

A watermark only deters if viewers notice it. A mark hidden in a corner at five percent opacity satisfies a compliance checkbox and changes nobody's behaviour.

Make it legible, not obnoxious

The viewer should be able to read their own email if they look for it. That is the whole deterrent.

Move it

A mark that drifts between positions cannot be cropped or masked out of a recording with a fixed rectangle.

Say that it is there

Deterrence works before the leak. A line in the course terms noting that sessions are marked does more than any amount of opacity tuning.

Further reading

Denis Konnov

Written by

Denis Konnov Verified

CMO at Kinescope

Denis leads marketing and content at Kinescope, working closely with the platform's engineers. This glossary is reviewed and kept up to date by the team building Kinescope's streaming, encoding and CDN delivery infrastructure.

FAQ

Can’t find the answer you need? Check out our Help Center.

Embedding an identifying mark into the picture itself — either a fixed logo or a per-viewer identifier. It does not prevent copying; it makes a leaked copy traceable to the session it came from.

A static watermark is the same for everyone — typically your logo — and serves branding. A dynamic watermark carries the individual viewer's name, email or account ID, so a leak can be traced to one account. Only the dynamic kind is a real deterrent.

It stops nothing directly. It changes the incentive: someone sharing a recording that carries their own email on every frame is handing over evidence along with the video. In practice that deters the casual sharing which accounts for most leaks.

A fixed overlay in a corner can be cropped. That is why dynamic marks move around the frame, and why forensic schemes encode the identifier into the pixel data where cropping and re-encoding do not reach it.

Yes — that is its main purpose. Because the mark is drawn into the frames the viewer sees, it appears in any capture made from that session, including a phone pointed at the monitor, which is the one route DRM cannot close.

For paid content, yes. DRM prevents the file being downloaded and replayed; it cannot prevent someone filming their own screen. Watermarking is the layer that makes that last route attributable, which is why the two are normally deployed together.