Watermarking

Every copy carries a mark. Every leak traces back.

An invisible marker embedded in a broadcast feed or a distributed file — so when a leak turns up, it can be traced back to the exact source it came from. Runs on two tracks under the hood: one built for a signal that's live right now, one built for a file that hasn't released yet.

Two tracks, one engine

Different problem, same forensic mark underneath.

A leaking live signal and a leaking pre-release file get stolen differently, so they're watermarked differently too — but both come out of the same engine, and both trace a leak back the same reliable way.

Live

Built for the moment it's airing.

For content most likely to be leaked in real time — a live sports event, a breaking broadcast. A forensic watermark is embedded into every frame, imperceptible to the viewer, while the signal is still going out.

Format support today
  • Codec: H.264
  • Container: MPEG-TS
  • Transport: SRT, UDP, or HLS (with CMAF support)

Best for: Live sports, breaking broadcasts, any single-source feed

VOD

Built for what hasn't released yet.

For pre-release and catalog assets — an unreleased season, a screener list, a review copy. Same underlying technology, delivered as a file instead of a live signal, so it fits into a distribution workflow instead of a broadcast one.

Format support today
  • Codec: H.264
  • Container: MPEG-TS
  • Delivery: Your S3 bucket, or your own cloud storage, over HTTP

Best for: Pre-release screeners, back catalog, review copies

Live, two ways

Delivered at the edge,
or built into the encoder.

Which one fits depends on what you're optimizing for — reaching every individual viewer, or preserving every bit of encode quality on a feed shared to a handful of receivers.

At the CDN edge

Session-level watermarking

We produce two versions of your stream — mechanically distinct, visually identical. Supported CDN partners serve each viewer their own unique mix of the two at the network edge, so every individual viewer's copy is traceable back to them, with no extra hardware on your end.
In the encoder

Embedded, hardware-level watermarking

For broadcasts where encode quality matters most, the mark is embedded directly in the encoder itself instead of a separate transcode pass — the way to go when the same source feed goes out to several receivers (affiliates, distribution partners) that need to stay distinguishable from one another.
When a leak turns up

Recovered, not just recorded.

A leaked clip — live or VOD — gets run back through the same engine to decode the mark. What comes back is a confidence level and the recovered watermark id, not just a yes-or-no on whether a mark exists. A longer clip gives the decoder more of the mark to work with, so it's the most reliable way to get a clean, fully-recovered id back.

Availability

Included as part of enterprise coverage.

Watermarking is scoped in as part of your enterprise engagement — see Enterprise pricing for how it fits into your coverage.

Watermarking FAQ

Common questions, answered.

It's an invisible, unique mark embedded into the video itself. Each recipient — a partner, a platform or a viewer session — gets its own marked copy. If a copy leaks, we read the watermark and tell you exactly which copy it came from.
A visible logo tells everyone who owns the content, but it says nothing about who leaked it, and it's easy to blur or crop out.
No, it complements them. DRM controls who can open content; watermarking tells you who let it out once it has leaked, for example through screen recording, re-streaming or a shared file.
No. The watermark is designed to be imperceptible — no visible patterns, flicker or color shifts — and audio is not touched. Picture quality stays in line with the original.
Marked content is delivered in the same format, at a bitrate matched to your original, so it fits into your existing players, platforms and workflows.
Live watermarking is built for content while it's airing and adds only minimal processing delay. The exact figure depends on your setup, and we confirm it during onboarding.
No, the watermark is impossible to remove. A pirate's attempts to remove it damage the content so much that the copy loses its value.
No. Reading a watermark requires secret keys that only Redflag holds, and the keys are unique to each customer.
No. Creating a valid watermark requires secret keys, so a pirate can't plant another recipient's ID in a copy.
Yes — that's what it's designed for. Pirated copies are almost always re-compressed, resized or converted before they're shared. The mark lives in the picture itself, not in the file, so it survives those changes.
Screen recording and re-streaming capture the picture, and the picture carries the watermark. Those copies remain traceable.
Mixing copies is a known technique for trying to confuse watermarks, and our approach takes it into account.
Send us a suspect copy, or a link to it, and we analyze it and return the recipient ID along with a confidence level. If you also use Redflag's monitoring service, leaked copies are found and checked automatically.
Longer clips give more reliable identification — a full program or a clip of 3 minutes or more is ideal.
Most files are analyzed in minutes, so you can act while a leak is still spreading.
Accuracy matters most here, so every result comes with a confidence level, and we only report a match when the evidence is strong. Content without our mark doesn't produce a match.
Yes. Each detection comes with a clear report: the recipient ID, the confidence level and supporting evidence.
Live watermarking supports H.264 video over MPEG-TS, delivered by SRT, UDP or HLS. VOD watermarking supports H.264 video over MPEG-TS, MP4 and MKV files delivered over HTTP or cloud storage such as S3. Contact us if you need other formats.
Watermarking runs either in your encoder or at the CDN edge for live content, and as part of file delivery for VOD. You keep your existing delivery path, and we add the watermark along the way.
Effectively unlimited. You can give a unique watermark to every partner, platform or viewer session, and adding recipients doesn't slow delivery down.

Trace the next leak back to its source.

Tell us what you're distributing and how — a broadcast feed, a screener list, a back catalog — and we'll scope the coverage that fits.