Dynamic Watermark
A dynamic watermark is text rendered into the video stream that’s unique per viewer — typically the viewer’s email address, account ID, or IP. Unlike a static logo (same for everyone), a dynamic watermark identifies the specific viewer the video was delivered to.
Why this changes anti-piracy
Most paid-content piracy starts with a single buyer who shares the link. Static watermarks don’t help — every viewer sees the same logo, so a leaked video is anonymous.
A dynamic watermark makes each delivered playback uniquely identifiable. If a course is reposted to a piracy site, screenshot a frame, read the watermark, identify the leaker. Most leakers stop sharing once they realize they’re identifiable.
How it’s rendered
Two approaches:
1. Pre-encoded burn-in. Render the watermark text into the video frames before HLS encoding. Each viewer gets a different encoded copy. Strongest protection (impossible to remove without re-encoding the whole video). Cost: separate encode per viewer = expensive.
2. Player-side overlay (more common). Render the watermark text via the player on top of the video frames at playback time. Cheaper because all viewers share the same encoded video; only the overlay differs.
Player-side is what most platforms (including AVCaption) use. It’s defeatable by clever JS injection but raises the bar enough that most casual leakers don’t bother.
Best practices
- Move the text periodically — every 30-60 seconds — so simple cropping can’t hide it.
- Position to overlap subject — corner watermarks crop off; partly-overlapping watermarks force the cropper to lose video area.
- Semi-transparent — readable but not so loud the legitimate viewer hates it.
- Include both text and a unique nonce — text is human-readable, nonce is server-traceable to the original token.
Forensic vs visible watermarks
A forensic watermark is invisible to the human eye but detectable by an algorithmic decoder. Used in cinema (Cinavia) and high-end OTT to trace pirated copies. Strong, expensive, requires specialist tooling.
A visible dynamic watermark is human-readable and serves a deterrent function. Cheaper, simpler, effective for most paid-content scenarios.
AVCaption ships visible dynamic watermarks on Enterprise.