Video Watermarking and the Battle against Piracy
By Mei Lam
In June 2017, full UHD Blu-Rays of Smurfs 2, Patriots Day and Inferno appeared on torrents sites. Not only did the pirates crack what was thought to be unbreakable encryption methods, this incident showed the increasing importance of tracking down offenders. This is where video watermarking technology fits in.
Video watermarking aims at combating post-decryption piracy. When the content theft occurs after decryption, Conditional Access System (CAS) and Digital Rights Management (DRM) solutions cannot help, and watermarking has proven to be an effective solution.
Piracy has a huge impact on the content industry; it affects revenue for all the players and threatens the integrity of content license agreements. If content is not adequately protected, service providers could face large fines and lose rights to content. Content owners, especially premium content owners, and distributors constantly need to find ways to better protect their assets.
While watermarking cannot directly stop piracy, it enables service providers to detect piracy, identify those who engage in it, and then do something about it. These reasons are why video watermarking has become more important to the video content industry: it is both a deterrent and a very powerful tool for managing content piracy.
What Is Video Watermarking?
Video watermarking embeds additional data (a digital message) within a media, so that this data will stick to the media all along its distribution path and copy operations, and makes this data difficult to erase, obfuscate or alter, without damaging the media itself. While we are referring to digital watermarking in relation to video, this technique can also be applied to audio and other media types.
Industry Adoption of Video Watermarking
First published in 2013 and updated in 2018, the “MovieLabs Specification for Enhanced Content Protection” suggest the following content protection requirements for forensic watermarking:
- The system shall have the ability to securely forensically mark video at the server and/or client to recover information necessary to address breaches.
- The watermark shall be robust against corruption of the forensic information, including collusion attacks, and transformations and capture techniques that leave the content still watchable.
- The watermark shall be inserted on the server or on the client such that the valid insertion is guaranteed during playback even if the device and its secrets are compromised.
- Forensic watermarks shall be tested for robustness by a third-party.
The specification put forward by MovieLabs indicates a serious effort on the part of the content owners to mitigate risks associated with piracy.
How Do Content Providers Use Video Watermarking?
The two main types of applications for watermarking are content tracking and forensic.
For tracking content, watermarking is used to embed information about the content itself (content identification), as well as possibly time codes. The applications can be multiple and include:
- Content usage monitoring (e.g. automated counting of broadcasts for commercials);
- Content integrity checking (e.g. automated checking that sequences with product placement have not been removed or altered);
- Audience metering; and,
- Source authentication.
For forensic purposes, watermarking is used to dissuade or trace piracy by embedding information that will allow identification of leakage sources.
Forensic watermarking can be visible (overt) or invisible (covert):
- Visible Watermarking: A simple overlay on the video (e.g. text characters), which does not intend to be unobtrusive.
- Invisible Watermarking: A complex alteration of the video designed to be imperceptible by the human eye, though computationally detectable.
Essentially, visible watermarking can help deter piracy, whilst invisible watermarking aims to remain undetectable by the pirates and increase the potential of detection. Both solutions support prosecution and are recognized by law enforcement agencies. For anti-piracy efforts, there is growing interest in digital watermarking for forensic uses.
Forensic Watermarking Applications
A forensic watermark can carry any information that may help identify the responsible party for a content leakage, such as the ID for a user or a subscription, VOD session, client device or smart card, a digital cinema projector, and even, a specific copy of a piece of content and the recipient of a live feed.
Four applications for forensic watermarking:
STREAMING SESSION OR USER ACCOUNT IDENTIFICATION
Used for unicast delivery (including live) and on-demand. At the time of playback, the ID of the user account or streaming session is embedded in the video. In the case of live, illegal streaming sites are monitored, and as soon as a user ID is retrieved, the streaming session can be taken down, and/or the user account shut down. If deemed appropriate, the individual behind the user account may be prosecuted.
After decryption and decoding the device inserts its identification before presenting the content, for example: 1) A set-top box can insert its serial number or the ID of the smartcard, allowing the individual owning the device or the subscription to be traced, and 2) For OTT, the player may insert information such as device type, OS, and player version, which can inform security changes, or to discontinue or degrade service on this device type.
Sender marks each copy or feed with a unique ID associated with the recipient, along with a legal warning. Used when a content owner shares content with third parties, for example: 1) Share movie copies with sub-contractors for dubbing or subtitling, or with commentators, and 2) Send live feeds to different networks for re-broadcast (B2B contribution). Can dissuade illegal sharing or re-distribution, as well as complacency and permissiveness with such practices.
Movies are distributed to theatres in high quality (encrypted) digital form to be projected by Digital Cinema Projectors, which must comply to a specification from DCI and be certified. Specification includes content security: encryption/decryption, key management and watermarking. Each projector embeds in the projected video the projector ID and a time code. Accompanied by enforcement processes, this method is very effective at managing screen-capture piracy (i.e., camcording).
How Pirates Tamper with Video Watermarks
Of course, even with content protection measures in place, pirates will look to find new and creative ways to circumvent security technologies like digital watermarking.
In the case of invisible watermarking, a pirate may attempt a series of transformations to a suspected watermarked video sequence, aiming to either make the detection of the watermark impossible, or the extracted watermark unreliable, and thus nullify its legal value or deem it erroneous.
Video pirates can use a multitude of tampering techniques when attempting to removing a watermark from a video. These manipulations include: geometrical transformations (flipping, zooming, cropping, trapezoidal, rotations); transcoding, changing framerate, filtering; modifying colours, saturation, luminance, quantizing; reverse engineering and alterations of watermarks, tampering of manifest (for server-based watermarking); and, collusion, or the mixing of several watermarked copies of the content.
Therefore, a large part of the robustness of a watermark technology relies on the secrecy of algorithms, difficulty of reverse engineering, unpredictability and spreading of marks in time and space, and “blindness” of pirates – meaning that pirates should never know when an attempted attack is successful or not.
So, for content owners and service providers not only is it important to ensure that they have a video watermarking solution in place, but also that this solution is robust to content manipulation.
The Future for Video Watermark Technology
There is clearly a global industry movement towards an increasing adoption of watermarking. Watermarking is seen as especially important for high value content such as UHD or live sports, which is constantly a target for video manipulation and illegitimate re-streaming.
Although video watermarking is far from being a ubiquitous technology, it is demanded increasingly by operators and content owners in markets with strong pirate activity. And because of pirates’ continuous attempts to break it, video watermark technology companies and content providers equally need to be vigilant and ensure their technology remains strong.
With its many effective applications, however, video watermarking can be a content provider’s strongest weapon in policing content distribution and the prosecution of pirates. <>
How can you check if a digital watermark technology is robust?
Cartesian tests the robustness of digital video watermarking technologies through our Farncombe Security Audit® Watermark. We developed this review through consultations with Hollywood studios, watermark solution vendors, MovieLabs, and the Motion Picture Association of America.