As operators and content producers alike continue to raise security standards in the battle against unauthorized redistribution, we’ve seen a piqued interest in watermarking as a deterrent and a growing list of questions regarding its adoption. In a recent webinar, Petr Peterka, CTO of Verimatrix, and I sat down with Ron Wheeler from Twentieth Century Fox to answer your questions.
However, before I dive into the Q&A, I wanted to highlight the case study Ron shared during the session, detailing the results Fox has experienced regarding the use of watermarking with Super Premium VOD (SPVOD) releases in South Korea.
Fox has found UHD/HDR and early release content represent the most attractive targets for piracy because of the quality and exclusivity of this content. Fox began early releases in South Korea in 2014 and has released 23 (now 24) films and counting. Its strategy for reducing piracy is three-pronged:
Watermarking of streaming home entertainment is becoming more and more prevalent-especially with the advent of UHD and early releases because traditionally, EVERY home entertainment release is pirated almost immediately. But by embedding a robust mark that can withstand attacks for the entire lifespan of the content , Fox has found that 43% of its releases in South Korea have not resulted in early release piracy. That is a huge win.
Moreover, criminal penalties have been brought to three individuals linked to the upload of Fox’s early release pirated content, and additional investigations are pending.
Like Fox, other providers of early release content and other premium video services can leverage watermarking to identify from which operator the content is leaked and close in on the specific device and/or subscriber account in order to shut down the leak, identify hacked or overly vulnerable devices and even pursue legal action against individual pirates and pirate release groups. This even applies to live streaming content, which requires very fast readout so unauthorized rebroadcasts can be shut down in real time.
We encourage you to watch the webinar in its entirety for more detailed information about how watermarking works and how it supports Fox’s success – and keep reading to find the answers to the questions participants asked during the live broadcast!
Q: What are the pros and cons of server-side and client-side watermarking?
A: The primary challenge for watermarking is ensuring that all playback devices are covered by some form of robust and effective watermarking. This is somewhat easier to achieve with a server-side implementation since the watermark is embedded before the content is delivered to the playback device, independent of the technology on that device. While server-side implementation is subject to malicious attack just like all other anti-piracy technologies, there is one less attack vector because, unlike client-side implementation, attackers cannot access the watermark inserters.
On the other hand, integrating and scaling server-side watermarking implementations can be challenging. In order to scale, a large number of vendors (encoders, origin servers, and CDNs) need to integrate the technology. Server-side watermarking must also support all content formats, codecs, containers, and unicast delivery systems.
Conversely, client-side watermarking does not require any preprocessing of the content in the headend because the technology is integrated into the SoC. It embeds the mark in the uncompressed domain and therefore can support any codec, container and file format, etc. Additionally, broadcast or multicast content such as sports is more suited for client-side watermarking because a single stream is sent to all client devices and cannot easily be uniquely marked at the head-end.
The challenge is that a complete client-side watermarking implementation must be integrated into all active playback devices. It requires operators to exercise a high level of due-diligence.
It is worth noting that server-side and client-side watermarking is orthogonal and can be used together to support different use cases.
Q: What about chipsets that don’t support client-side watermarking? Is there a secure software approach?
A: Verimatrix has worked hard to pre-integrate our VideoMark technology into the silicon of all major SoC vendors, Smart TVs, and STB platforms. Because the technology already exists in the hardware, an implementation of watermarking can be activated immediately by a software license from Verimatrix.
However, if a chipset does not support client-side watermarking, there is an option to use the trusted execution environment (TEE) and work within the secure video path, but not completely inside the hardware. If your business model is delivering to devices to which you have no control, then server-side watermarking gives you an advantage there.
Q: Is there ever a need to implement both server and client-side watermarking?
A: Yes. In a single-distributor situation, the distributor may have some playback devices that are within its control (e.g., a Set-Top Box or STB), and some that are not (e.g., a smartphone). In order to ensure that the watermark is inserted in each playback, it might be necessary to use a client-side implementation for the STB and a server-side implementation for the smartphone. Another example is a multi-distributor situation, where some distributors have client-side implementations and others have server-side implementations (or a combination of the two). In all cases, for content providers like Fox, the most important consideration is that each of its distributors ensures that all of its playback devices are covered by some form of robust and effective watermarking.
Q: Is visual watermarking an option or do the studios only accept forensic watermarking, or a combination?
A: Visible watermarking can’t be used where it would interfere with a legitimate end-user’s enjoyment of a movie or TV show, and there is also a security benefit insofar as invisibility makes a watermark harder to scrub from the video. But visible watermarks can certainly be used in a wide variety of B2B and other situations where a pristine copy is not required (e.g., vendor workflows, some review screeners, etc.). Moreover, when it comes to live sporting events, the immediate extraction and robustness of the mark during the event window is a key factor and could be considered more important than invisibility and security. A semi-visible (one or two frames) or fully visible mark provides for live events can be beneficial. Studios like Fox have been known to use a combination of visible and invisible watermarking, which serve different purposes. It depends on the situation.
Q: How do you find watermarked content in the field?
A: Studios have partners that constantly monitor the web for unauthorized leaks of video as a routine matter. When a file is found, it is sent back to the content provider. Studios first look for the distributor mark which identifies the distributor from which the file came. The file is sent to the distributor and they examine the file for evidence of the session-based watermark which will identify the specific device involved in the leak.
Q: Can you protect against pirates using VPNs to access and rip content? What about users creating new accounts for each infringement?
A: Pirates using VPNs to access early-window and other premium content offerings might succeed in hiding their true IP address, but in the case of transactional VOD/EST or any SVOD or other subscription-based business model; they still have to register an account and pay for the content before it is sent to them. That account registration and payment information will be captured in the watermark payload and, once recovered, will be available to law enforcement authorities, along with the date and time of acquisition and any playback from the distributor’s servers. Even if the authorities are not able to see the pirate’s true IP address in connection with the acquisition and initial playback of the file, the subsequent upload of the file to the Internet will also produce evidence as to the identity of the individual or group responsible for the upload.
In the case of users creating new accounts for each infringement, each such account creation will create its own evidence package, which can be linked together by the distributor and by law enforcement authorities.
In short, effective use of forensic watermarking and related techniques always provides content providers and law enforcement authorities with valuable anti-piracy evidence that would not otherwise exist and which raises pirates’ risk of getting caught.
Q: How good are the latest watermarking technologies at surviving attacks such as collusion, down-res (blurring), mirroring and cropping?
A: One of the requirements for a good watermarking technology is that it needs to resist a certain level of attack and we certainly test against collusion attacks. What’s interesting about collusion is you may actually end up revealing all of the sources that colluded. Watermarking is not like cryptography where if you are trying to break something, eventually you know if you have succeeded because you see clear video. With watermarking, you never know if you’ve succeeded in removing or damaging the mark or whether you’ve just left the fingerprint of everyone who participated.
Studios and other organizations test visibility/perceptibility and robustness of watermarking technologies. Transcoding, lowering resolution, cropping, camcording and other attacks are included in these tests. There are certainly limits to what watermarking can survive but at the same time, if the stolen copy is of such a poor quality that the watermarking could not be recovered, that content has likely very little value.
Q: Can anything be said about the type of content and associated time it takes to apply/detect the watermark?
A: The most efficient insertion technique for watermarking is the hardware-based, pre-integrated client-side insertion which supports UHD quality video. A pure software based client-side insertion is still efficient, but due to bandwidth efficiency may only be suitable for HD video. In that case, operators may think about utilizing a pre-integrated server-side solution.
Watermark extraction or detection is highly dependent on the quality of the pirated content. It takes a matter of minutes to extract the watermark from standard definition or high definition content without distortions. If there are several distortions on the file, the mark can still be extracted, but it takes a bit longer.
Q: Is there a Verimatrix API/SDK integration option for server-side watermarking?
A: There are several options related to the API/SDK integration which we are implementing. We currently support watermark insertion for ABR-particularly HLS and DASH formats.
You may have seen a recent press release announcing these features as well as our Standalone VideoMark forensic watermarking support for third-party conditional access and DRM systems. We encourage you to contact us for more information on how you can start implementing a watermarking solution today.
Still have questions about watermarking? Ask us in the comments.