How Netflix protects its content — Part 2
What is Forensic Watermarking and Why it is needed
In Part 1, we looked at the ‘multi-DRM’ technology that online video service providers such as Netflix use to protect their content from illegal use.
Because DRM-enabled content is distributed in an encrypted format, anyone who does not have a usage rights (DRM license) cannot play the video files.
But DRM technology alone can not completely protect content from illegal distribution.
The authorized user’s device needs to convert the encrypted DRM content to its original form through the decryption process for playback. In this playback process, there are technical limitations that make someone can leak the content.
Therefore, ‘forensic watermarking’ technology is needed to keep track of who leaked the content and prevent further leakage.
This post is an English translation of my Korean blog. If you prefer Korean, please check the original post here.
What is Forensic Watermarking?
‘Forensic Watermarking’ is a field of a technology called ‘digital watermarking’. ‘Digital watermarking’ refers to the technology of inserting and managing confidential information such as copyright information in various digital data such as photographs and videos. (Ref. 1)
Both terms ‘watermark’ and ‘watermarking’ are used interchangeably in most cases. Mostly, ‘watermark’ refers to the data that is inserted into the content, and ‘watermarking’ often means the technology or solution.
While conventional digital watermarking is for claiming copyright by embedding copyright holder information in the content, ‘forensic watermarking’ inserts information of content users to track illegal distribution.
When a content copyright holder or content service provider discovers illegally distributed content, it can detect the watermark, track the user, stop the user from using the service, or take legal action to prevent further illegal spillage.
Why Forensic Watermarking is needed?
If I can stop illegal use of content with DRM, why do I need forensic watermarking?
You may have questions about the need for forensic watermarking. However, even if DRM protects the content, it is difficult to prevent illegal leakage completely.
Though the content is transmitted securely to the client (playback device) in an encrypted state, eventually the client should decrypt the content for playback.
To protect the decrypted content in this final stage, you can apply various technologies such as ‘Hardware-level DRM’, ‘Trusted Execution Environment (TEE)’, and ‘High-bandwidth Digital Content Protection (HDCP)’.
But still there are cases where these technologies alone can not prevent content leakage.
Leakage case 1 — Screen recording using camcorder
One of the most traditional content leakage cases is the so-called ‘cam version’ footage.
This is to record the video played on the screen with a separate camcorder or smartphone camera. DRM technology can not prevent this type of content leakage.
This is called Digital-to-Analog (D2A) or Analog-to-Digital (A2D) conversion, one of the attack methods for watermarking.
With the increasingly smaller and higher-performing devices, it is now possible to record video with almost the same quality as the original at 4K/UHD resolutions beyond full HD.
This is a more serious problem for OTT services like Netflix than movie theaters where the camcording can be prohibited manually.
Leakage case 2 — Video recording through PC screen recorder
As we learned in Part 1, support for DRM content playback through PC web browsers and HTML5 players is one of the most important target platforms for OTT services.
However, since some web browsers use software based DRM, you can easily save the DRM video as normal MOV or MP4 format video file by using various screen recording tools.
Specifically, PlayReady DRM applies to IE11 (Win 8.1 or later) and Edge (Win 10) browsers of Windows OS. Here, it is impossible to capture the screen shot or video due to the OS level screen capture prevention function.
In Mac OS, FairPlay Streaming DRM in Safari browser also prevents screen capture using a screen recorder.
Chrome and Firefox browsers, however, do not prevent image capture because of software-level Widevine DRM on most supported operating systems, including Windows, Mac OS, and Linux. (Ref. #2)
The topic of hardware/software level DRM and screen recorder issue will be discussed in a separate post.
The requirement from Hollywood studios
In 2007, major Hollywood studios, including Disney, Sony, Warner Bros., Universal, and Paramount, established a nonprofit organization called Motion Picture Laboratories (MovieLabs) to study technology related to the distribution and security of movie content.
To prevent the copyright infringement due to the illegal leakage, ‘MovieLabs Specifications for Enhanced Content Protection’ require the application of hardware-based DRM and forensic watermarking technology for high-value content such as UHD movies. (Ref. #3)
Basics of Forensic Watermarking Technology
Two of the most important requirements for forensic watermarking are ‘imperceptibility’ and ‘robustness’.
- Imperceptibility: The difference between the original image and the watermarked image must not be visually recognizable.
- Robustness: Watermark information should survive various attacks such as re-encoding, cropping, and filtering.
‘Visible’ watermarking is not suitable for a forensic watermarking solution because it may degrade the quality of the original content and can be removed easily through editing.
To satisfy non-visibility, most forensic watermarking techniques insert watermark information into invisible areas of the video frame images.
As for the watermarking solution that satisfies the invisibility, the watermark image should look the same as the original image, as shown in the above reference image.
A high-intensity watermark for ‘robustness’ from attacks would cause the watermark to become visible. Applying the right level of watermark to meet both requirements is a key technology of forensic watermarking.
In addition, it requires the ‘uniqueness’ of the watermark to track the end user. To do this, ‘session based’ watermarking technology inserts unique user information into the watermark for each playback session.
Session-based watermarking technology can be divided into client-side and server-side technologies.
- Client-side watermarking: Watermark is inserted by the user device (client) playing the content. It requires the minimum modification of the back-end server system, but there is a disadvantage that it can support specific clients only.
- Server-side watermarking: The server system of the content service embeds the watermark. The advantage is that there are no constraints on the client player.
Application Areas of Forensic Watermarking
Forensic watermarking technology has been developed and used in various fields to track illicit leakage of video contents.
- Screener (Pre-release): Movie content that is distributed in advance in the form of files or discs to internal/external stakeholders of the studio for review before release. In case of illegal leakage, it causes great damage to the copyright holder, so it inserts watermark information about the distribution channel and target to prevent leakage.
- Digital Cinema: Watermarking can be applied to movie content that is displayed in a theater in the form of digital film. It is necessary to insert information such as the theater and screening time into a watermark, and link it with a digital cinema system.
- OTT VOD service: When forensic watermarking is applied to the premium contents of online movie service (like Netflix), it is possible to insert the end user information of the service in real time to track the illegal leak.
- Live stream service: For live events such as World Cup, forensic watermarking can be used to detect and immediately block illegal stream retransmissions.
We will discuss the classification of watermarks by inserting subjects and details of various implementation methods in a separate post.
Multi-DRM and Forensic Watermarking
While DRM is a technique to prevent unauthorized use of content, watermarking is a technology that tracks illegal content leaks by authorized users.
‘DRM’ and ‘watermarking’ complement each other and must be applied together to premium content that requires a high level of security, such as the latest Hollywood movies.
PallyCon Forensic Watermarking and Multi-DRM
In order to quickly and easily apply forensic watermarking and multi-DRM, which are essential for premium content security, it is recommended that you use a professional company that can provide both solutions as a unified service.
‘PallyCon Forensic Watermarking’ is a cloud-based forensic watermarking SaaS service provided by INKA Entworks, a content and application security technology company. It is integrated with ‘PallyCon Multi DRM’ service in a unified content workflow. (Ref. #4)
‘PallyCon Forensic Watermarking’ guarantees robustness to withstand various attacks such as re-encoding, cropping, filtering and camcording. It passed through the watermark robustness tests of several Hollywood studios to be reliably applied to the latest Hollywood movie content.
If you sign up the PallyCon service now, you can experience forensic watermarking and multi-DRM services for one month through ‘Free Trial’.
References
- #1 Digital watermarking — Wikipedia : https://en.wikipedia.org/wiki/Digital_watermarking
- #2 You Can’t Protect Your Videos From Being Stolen — Foliovision : https://foliovision.com/2017/01/stealing-videos
- #3 MovieLabs Specification for Enhanced Content Protection : https://movielabs.com/solutions-specifications/enhanced-content-protection-ecp/
- #4 PallyCon Forensic Watermarking : https://pallycon.com/forensic-watermarking/
