This blog first appeared on Videonet.

Last week I spoke to Petr Peterka, CTO of Verimatrix, about the Hollywood studio’s latest attempt to avoid following in the music industry’s footsteps and the new Verimatrix’ watermarking product.

The general context we addressed appears bleak. The musical CD is no longer relevant for mass consumption and the DVD looks as if it is following in that direction. The MP3 format has become ubiquitous for music files. Will formats like MP4 or h.264 files doing the same for video? Just as P2P seems at last to be slowing, is streaming now taking up the slack?

Book publishers are also alarmed at what their future might hold for them. Tim Bradshaw, the FT’s digital media correspondent noted on May 17th that according to a recent survey, one in eight female tablet or e-reader owners over the age of 35 admits to downloading “unauthorised” copies of e-books. It’s not so much the 1 in 8 figure as the image of women over 35, which catches the imagination. Illicit content consumption is not just from acne-ridden boys. So if we’re all potential pirates, could digital watermarking be the deterrent that might finally work where previous threats have failed to stem the tide of piracy?

With session-based, forensic digital watermarking small changes are made to the video file. These cannot be viewed or heard by the viewer, but can be extracted out again from the file with the tools provided by the watermarking solution vendor. If someone redistributes his or her copy of a film, a unique “fingerprint” remains embedded in the video file. So after investigation, a finger can be pointed at the person who first redistributed their copy.

So that’s the stick, but the movie industry is trying to put together a carrot by creating new early release windows for movies. As users are prepared to pay for convenience, it seems only natural to expect them to pay a premium to get access to a movie more conveniently. Organising a movie party at home to show friends a film that might still be in movie theatres would be worth spending a bit more than on a normal $5 VoD (current US VoD rights start 90 days after theatre debut, the new “Home Premiere” would be available 30-60 days after debut instead).

The difficult question is exactly how much more is that worth? Studios have been talking of up to US$ 25 more for a rental period. TDG recently analysed the potential demand and was not convinced that this kind of proposition can fly. According to the TDG analysis, users would only be prepared to spend an extra $5, and only if the movie were available within a week of theatre release. Further, the TDG study notes that there is still some leeway in playing with both the price and time frame so there may be room for a new offering even if studios might have to lower their expectations a bit.

However, more people are betting on ad-funded models with free or cheaper content made available. Many small transactions are more in fashion than fewer big ones. As digital watermarking requires session-based processing, it will be too expensive for micro-transactions. Time will tell if offering something better for a higher price will be going against the current in the entertainment industry or not. Sometimes bone-headed stubbornness has paid, albeit rarely.

Verimatrix is involved in enabling this early release VoD window but Petr rightly used the term “experiment” when describing this initiative from the studios.

He told me that they had been waiting for three specific requirements to be fulfilled before embarking on this project:

• Firstly secure encryption was required. This has been achieved and consolidated in the last decade
• Disabling analogue output was the second requirement so as to make it harder to record. This was only achieved recently in the US when, last year, the Selectable Output Control issue was finally resolved last year.
• Finally, studios required session based forensic watermarking which is what the new Verimatrix product is all about.

Verimatrix released a product called VideoMark 5 years ago, which was able to insert a digital watermark payload into the video outputs of a set-top box in real time. This addressed a range of re-broadcast and re-distribution threats, but required device-by-device software integration. The novelty of the newly released StreamMark product is that the watermarking can now happen in Head End or in the Content Distribution Network (CDN). Indeed the watermark can be inserted into a compressed and encrypted stream. By removing the requirement of decoding and decrypting, this new solution requires only a very small processing overhead. But as this solution is technically a stateful one, scalability will still imply some cost, even if it's a small one. StreamMark is designed to insert a unique watermark for each user and is therefore primarily useful for unicast not multicast, i.e. on-demand applications.

No information is required on the end device so this kind of architecture is well adapted to multiscreen deployments that at the moment are on everyone’s mind.

The basic challenges of any watermarking technology are that the watermark must be:

• invisible
• easy to extract
• difficult for pirates to see and remove.

As these challenges can be somewhat contradictory, a finely tuned combination of different algorithms is necessary, and achieving this is probably one way we will see competing solutions differentiate.

Verimatrix' solution requires three separate steps:

• Firstly, the video assets must be pre-processed, usually when the operator ingests new content. This stage identifies and stores the location of potential marks. For each of these locations, the information that would be inserted is also prepared and stored for future use.
• The second stage is to embed the watermarks at distribution a point where a unique session is set up. This can be in a central server or regional VOD server or in an edge CDN server. The original file has extra info added to it that embeds a unique transaction ID.
• The final stage will only happen when pirated content is discovered. For now extracting the watermark is only done in Verimatrix’ labs.

The technology was reviewed by the major studios during the soft-launch period and a third party audit was carried out by Jian Zhao, a distinguished researcher in watermarking theory and practice.

The Verimatrix pricing structure is based on unique payloads i.e the number of transactions and is volume dependent. On June 7^th, StreamMark, which has been in soft launch for a year, was be released commercially.