Posted by: Peter Burrows on May 31, 2007
It’s called Vobile Inc. In reporting a story on the surprising progress in video fingerprinting technology, the company was mentioned time and again by sources—even by rivals—as having a technology that stands at or near the top among the twelve systems that have been tested by the Motion Picture Association of America over the past six months. That testing is now pretty much completed, and the MPAA has planned a conference to discuss the results with the leading Hollywood studios on June 21, according to one source. Vobile is already in talks with a number of studios, and its technology is being tested by MySpace, Chinese video site Tudou.com and others. The company has already announced a deal with Pohomo, another large Chinese video site.
More after the break…
The reason for the enthusiasm is that Vobile's technology seems to be able to efficiently spot a content owners's creations out on the Net--even clips that have been compressed, cropped, or knowingly altered to avoid detection. (Like most fingerprinting system, Vobile's technology is a two-part system. The content owner needs to install a system to store a database of fingerprints of its shows or films, and video sites need to install systems, that would search for matches when users upload or try to view a piece of content).
While he's short on details, Vobile CEO Yangbin Wang says the company has come up with a way to extract what it calls Video DNA, in a way that is more foolproof than most of the methods that have been tried over the past decade or so. For example, many suppliers have focused on spotting pirated videos by IDing the soundtrack that goes along with them--not much use if the movie has been dubbed into another language. Others focus on fingerprinting only key scenes--leaving the rest of the show or movie vulnerable. But Vobile claims it can find even a few seconds--any few seconds--of a video. In one demo he showed to studio execs, Vobile’s software was able to find a few seconds of a Star Trek movie that appeared inside a mash-up that had been reformatted to be viewed on an iPod. “We like the term “Video DNA” better than fingerprinting, because DNA is more accurate than fingerprints [at identifying criminals],” says Wang.
Just as important, sources say that Vobile's technology is proving to be very good at avoiding false positives--that is, at not IDing a clip as something it is not. This is critical, because no studio or video site wants to be in the position of incorrectly accusing a consumer of piracy. That not only spells lawsuit, but a surefire PR fiasco given the raw nerves that surround the whole issue of copyright protection on the Net. “False positives would be an instant death knell for DavidLiu.com,” says Jeffries & Co. analyst David Liu. “Nobody wants to be the guy that makes a lot of false accusations.”
But there’s another requirement for any video fingerprinting system to actually be used: that it be able to analyze lots of movies, with relatively few computers. That's to hold down the cost. Multiple sources say Vobile stands out in terms of scalability. “We’re not talking about a room full of equipment; we’re talking about one rack [of gear]," says one potential customer. Wang claims one such rack would be enough to scour as many movies as there are in Netflix' entire inventory.
CEO Wang and CTO Jian Lu, a former Apple engineer, created the company two years ago, with a goal of creating a kind of eBay for videos. But they realized they’d need a way to ensure they weren’t facilitating trade of content that copyright holders didn’t want traded--and ended up focusing all of their efforts over the next two years on developing that technology.
The company has plenty of challenges, starting with finances. The company has been hurt by a lack of interest in the video fingerprinting space from the venture capital community. Besides the sheer technology risk that the product won’t work well enough, there’s also the question of who will pay (studios think the video sites have the responsibility to know whether they’re hosting infringing content, while the sites think the studios should foot the bill for protecting their products). And more fundamentally, many VCs doubt video sites will rush to deploy a technology whose purpose is to filter out content consumers want to see. “We have enough money to make it for a while, but we need to work out our business model before we’ll have a credible story for the VCs,” says Vobile chairman Vernon Altman, a long-time Bain consultant. “We’ve got technology that’s off the charts, but we have to prove there’s a business model to go around it.”
Vobile must also overcome the fact that it's an unknown quantity. Many studios and video sites would prefer to work with established players such as Philips Electronics, Thomson or Audible Magic, which all know how to deploy technologies reliably and in ways that integrate with the way their businesses work. One possibility is that Vobile might cut a deal to have its technology integrated into their offerings. Audible Magic CEO Vance Ikezoye says he's interested. While his company plans to roll out its own video fingerprinting technology in coming weeks, he says that “Our technology doesn’t have to be the total solution.”
Still, the folks at Vobile are flush with confidence, certain that they are a diamond in the rough that's yet to be discovered. “We’re the new kids on the block, but we're getting to the point that people feel comfortable enough to work with us," says Wang. He claims he has turned away feelers from rivals interested in buying Vobile. We'll see if his bold confidence turns out to be justified. But at least at this point, Vobile has won a spot at the table at a time when many in Hollywood feel video fingerprinting has finally emerged as a viable technology.