Images of the Future

Leonardo Chiariglione is the brains behind MPEG — and he’s about to make metadata the next leap forward.

Pick apart Coca-Cola’s business model, and one fact leaps out. The Atlanta-based company spends only a tiny fraction of its revenue producing famous brown syrup. The real muscle of the company is in labeling, packaging, and branding.


Should multimedia companies be following Coke’s example?

At first, the question sounds like heresy. We’ve grown up in a world where it takes big money and incredible technical expertise to create the movies, television shows, and music that millions of people want to enjoy. We expect that Dan Rather and Julia Roberts will command multimillion-dollar salaries for their work. We hardly ever think about the people who write the capsule summaries for TV Guide or the back-cover copy for a DVD box. If we do, we assume that these are dull clerical tasks that can be carried out as low-cost afterthoughts.

But maybe that analysis is becoming dangerously out-of-date. In a world where it’s easier than ever to create content — and maddeningly hard to know exactly what’s out there — a surprising amount of power could shift to those who do the best job of indexing and cataloging.

Go one step further, and it’s easy to imagine that the next multimedia company with a $10 billion valuation won’t be a movie studio or a record label, but instead will be an upstart that organizes what is already out there and steers consumers to what they really want. The 20th century belonged to movie studios like Fox. The 21st century will belong to companies that can bring order to an information explosion — companies that can become the entertainment industry’s equivalent of Reuters or Dun & Bradstreet.

That’s the provocative vision of Leonardo Chiariglione, an Italian scientist who is a legend in engineering circles — and not nearly well-enough known or appreciated in the business world. He is a top researcher at Telecom Italia Lab, but his real power comes from his behind-the-scenes leadership of the Motion Picture Experts Group, or MPEG. Since 1988, this remarkable global brain trust has been the place where companies as diverse as IBM, Philips, and Sony help set the standards that define our digital future.

Even in his student days, Chiariglione broke away from the pack in flamboyant style. When he was exempted from Italian military service, he decided to use this suddenly available slice of time to go abroad for advanced study in electrical engineering. The United States would have been the obvious destination. But that was too easy, he decided. Instead, he enrolled in the University of Tokyo to see how he would fare with all of the instruction in Japanese. Three and a half years later, he won his PhD.


In the past 10 years, Chiariglione and friends have unleashed MPEG-1, MPEG-2, and MPEG-4 on the world. Those three landmark standards define video compression, and they have made it possible for untold millions of people to send and play short-movie files over the Internet. Chiariglione’s influence also extends to the architecture of the MP3’s digital-music format and the design of microchips that handle digital-media signals.

Now Chiariglione’s most ambitious project is about to go public. It’s called MPEG-7, and if it works as planned, it will allow people to index and search digital video and audio files as never before. Looking for that song that goes “ba-ba-ba-boom”? Key in the first few notes, and MPEG-7 technology will find everything that matches up. Want to find helpful — or menacing — uses of your trademarks in video footage worldwide? Enter descriptions of your logos and emblems, and this new technology will do the rest.

The full impact of MPEG-7 may take years to play out. But the basic principle is clear. To make sense of all of the multimedia files out there, we need more than just simple text-based file names. (Go to any current search engine, type in “Birthday.jpg,” and you’ll see the problem.) We need much more sophisticated ways of looking for specific images and patterns. And we need one standard way of doing that — so that we aren’t stuck in a world where even the best tools don’t let us search more than 1% or 2% of the Web.

Creating all of this metadata — information about content — won’t be cheap. In fact, “it might cost more than the content itself,” Chiariglione mischievously says. But that doesn’t bother him one bit. In a world where we are all bombarded with too many media choices, consumers will take content for granted and cherish anything that can help them stay organized.

What do the established heavyweights of entertainment, software, and consumer electronics think about this? Most of them are cheering on the initiative. Plans are under way to rally MPEG-7’s biggest potential users for a multiday public demonstration in Washington, DC. Among the companies due to make presentations are Canon, Mitsubishi, Ricoh, Sharp, and Sony for Japan; Philips, Telematica, and Thomson for Europe; and IBM for the United States.

Individually, each of these companies might be wary of getting too far ahead of what consumers want. After all, right now, there are dozens of half-built attempts to create better metadata for multimedia — none of which has yet taken the world by storm. But knowing that a well-built system can work everywhere, industry veterans say that a unifying standard will make it much easier to march forward.


“With good metadata, you can create new applications,” says Peter van Beek, a project leader for Sharp’s U.S. research lab in Camas, Washington. “And it’s going to make the existing applications more valuable.”

One of the ideas that Sharp Labs is playing with involves developing user profiles that are fully portable. Watch movies at home, and software can keep track of your favorite actors and genres. Board a plane or check into a hotel, and that same software can point you toward a menu of choices that you might like.

Another Sharp Labs project involves automating the otherwise slow and laborious process of identifying particular segments of video content. Lately, van Beek and his researchers have been cooking up ways to identify home runs or soccer goals automatically in sports videos. There are enough common elements of a home run, he says — the crack of the bat, the camera zooming into the bleachers, the slugger circling the bases — that his team’s accuracy rates approach 99%. That kind of ability makes it easy to label or repackage game footage so that fans can zoom ahead to the highlights if they wish.

In fact, almost everyone associated with the MPEG-7 project has already advanced some favorite ideas about how to harness this powerful new technology. Talal Shamoon, head of business development for InterTrust Technologies Corp., a digital-rights-management company in Santa Clara, California, sees a day when news organizations can search through old footage for relevant clips as never before. “Say you wanted something from the archives on Afghanistan,” he says. “Now you would be able to get information quickly and efficiently, without having to scroll through endless old tapes.”

Add up all of these initiatives, Chiariglione says, and “there will be new companies, and maybe even new industries, formed around metadata.” The hardest work will be performed in the first few years, he says, which will be spent translating today’s standards into fully functioning metadata services. But he believes that once consumers begin interacting with these new systems, the momentum to do more will become unstoppable. As Chiariglione wryly puts it, “Producers who care to have their programs watched will have to describe their content as fully and as engagingly as possible.”

George Anders ( is Fast Company’s Silicon Valley bureau chief. Find a catalog of his columns here. Learn more about Leonardo Chiariglione and Telecom Italia Lab on the Web (