Welcome to the Metaverse
By Mark Frauenfelder, Wed Jun 12 00:00:00 GMT 2002

Suddenly, my image is replaced by that of a slightly manga-esque looking young Japanese woman with a sensible haircut and an expensive business suit...

I just sat down in front of the computer in Eyematic's laboratory in Los Angeles, California and I'm not quite sure what to expect. There is a little webcam mounted above the PC, pointed at my head. My live video image is being displayed in the PC's window. In the lab with me are Orang Dialameh, Eyematic's CEO, Hartmut Neven the chief technology officer, and chief scientist Jaron Lanier, the famous virtual reality pioneer.

Dialameh taps a couple of buttons on the keyboard and suddenly a pattern of dots, connected by lines, is floating over the image of my head. It looks as if the kabalistic tree of life has been tattooed on my face. Actually, these dots are positioned over certain facial features, such as my nose, eyes, eyebrows, and the middle and corners of my lips. If I turn my head or move it from side to side, the dots followed me like a swarm of flies.

Dialameh clicks a couple of keys again. Suddenly, my image is replaced by that of a slightly manga-esque looking young Japanese woman with a sensible haircut and an expensive business suit. I start to laugh. And like some kind of magic mirror, the Japanese woman laughs too. I raise an eyebrow. So does she. I open my mouth. I wink. I nod. I roll my eyes. She copies my every move. She is a kind of virtual marionette, but instead of being controlled by strings, she is responding to the changes in my face. She's an avatar, not one in the sense of an incarnated Hindu deity, but in the way science fiction writer Neal Stephenson described them, in his now classic cyberpunk novel Snow Crash.


In Snow Crash, Stephenson envisioned a high-fidelity virtual world dubbed the Metaverse, which was like a fully immersive 3D version of the World Wide Web. People could buy off-the-shelf animated characters, or avatars, that they could use to interact with other people visiting the Metaverse. As Stephenson explains: "Your avatar can look any way you want it to, up to the limitations of your equipment. If you're ugly, you can make your avatar beautiful. If you've just gotten out of bed, your avatar can still be wearing beautiful clothes and professionally applied makeup."

In the novel, the most popular avatar for women was called the "Brandy," and the men favored the "Clint" model. Here's how Stephenson described the avatars: "Brandy has a limited repertoire of facial expressions: cute and pouty; cute and sultry; perky and interested; smiling and receptive; cute and spacy. Her eyelashes are half an inch long, and the software is so cheap that they are rendered as solid ebony chips. When a Brandy flutters her eyelashes, you can almost feel the breeze." And Clint is "the male counterpart of Brandy. He is craggy and handsome and has an extremely limited range of facial expressions."

Stephenson pretty much nailed the future. Here at Eyematic, the choice in avatars is even better than the ones available in the Metasphere. To demonstrate Dialameh grabs the mouse and moves it to a pull-down menu, selecting a new character. This time, I'm a caped superhero with a jaw the size of a toaster. Then, I become a fat cartoon dragon. This is fun! It becomes even more fun when an Eyematic employee in another part of the building has a videoconference with me, using avatars instead of our real video images. He is some kind of purple monster, and I am Humphrey Bogart, or at least someone who looks just like him. About all I can tell him was, "This is really neat!"

Mobile Masks

The even neater thing about Eyematic's avatar technology is the way it can be used in low bandwidth situations, such as over mobile phone connections. That's because Eyematic's software isn't actually sending fully rendered 3-D animation over the wires - that would bring most wireless networks to their knees. Instead, the real-time facial tracking software analyzes the movements of the user's face, extracts the relevant information, then sends the instructions to move the corresponding features on the avatar, which has been previously downloaded onto the device. It's like pulling the marionette strings over a distance instead of sending the marionette itself. The company claims that its "lightweight" data format requires less than 1/100th the bandwidth of standard video.

Eyematic hopes to get wireless operators interested in using this technology, and is offering a package of applications they call "Eyematic Synthetic Video" that provide an end-to-end suite of animation services for low bandwidth applications. The technology is a combination of authoring applications, services on the backend that take care of content delivery, and the player software that actually runs on the phones themselves.

So far, Eyematic has managed to grab the attention of several high-profile telecom-related investment firms, who have ponied up $16 million in funding in January. Investors include Texas Instruments, Deutsche Telecom's T-Telematik Venture Holding GmbH, and Japanese investment firm INTEC, which is developing rich media delivery for Japan's mobile Internet content market.

In March, Eyematic and Texas Instruments announced that they were working together to add Eyematic's software capabilities to 2.5 and 3G cellular phones and mobile handhelds that use TI's high-performance, ultra-low power OMAP application processors and wireless modems. Steve Hartford, Eyematic's VP of marketing, says the OMAP processors can process multimedia content at close to desktop quality without sucking phone batteries dry in a matter of minutes.

Eyematic has ported its multimedia engine over to 22 devices. In some cases, the software is embedded in the phones' firmware, and in other cases, it's downloaded on demand into the phone's memory. Eyematic's technology is already running in commercial applications in Japan and should be up and running in a major wireless carrier by the end of June.

The current applications include a horoscope avatar that reads your fortune every day, a 3D multimedia messaging application that lets you select a character to announce your messages, and a 3D adventure game. They've also been running tests on a perky avatar assistant that works in conjunction with your scheduling program to appear on your phone's display and cheerfully announces upcoming appointments.

In the coming years, Eyematic will work closely with device manufacturers, because they're interested in the potential for avatar videoconferencing using video cameras mounted in telephones. For now, however, Eyematics avatars can be triggered by speech alone, or even by plain old text that runs through a text-to-speech generator.

Jeepers, I'm a Veeper!

About 400 miles north of Eyematic's Los Angeles headquarters, I find myself in the offices of another company that's creating avatars for a wide range of purposes. Pulse Entertainment, a privately held company with recent investments from Softbank, AOL/TimeWarner, Autodesk and El Dorado Ventures, is set in a funky brick building on the upscale side of Market Street in San Francisco. I'm here on a sunny Friday morning, posing against a whiteboard while Pulse's laid back and genial CEO, Fred Angelopoulos, snaps my picture with a digital camera.

"Ok, you're going to make your own Veeper," he announces, steering me to a laptop. (Veeper, as I find out, is what the software engineers were using to designate the VPs, or "virtual persons" they were developing.) After transferring my photograph and loading up the program, I use the mouse to designate some key "data points", such as my head outline, eyes, and mouth. I follow a few more step-by-step instructions, and in the space of about one minute, I'm staring at a reproduction of myself. The image bobs its head slightly, and blinks every once in a while. It feels alive. I can press different buttons and make it appear happy or sad. Next, I say a few words into a microphone and hit the return key. The avatar repeats them, moving its lips in sync with the words. Freaky!

Then, just for fun, Angelopoulos has me type some words into a text box and asks me to select one of the different text-to-speech generators available from a pull-down menu, such as a male Scotsman, a valley girl, and an enthusiastic sports announcer. Suddenly, a brogue is coming out of my... er... my avatar's lips.

Like Eyematics, Pulse's technology is optimized for low bandwidth conditions. Even though it uses very little bandwidth, Angelopoulos says "people think it's streaming video." In fact, that's one of the things Pulse has trouble with when they demo the technology. People think they've seen this before, but they are mistaken. Very little data is going over the wires when a Veeper is in action.

Also like Eyematics, Pulse is working with Texas Instruments to develop Veepers for multimedia messaging, enterprise, entertainment and instant messaging wireless applications on 2.5G and 3G mobile devices that use TI's OMAP application processors.

Pulse showed me a couple of demos on mobile devices. One was a color 3D avatar running on a Pocket PC, which looked as good as the desktop demos I'd seen earlier. But even more impressive was the simple 2-bit graphics playing on the monochrome screen of an ordinary mobile phone. The character had a sense of life to it that transcended the clunky pixels on the screen. And it will only get better when color smartphones hit the mainstream.

After seeing the demos at Eyematic and Pulse, I left impressed, and convinced that 3D characters on wireless devices to enhance e-mail, voicemail, games, and other services will definitely be part of the future of the mobile Internet.

Want to play with my Veeper? You can visit my veeper page at http://demo.pulse3d.com/wired_markf/. (Just don't make me say anything that could get me in trouble.)

Mark Frauenfelder is a writer and illustrator from Los Angeles.