michael naimark

Leonardo Electronic Almanac, Vol. 9:5,
MIT Press
May 2001


VR Today

By Michael Naimark

I never liked the phrase "virtual reality".  For one thing, it immediately spawned the phrase "really real".  (Jaron: "Look! This new graphics engine makes these virtual worlds look really real."  Me: "What? It's a friggin' cartoon.") To Jaron Lanier's credit it was the promotion of VR's "realness" that provoked community consciousness to imagine what immersive virtual environments could be. The down side, of course, was that such promotion - some would say hype - led to grossly mismatched expectations between the dream of VR and the actuality of what was technically possible. For those of us working in the trenches, it was a mixed blessing.

Ten years ago, VR rocked.  It was everywhere.  Jaron, who coined the phrase "virtual reality," was on the front page of the Wall Street Journal.  Marvin Minsky, co-founder of MIT's Artificial Intelligence Lab, spoke at Ars Electronica on the topic.  Timothy Leary came back (actually was sort of adopted by the VR community, and bless him, added a brilliant mix of insight and outrageousness).  Corporations, particularly Japanese ones, announced major VR initiatives.  VR goggles and gloves became a fashion statement. " Cyber-sex" (a phrase even more oxymoronic than virtual reality ) was seen in supermarket tabloids and discussed at art openings.

Then what happened?  It seems as if VR bit the dust as fast as it rose to fame. Corporate VR labs, particularly at Sony and Fujitsu, shifted focus.  Many of the first VR companies, including Jaron's VPL, folded.  The great orators moved on.  The spotlight had shifted.

This was around 1992, and - no surprise here - the spotlight had shifted to the Internet and the Web.  This is significant, and I'd like to offer some observations and speculations.

First, VR isn't dead, and with the spotlight off it, more serious and focused work has been able to proceed unhampered.  Head-mounted displays - the goggles - have gotten higher resolution and lighter weight, with promising work on "virtual retinal displays" led by the HIT Lab at the University of Washington.  Fakespace Labs, a first-wave VR company, is doing well building innovative immersive viewers and screens.  Tracking technology - the gloves - has miniaturized and diversified, with tracking devices integrated into every human appendage, prosthetic, and tool imaginable (a charge often led by artists).  There's some excellent work in this area continuing at the University of North Carolina.  "Image-based modeling" and "light fields," recent sub-disciplines of computer vision, have transformed the range of 3-D computer models from built-from-scratch to camera-based, led by landmark work at Stanford, UC Berkeley, and MIT.  And in a most-forward thinking way, the artists and scientists at the Electronic Visualization Lab of the University of Chicago invented the "CAVE," an immersive projection space for group viewing capable of live-networked applications.

The much-quoted 1965 decree of Ivan Sutherland articulated the original dream of VR: "The screen is a window through which one sees a virtual world. The challenge is to make that world look real, act real, sound real, feel real."  Ten years ago, this was the world of eyephones and datagloves; of SGI Reality Engines and high-level modeling languages; and of flight simulators and location-based theme park attractions.  Today it's postage-stamp-sized silent images of Jenni-cam and the Western Wall, updating (on a good day) at one frame per minute.  But wait, you may say, it's even more real because it's live and it's interactive.  Clearly, realness means different things to different people.

Let's look at the numbers, using moving images as a frame of reference.  I believe there are five distinct levels, each with its own distinct industry.  The hottest one right now is the lowest-resolution one, the "streaming video" level, since everyone wants video over the thin pipe of the current Internet.  Streaming video often has to pass through a 56 kilabit per second modem, and is rarely more than 1 megabit per second (mb/sec).  But streaming video never looks as good as the movies we rent on VHS.  VHS quality, and home video in general, is the next level. VHS resolution is approximately what the MPEG-1 standard is, with a bandwidth or bitstream of 1.5 mb/sec (originally for CD video).  It's noteworthy that moving from streaming video to home video crosses an industry line from computing and networks (Apple and Akamai) to consumer video (Sony and Blockbuster).  These people drink in different bars.

The next level is broadcast-quality video.  MPEG-2 was made for this, with a bitstream range from 4 to 9 mb/sec.  Everyone agrees that broadcast video looks much better than VHS, and high-end broadcast equipment typically costs ten times more than consumer video equipment.  Uncompressed broadcast video often travels around production studios at 45 mb/sec.  Then there's cinema.  The silver screen of the local movie theater appears much bigger than the television in the home, and requires that much more bandwidth.  We just crossed another industry line from video to filmmaking, from CNN to Panavision.  Typically, movies are shot with 35mm motion picture film, but the quest to replace film with high-definition digital video has its roots almost 20 years ago.  Today, various digital HDTV bitstreams range from 20 mb/sec for highly compressed HDTV to as high as 1,000 mb/sec, 1gigabit/sec!

But we're not done.  "Special venue" cinema, the sort of immersive movies seen in theme parks and world's fairs, are typically 10 times the bandwidth of theatrical 35mm film.  This is yet another industry, with formats like Imax (wide 70mm film), Showscan (70mm film at 60 frames per second rather than 24), Stereo-70 (twin 70mm film for 3D), and CircleVision (nine 35mm screens in a panorama), each presenting different offerings of what's "really real."

So the range of current versions of moving images is from roughly 0.1 mb/sec to 10,000 mb/sec. It's both true and ironic that the Real.com folks and the Imax folks, or the QuickTime VR folks and the CircleVision folks, have very little to talk about.  But when one looks at the continuum, the drive at each end becomes clear. At one end is sensory verisimilitude. Some might say, "looking real." But others might say "dead," in that it's never live and at best barely interactive. At the other end is "live," telematic, participatory, and interactive, even if the cost is sacrificing a four-story high 3D screen for a postage stamp screen. Again, realness means different things to different people.

There's a noteworthy other phenomena, dealing with the politics of access, best summarized by French art theorist and UNESCO Webworld Director Philippe Queau: "Maximum hits per bits." Its extreme interpretation is that there's something unfair about concentrating a large number of bits for a small number of people. Immersive theaters and art installations are out, websites accessible to all are in. The reason is beyond theory:  53% of the world still hasn't made a phone call.

Incredibly, this polarized situation is temporary.  Add Moore's Law and optical fiber, easily capable of over a million mb/sec (!), and the conflict disappears. It's inevitable, with the only questions being, when and how? The lever is large because the status is embryonic.

Here are some examples of interim strategies doable today.  One strategy is simply to continue doing what we can with the narrow pipe of the Internet.  It's relatively cheap, easy, and uncharted (particularly when one ponders the World Wide part of the Web).  And the pipe is quickly getting wider via DSL, cable modems, and broadband.

Another strategy involves hybrid investigation.  For example, there are lots of stereoscopic and panoramic images on the web, even if they're small.  "Interactive films" have existed since 1967 and have managed to present at least the illusion of control. For example, the world's first interactive movie, at Expo '67, involved red/green pushbuttons for everyone and two projectors running in sync, where the operator swapped the lens cap depending on the vote. Its director, Raduz Cincerra, alive and living in Prague, told me "I did it as a joke but everyone believed it."

A related strategy is based on simulation: making web video more cinematic in style and making motion pictures more webcam like, for example.  There's evidence of this happening today.  Another hybrid strategy exploits high-octane cheap computers at each end of the narrow pipe.  We're beginning to see examples of the Sony Playstation II, whose power is similar to the SGI Reality Engines that drove first-wave VR, being used for live-networked gaming and beyond.

A most intriguing strategy turns the access argument into a feature.  One can use democratization ideology to justify piping 100 times the bandwidth available for one person into a collective space designed for 100 people (whatever the scale).  Public space as public research as public spectacle as community experience.  Such media-rich public experiments have a lively history, most notably 100 years ago around the birth of cinema.  An enormous opportunity exists for museums, libraries, alternative art spaces, and other public places to collaborate with scientists and researchers.