VR: Theory, Practice & Promise




"Building Cyberspace: the development of the Virtual Reality Modeling Language"

Easily the most exciting development this past year has been the introduction of VRML, the "Virtual Reality Modeling Language." Proposed at the first International Conference on the World Wide Web in Geneva, Switzerland, by David Raggett of Hewlett Packard Laboratories, the idea of creating a three-dimensional interface to the World Wide Web (a "Virtual Reality Markup Language" as Raggett originally termed it) immediately generated a great deal of interest and enthusiasm. According to Raggett, VRML would be "a logical markup format for non-proprietary platform independent VR." It would be compatible with Web interface standards, such as HTML (hyperlinking) and HTTP (Uniform Resource Locator addressing system), and would allow for various effects such as texture wrapping, weather effects and virtual presence teleconferencing. At the same conference, Mark Pesce, Peter Kennard and Toni Parisi announced their own ideas for a World Wide Web-based three-dimensional interface. In a paper entitled Cyberspace, they called for a new Cyberspace Protocol (CP) which would not only be compatible with standard Web interface protocols, but would also work in conjunction with a viewer application they had developed called "Labyrinth." This new tool, in conjunction with a parser, would be able to interpret and display ASCII text files containing formatted descriptions of three-dimensional scenes, objects, hyperlinks, and information on light and color. According to Pesce, such a three-dimensional interface to the World Wide Web would lead to a more "human-centered" experience of the Internet. Encouraged by the favorable reaction their ideas received at the conference, Pesce and Parisi returned to San Francisco and, with the assistance of WIRED magazine, set up a WWW Home Page, the VRML Forum .

The goal of the Forum and its accompanying mailing list was to open up development to the community itself: to discuss the new specification and to solicit proposals for its realization. Rather than develop a brand new specification, the Forum participants chose to seek out an existing implementation as the basis of VRML, hoping to speed up the process. Several proposals for language implementation came in. These were: Paul Strauss' and Gavin Bell's SGI "Open Inventor", The Geometry Center of Minnesota's "WebOOGL", Autodesk's "Cyberspace Description Format" (CDF), Bernie Roehl's and Kerry Bonin's "File format for the Interchange of Virtual Worlds", Pesce's and Parisi's "Labyrinth", the "Manchester Scene Description Language", and Marc de Groot's "Meme." Through an online voting procedure, the ASCII format of Open Inventor was chosen.

The resulting specification, renamed "Virtual Reality Modeling Language", was written by Bell and Parisi. As an open standard and ASCII text-based scene description language, VRML 1.0 is compatible with HTML and HTTP and is extensible to other applications. When a three-dimensional scene has been produced and translated into VRML file output (the file extension is termed .wrl for a VRML "world"), the file can be put on a server, downloaded from a Web page and viewed within a specially configured viewer. Once the data is transferred to a viewer on the end user's home computer, a three-dimensional navigable scene appears. As currently written, the specification does have one major drawback, in that it does not account for object behavior. This means that objects within VRML environments do not have complex differentiated behavior from within the scene itself. This will, however, be corrected in version 2.0.

In light of the potential of VRML to make the Internet a much more visually exciting environment than it currently is - and hence one more likely to attract a mass audience, it came as no surprise that various high-tech corporations began to show an interest in the new technology. On April 3, 1995, Silicon Graphics, Inc. of Mountain View and Template Graphics Software of San Diego announced their "long-term development relationship" for the VRML specification, as well as "WebSpace", the first commercially available viewer for VRML. Compatible platforms named were SGI, Sun, IBM, Windows 3.1 and Windows NT 3.5 systems. Many companies and organizations came out in support of VRML, promising integration of the specification into their products and services. These included: Netscape Communications, Spyglass, Wavefront Technologies, Digital Equipment Corporation, Intergraph, NCD, NEC, 3Dlabs, Accel Graphics, Oki Advanced Systems, the San Diego Supercomputer Center, CERN, Viewpoint Datalabs, Brown University, and the University of Darmstadt. Silicon Graphics and Template Graphics also announced that a viewer for the Macintosh would be developed within the coming months.

A variety of people and organizations began to experiment with VRML once Pesce, Parisi and Bell made the source code of the new specification public. The results of these experimentations can be found all over the Web. One fascinating project was developed by David Blair and Tom Meyer, who turned Blair's art house movie, "WAX: or the discovery of television among the bees", into a giant hypermedia Web site displaying VR simulation sequences, along with movie clips, photographic stills, audio tracks, and computerized three-dimensional models. To build on the film's themes of traveling through a "Void", Blair and Meyer integrated over 250 VRML scenes into the site using a special text-based MOO program which Meyer adapted for the project. This giant hypermedia database narrative is the first of its kind and is truly impressive.

Another site to pioneer the use of VRML was the "VRML in Chemistry" project, sponsored by The Institute for Physiological Chemistry at the University of Darmstadt in Germany. Located at http://ws05.pc.chemie.th-darmstadt.de/vrml/, the site contains a large collection of 3D models which demonstrate different components of the cancer research enzyme Cytochrome P450. The visitor is given the opportunity to explore the structure of the amino acids of the enzyme, which are modeled in a variety of ways: as wireframes, capped sticks, or as balls and sticks. Navigating one's way through these colorful VRML models which float against a background of black space, one is struck by the beauty and complexity of the enzyme's structure. The site has an additional interactive element: visitors are even given the opportunity to make their own model of Cytochrome P450.

However, by far the most elaborate VRML project to date has been the modeling of the Interactive Media Festival's event showcase at the Variety Arts Center in Los Angeles. Sponsored by Motorola and other corporations, the 2nd Annual Interactive Media Festival took place during the first week of June, 1995. The goal of the modeling project was to recreate the exterior and interior of the Variety Arts Center (the location of the Festival) as well as the twenty-two installations which comprised the festival's content.

The IMF/Variety Arts Center VRML realization was a genuine team effort. Undertaken in San Francisco over a period of three months by a mostly volunteer group of network scientists, engineers, and 3D artists, this project was the most elaborate investigation yet of the possibilities and the limitations of the new language. Working together for the first time, this group of 13 men and 1 woman ranged in age from the late teens to 40 and up. Team members spent endless hours thinking, writing, discussing, reworking, arguing, designing, modeling, translating and testing VRML output. Communicating progress through a closed mailing list, the members of the team shared their thoughts, raves and flames remotely.

Led by Mark Meadows, the project coordinator and IMF Webmaster, the group consisted of Thomas Caleshu (modeling), Michael Gerstein (modeling), Todd Goldenbaum (modeling), Adam Gould (post-production and special effects supervisor), Michael Gough (principle modeling architect), David Lewis (system administration), Annette Loudon (texture mapping and Web administration), Jeffrey Gray (camera positioning and hyperlinking), Ian Kallen (team member), Steven Piasecki (file structures and naming conventions), Jim Race (image tracking), and James Waldrop (back-end tools and Perl scripting).

The implementation of the project was extremely difficult, due in large part to the fact that none of the available modeling or architectural applications had at that time direct VRML file output capability. Because Michael Gough, the project's chief architect was most at ease working with the architectural program Form Z, it was used to model the building's exterior and interior, whereas the rest of creative team used 3D Studio or Strata in order to model the festival's exhibits. Not only were things made difficult by the fact that none of these applications was capable of outputting VRML files, but the DXF files which each application did output, happened to be incompatible with one another - making it very easy for bad translations (images that were inside out, for example) to occur between the various formats. As Gough points out, a great deal of technical improvisation had to take place: "We often had to build our own tools or hack somebody's else's existing tools in order to get this thing done."

In spite of all the technical difficulties the team had faced, the project was completed in time for the festival. Festival attendees and visitors logging in remotely over the Internet, could navigate a scaled model of the Variety Arts Center and the various IMF exhibits. The model was fully hyperlinked: clicking on any object in a scene transported the visitor to another virtual location or to Web pages containing information about the exhibits. Ken Feingold's interactive piece, "where I can see my house from here so we are", was especially beautifully represented in VRML. Once the file rendered, three grey robots, sitting next to each other in triangle formation, appeared as a nightmarish Mad Hatter's Tea Party, perfectly composed and swathed in VRML darkness. As the viewer moved in for a closer look, the robots' yellow eyes became sharper and more mesmerizing at every turn. The scene was quite eerie.

To make sense of the IMF project and the significance of VRML, a special panel convened Tuesday afternoon, June 6th. Led by Jeannine Parker, the President of the International Interactive Communications Society, this panel consisted of the IMF/VRML team, David Blair and Tom Meyer of Waxweb, Mitra of Worlds Inc., Mark Pesce and Clay Graham (SGI's resident 3D modeler and cyberspace theorist). While some of the discussion was about practical and immediate matters such as the future of VRML in relation to other new Internet applications like Hot Java (SunMicrosystem's animation behavior specification), much of the time was spent coming to terms with the more abstract philosophical problems raised by the new technology. Graham, an outspoken proponent of the importance of design in cyberspace, argued that greater attention should be paid to the purpose and meaning behind the virtual structures themselves.

One of the more controversial philosophical subjects discussed by the panel was the notion of creating a unified "spatial" cyberspace. Possibly influenced by the "cyberspace-Matrix" described in William Gibson's cyberpunk novel Neuromancer and the "Metaverse" of Neal Stephenson's novel Snow Crash, some members of the panel agreed that there was a need for this type of unified experience. A vigorous objection was raised by Worlds Inc.'s Mitra, who argued that a unified cyberspace implied one worldview and that had unacceptable overtones of cultural imperialism. While a unified cyberspace is still a long way off, a new technology that was also featured at the Interactive Media Festival, aroused much interest from the panel participants as a possible way of organizing such a unified cyberspace. Essentially, a rotating earth-globe that is texture-mapped with photos taken from a Global Positioning System satellite, the T-Vision project from Art + Com of Berlin, is easily manipulated by a track ball. From an orbital position, using the cursor, one can designate an area within a city until just a few blocks away, the buildings, blocks and cars become distinguishable. At the panel, Pesce announced he would be working with the developers to see if it were possible to integrate VRML technology into the project.

Up to this point, the technological infrastructure to create VRML output has been extremely difficult and cumbersome. For this new technology to become as widespread as the Home Page phenomenon, there needs to be an end-user friendly way of creating VRML files. Recently, two companies have promised to make this a reality. Developers from ParaGraph International, which is based in Sunnyvale, California, have announced their three-dimensional authoring tool for the World Wide Web. A tool which will directly output to VRML, "Home Space Builder 1.0", allows users to render three-dimensional graphical environments using coordinated panel displays. Some of the displays include: a top down architectural blueprint maker, a texture-map library, a graphical interactive three-dimensional model, a point-and-click color library, and a special effects panel. Not only can the user prepare a three-dimensional space, but images, sounds, morphs, movies, animations, moving texture-maps, and hyperlinks can be added to the scene. Another product, the "3D World Builder" from Caligari Corporation of Mountain View, is also a VRML compatible authoring tool. Caligari promises that 3D World Builder will (the project was still in development at the time of writing), "incorporate many of the special features available in our flagship professional 3D graphics and animation package: "trueSpace2."

All in all, then, it isn't too far-fetched to predict that Home Pages will soon be superseded by Home Spaces or Home Rooms. It is to be hoped that a whole creativity industry will be spawned by new technologies such as VRML as corporations and developers vie to lure the online public with ever more exciting and involving Internet environments.

HOME, JAMES