Web Techniques Magazine
September 1997
Volume 2, Issue 9

Special Report

VRMLocity and 3D Design

By Sue Wilcox

Intensive tutorials by industry experts are the special appeal of the 3D Design Conference, and this time around (June 1997) they drew large, rapt audiences to the San Francisco Marriott. This year the conference devoted an entire day to seminars focused on 3D on the Internet. Titled "VRMLocity," this day offered plenty of VRML-related fare as well as tutorials on a wider range of 3D experiences: becoming a 3D artist, using AutoCAD 14, creating special effects with 3DStudio MAX, and sculpting, digitizing, and animating characters. The other three days comprised four extended sessions per day, each offering six different tutorial-based classes that included product-specific sessions, character creation, 3D artistry, creating textures and special effects, and using VRML. Attendance this year increased over three times from last year, so the organizers must be meeting a real need.

VRMLocity

VRMLocity attempted to cover all the bases of concern to 3D artists designing for the Web. Bruce Damer, codirector of the Contact Consortium, gave the keynote: "A Journey into Our Future." "VRML is the mechanism to carry your content to a mass audience," said Damer, and "the future of the Internet is communication and personal expression space." Damer discussed uses of VRML for character creation (Floops from Protozoa), avatars for communication (OnLive's talking heads), and his research into L-systems, which algorithmically generate large (250K) VRML creatures from tiny (400-byte) seed files. He demonstrated VRML 2.0 features for adding behavior to objects using a dancing lobster that can sense when the mouse is near it and react. He also showed large-scale worlds being developed by Circle of Fire that have thousands of registered inhabitants and about 500 constant users. Encouraged by the popularity of virtual communities and the growing use of 3D accelerator cards, Bruce foretells a great future for VRML as an art and communication medium.

The audience's questions revealed frustration and confusion over the present state of VRML browsers. Damer could only sympathize and remind us that "this is all early adopter stuff now" and that "the Web is trivially easy compared to this." The good news is that VRML has attracted some major investors and that browsers are improving. The platform battle between Microsoft and SGI/Netscape should drive improvements in browser technology and ensure fast implementation of the rest of the VRML 2.0 specification. The ISO process is nearly complete, and this will further lock down the core of VRML. Meanwhile, the balancing act between standards and experimentation continues, with VRML providing the interoperability essential for content providers to try out new uses for it.

Avatar technologist Kirk Parsons, creator of avatar-authoring software and a talented artist, demonstrated state-of-the-art 3D avatar creation using his Figure Sculptor tool. According to Parsons, multiuser technology is VRML's killer app, but reliability and stability are the biggest barriers to entry for new users. His work at Black Sun Interactive has convinced him that although maintaining open standards and stability is difficult, "multiuser will be part of every Web site." He took us through the production pipeline, recommending 3DStudio MAX, Nichimen, Softimage, Lightwave, and Alias/Wavefront as the best tools for character modeling because "getting the right geometry is the critical part; output to VRML is not so bad." Texture mapping is also really important, as it can make a low-resolution figure look hi-res. Kirk shared some of his secrets for taking texture maps from detailed figures, then using them on reduced polygon figures. He warned, however, that in the absence of texture maps, it's better to work in low-res from the start.

Adrian Scott, publisher of VRMLSite, chaired the VRML and Commerce panel, which mixed artists with technologists and entrepreneurs. One topic was users' perception that VRML software doesn't work reliably-it needs to work out of the box, or preferably come built in to the main browser. Evolution is happening. As David Colleen of Planet 9 Studios put it, "browsers are crawling out of the ooze," and his work demonstrates impressively what can be done with VRML now. Dan Greening of LikeMinds believes that "personalization of 3D spaces should make them more popular," so his company is concentrating on collecting preference files on users to match them up with each other and with personally targeted advertising. Chip Morningstar of Electric Communities saw expressiveness and stability as key issues for VRML. He sees VRML 2.0 as stabilizing but feels the multiuser-technology side is not yet well thought out: "...standards are down to implementations that become de facto standard." The panel then debated which browser was likely to win the standards battle. Chip's sound advice to artists: "concentrate on honing your skills as a designer and 3D artist. The tools will catch up with you."

Saul Kato of Sven Technologies explained his work with photo-based textures. He describes the human face as "the epicenter of recognition" and concentrates on precise creation and application of textures from photographs. Although he works with high-res images, he uses proprietary polygon-reduction technology to slim down the geometry before exporting it to VRML. (In the exhibit hall Sven displayed a whole-world polygon-reduction tool that worked with just a slider control.) Saul also gave us his wish list for VRML: to progressively stream data geometry and textures, to handle multiresolution geometry, renderer standardization, low-bandwidth animation, and to support new interfaces such as force-feedback, motion capture and shutter glasses.

Eric Chen, inventor of QuickTimeVR and now with LivePicture, presented his Image Worlds specification, which combines VRML with panoramic technology. He and Pete Falco, a LivePicture engineer, took us through the addition of photographic panoramas and image based objects (IMobs) to a regular VRML scene graph. Besides allowing spherical panoramas to be part of VRML scenes, the new specification shows how to add screen elements that are always visible, like the gun in first-person shoot-'em-up games, animated sprites, video textures, and IMob sensors. LivePicture's FlashPix format allows the addition of multiresolution zoomable pictures or objects to the panoramic scenes. Eric gave us a quick look at his new PhotoVista tool, which combines twelve photos into a panorama with only a couple of minutes work.

Dave Marvit, moderator of the panel discussion on VRML content, explained that "a good artist has to push way beyond the technology," and this panel pushed from many directions. Linda Hahner of Out of the Blue Design showed her work from a painterly fine art background, Zara Houshmand of New Ether Studios spoke as a world designer from a theatrical point of view, Mark Meadows of Construct uses comic-book formats in his creations, while Parker Morris of Electric Communities is inspired by film to achieve a series of camera viewpoints to add atmosphere to a scene. The combined effect is of rich artistic inspiration producing beautiful, compelling, interactive scenes and stories. These VRML artists show they can take established art forms and grow them into a new medium whose possibilities are clearly worth exploring.

Maclen Marvit explained adding behaviors to VRML characters. He defined behaviors as actions that modify the state of the world, or time that makes the world change, or animations that produce change in the object animated. Maclen looked at the basic elements and how changes to a world are effected in VRML. VRML uses sensors, routes, and effectors as the components of a behavior. But how well the components made in one authoring tool work in another's browser brings up the subject of interbrowser compatibility, today's main constraint.

Intervista fielded a complete VRML production team to explain how they work together to produce complex VRML projects. They took us through all the roles, from the VRML foreman, who lays out the design, models, and supervises the whole process; to the character modeler, who works on the animations; the programmer; and the integrator, who prototypes original ideas, edits VRML source, and sets up behaviors to check the inline files work correctly. All the roles overlap, and the team approach is crucial to success. Unless the modeler understands the need for low polygon count, the scene will run too slowly, unless the programmer understands what the artists are trying to achieve, the code optimization won't give the right visual effect. This presentation really brought out the interdependence of the members of the production team and the fact that VRML is a hybrid medium. Intervista's WorldView browser is now complete so the team has a stable, fully implemented viewer on which to test their VRML. Look for it as a built-in feature of Internet Explorer 4.0 or download it from the InterVista site.

Exhibits

Unsurprisingly, the exhibit hall contained many graphics-accelerator boards. At one booth I found the new chair of the VRML Consortium, and 3D Labs VP of marketing, Neil Trevett. Elected as a neutral candidate, with no browser affiliations to cloud his vision, Neil acknowledges that the confusion and a lack of certainty about VRML-browsers functionality is reflecting badly on the specification. He seeks to reassure users that the VRML 2.0 specification is complete, that browsers are complying with the specification, and that there are solid business applications for VRML. He sees the Consortium as a leverage point-a catalyst and coordinator of the research being done by the consortium's working groups.

Several booths held particular promise for 3D designers and developers:

Draco Cybersystems. Draco was looking for artists and developers to produce content with the beta version of its V3 Virtual Reality Technologies and its World Processor Development Environment. This unique multiuser environment supports objects that are independent programs. One use is production of avatars containing genetic algorithms (GA), a way of learning from experience and "breeding" more successful programs. They envisage game characters that can evolve to assist players and even take control for themselves, following human instructions, to act semi-autonomously as intelligent agents. Draco also plans to use its technology for data-mining stock-market information and in chat rooms with an attribute grammar system they are developing that will let a computer understand commands typed in English.

URL Resources

3D Design Conference
www.3dshow.com/frames.html

3D Labs
www.3dlabs.com

VRML Consortium
vag.vrml.org

NewTek
www.newtek.com

Sven Technologies
www.sven-tech.com

LivePicture
www.livepicture.com

Intervista
www.intervista.com

LikeMinds
www.likeminds.com

Planet 9
www.planet9.com

VRMLSite
www.vrmlsite.com

Electric Communities
www.communities.com

Out of the Blue
www.outoftheblue.com

Credo Interactive
fas.sfu.ca/lifeforms.html

Datapath's RealiMation
www.realimation.com

IBM IntelliStation
www.pc.ibm.com/intellistation

Draco Cybersystems
www.best.com/~mantic

NewTek. The company's LightWave 3D 5.5 has a facial animation tool, Morph Gizmo, that caught my eye. After attending Georg Mastri's class on creating and animating the face, it was a relief to see this tool simplify morphing between an object and several targets. Using this for animating facial expressions or phonemes for lip synch is as easy as dragging a slider and watching the result in real time on a standard PC. The VRML 2.0 export support from this product is good (it handles animation, routes, and sensors) and is getting better.

Credo Interactive. Tucked in a corner of the ElectricImage booth were Credo's stunning Life Forms animation software. This tool lets you easily and rapidly animate single or multiple people or multilegged creatures. Their unique key-shape feature is really useful for seeing how an entire animation sequence looks while still in the keyframer. And it exports to VRML so you can incorporate animations into your Web site.

Datapath. RealiStorm, a realtime 3D manipulation and visualization tool is a plug-in that runs inside your favorite design tool (AutoCAD, 3DStudio MAX, or MicroStation) converting the slow running original format to RealiMation and letting you interact with it in this form, then convert back or export in a variety of formats. VRML is just one of many export choices, or you can output a demo AVI file, or stay in RealiMation, add actions, and use their viewer for Internet distribution.

IBM. The Intellistation Z Pro with Active Matrix thin-screen monitor attracted a lot of attention. Weighing only 20 pounds and a couple of inches deep, the monitor can be hung on the wall or held at a convenient height using its heavy stabilizing stand. It's more robust than a conventional monitor, and uses a lot less power: 20 watts compared to 200 watts in a regular monitor. The only snag is the price:$5,000. The box itself is a 200-MHz Pentium Pro.


Sue is co-author of the game-theory book EZ-GO: Oriental Strategy in a Nutshell and is working on a new book on 3D avatars. She regularly writes abut virtual reality, VRML, and 3D graphics for Web publications. Sue is VRML Pro at inquiry.com in San Mateo, CA.
Copyright Web Techniques. All rights reserved.
Web Techniques Magazine
Last modified: 2/5/97