SIGGRAPH Special Report:
Web 3D graphics get boost as
VRML 2.0 specification is approved

Prior to this announcement, there were a handful of VRML browsers from Silicon Graphics, DimensionX and Sony. But with the specification's approval, developers and users can expect a plethora of dedicated VRML browsers, VRML plug-ins to Web browsers and increasingly, more sophisticated authoring software and toolkits.

A few trailblazing companies, such as Construct, Worlds and Planet9, have been creating 3D worlds on the Web using VRML 1.0, but the developer community had been awaiting anxiously for the latest specification to be finalized because it represents a major leap forward in terms of features and interactivity. It also signals that the dust has settled -- at least for now.

As good a start as VRML 2.0 is, not all issues were resolved when it was finalized on the first day of SIGGRAPH. Another barrier was overcome several days later at an after-hours VRML birds-of-a-feather session when Apple, IBM and VRML authoring tool vendor ParaGraph announced their joint support of a new compression standard, called Compressed Binary Format. VRML environments were already designed to be lightweight; now they'll be able to be sent efficiently over the narrow pipes of the current Internet network.

Technical issues aside, VRML 2.0 also represents a triumph of how the virtual Internet development community can work together to generate the framework within which it wants to operate. Initially through the VRML Architecture Group (VAG) and now through the newly formed VRML Consortium, the developer community has used e-mail lists and newsgroups -- the very stuff of the 'net -- to build something important. Though the newly approved spec is based on one originally proposed by Silicon Graphics, no one owns VRML.

A primer on VRML
VRML is a text format for describing 3D shapes and interactive environments on the World Wide Web. In addition to 3D animation, VRML 2.0 supports such multimedia features as text, graphics, audio and video. It also gives developers the capability to embed URLs in virtual worlds, so that a user could click on a door, for example, and be transported to another VRML world on the Web. VRML has been designed to work in conjunction with Java and JavaScript.

Why VRML?
Three dimensionality is important because it is closer to human experience. We live in a 3D world. Viewing 3D space on a monitor (or 3D-IMAX at the cinema) gives viewers a more immersive environment, a world in which minds and imaginations can delve more deeply than is possible when looking at a two dimensional representation, such as a photograph. With 2D, there's always the sense of the viewer and the thing viewed; with 3D, the potential exists for viewers to "forget" where they are and to enter another world. Until now, the Web has existed within the 2D confines of HTML pages, where even GIF or JPEG images are two dimensional.

A VRML world can be thought of as a more perfect human interface to the Web. Like HTML pages, VRML worlds have hyperlinks to either text data on HTML pages, or other VRML worlds. If you can imagine a time when VRML worlds have proliferated across the Web, linking between these 3D Web pages will be like moving through rooms in a building that knows no limit.

And it's not just about games and entertainment. While early VRML initiatives are exploring the addition of 3D worlds to chat rooms, or creating game sites, very quickly we are going to see interactive, three dimensional shopping malls and investment or banking sites where financial data is represented by 3D objects that change dynamically as the numbers fluctuate.

Another use being investigated by developers is the representation of abstract data. The online Yellow Pages service BigBook, Inc., based in San Francisco, is a creating an online 3D application that integrates a comprehensive business database with 3D replicas of major cities. Users seek out information by entering virtual buildings, with data being represented proximally.

VRML worlds will not replace HTML pages, but both options will give developers more choices for communicating their messages. And, VRML offers the promise of enlivening Web sites, attracting customers to an otherwise static site. Ultimately, the effect of VRML should be to bring users to the Internet who might not otherwise have ventured in.

Moving from VRML 1.0 to VRML 2.0
According to the specification overview,"VRML 1.0 provided a means of creating and viewing static 3D worlds; VRML 2.0 provides much more . . . a richer, more exciting, more interactive user experience." (See Resources, below, for a link to the full text of the spec.)

Specifically, VRML 2.0 provides the following extensions and enhancements to VRML 1.0, which was based on the Open Inventor file format:

• enhanced static worlds: More realism is defined, such as the ability to create irregular terrain instead of flat planes for ground surfaces and 3D spatial sound-generation. Optimizing and parsing files are easier, which is good news for developers writing browsers.
  
• interaction: "No more moving like a ghost through cold, dead worlds," reads the specification. "Now you can directly interact with objects and creatures you encounter." Developers can achieve this interactivity by writing sensors that set off events or keep track of the passage time. Collision detection means no more walking through walls.
  
• animation: "Interpolators" allow developers to create pre-defined animations, some of which are moving, (doors closing, birds flying), changing color (the sun setting), or morphing (from one shape to another).
  
• scripting: a script receives input from sensors and generates events that can impact the entire "world." Scripting is what lets object and creatures move and respond intelligently.
  
• prototyping: This new feature allows VRML developers to share material with each other, and incorporate aspects of each other's worlds.

Speaking at a SIGGRAPH panel he said VRML was now at the place computer graphics was about a decade ago: the "flying logos" stage. "We're going to see bad content for quite a while, but once the tools and browsers are strengthened, we'll start seeing 'smart' animations and interactions."

Binary compression standard proposed
Although the 2.0 spec covered a lot of ground, not all issues were finalized before it was released. Perhaps the primary outstanding hurdle to overcome was finding agreement on a cross-platform, high data rate compression standard. The announcement that IBM, Apple and ParaGraph had come to agreement on a unitary binary format compression scheme was greeted with cheers all around at the VRML Technical SIG held during SIGGRAPH. The VRML 2.0 Compressed Binary Format proposal achieves up to 50:1 compression ratios for ASCII 1.0 files, and up to 10:1 with respect to so-called "gziped files."

Images referenced by VRML will still be compressed using GIF and JPEG. "There's no need to re-invent the wheel," quipped Mitra, chief network technology officer at ParaGraph, who is a co-author of the binary format proposal (and who goes by one name).

"The implication of having a binary spec is that it will help proliferate VRML in the PC developer mainstream," explained John Alfano, who oversees 3D hardware and software at Apple. "It will be easier to make VRML work and transmit information with fewer barriers."

"Three-D graphics abuse computers," Alfano said.

The signatories to the agreement decided to break down their roles this way:

• IBM will serve as the lead to the combined engineering effort, working on geometric as well as field-for-field compression. The compression algorithm is highly compatible with the ASCII format.
  
• Apple will serve as the evangelist to the cross-platform multimedia developer community.
  
• ParaGraph International, of Campbell, Calif. and Moscow, Russia, will develop the cross platform code that supports the specification.

As one audience member, consultant John Chapman of Information Assets in Houston, put it: "The VRML community will drop kick it if IBM decides to charge for it."

The proposal offered in New Orleans was itself a work in progress; to follow the debate and see the final approach decided upon since SIGGRAPH, log into IBM's Web site on the subject. (Resources, below.)

In process: avatars
Another issue unresolved by the 2.0 ratification is how to standardize the use of avatars, VRML's non-carbon-based life forms that stand-in for users in a 3D world. To move this aspect of VRML development forward, a working group has been created sponsored by Chaco, Velosity and Worlds, Inc. A draft report called Universal Avatars addresses the data base issues that underlie the problem of how to let users create avatars that look the same across worlds created by different artists or vendors. (See Resources, below, to participate in the avatar debate.)

Although there currently is no universal standard, many companies are already populating their worlds with avatars, Black Sun and Oz Interactive, among them. These companies are exploring the possibilities of muli-user worlds in which one's avatar not only appears in a 3D environment, but can interact with avatars representing other people.

"The world of multi-user avatars is exploding as fast as VRML itself," said Maclen Marvit, vice president development of Worlds, Inc.

Moses Ma, chairman of Velocity, in San Francisco, listed the following issues as requiring further research:

• user profile information (willing to share or not)
• definitions (transitive behaviors)
• trust and transactions (real time commmunication with directory services)
• interworld mobility
• preserving information about your avatar in a secure manner (history)

The VRML players at SIGGRAPH

"As the vendors start making VRML tools and applications, my advice is to embrace and extend. You don't want companies competing on extensions like what's happening with HTML. Extensions have to be open. This whole area is far too important to be sacrificed to the big guns."

-- Andries Van Dam, speaking at SIGGRAPH '96
panel "VRML: Prelude and Future."

Netscape's 3D plans
While Netscape did not have a booth at SIGGRAPH, it had a presence on the podium at both VRML birds-of-a-feather sessions in the form of Yon-Jan Hardenbergh, who has been an active participant in the VRML Architecture Group deliberations leading up to the new specification. According to Hardenbergh, Netscape will be including VRML 1.0 support in Navigator 3.0. In addition, Live3D, Netscape's name for its overall initiative within the LiveConnect framework, will be a component of a Netscape Software Developers' Kit (SDK).

Netscape plans to integrate the VRML 2.0 spec into all its products within the next couple of months, said Hardenbergh, whose title at Netscape is, appropriately, VRML Practitioner.

Currently, nearly all third party VRML software vendors have or planning Navigator plug-ins, including Axial, Black Sun, and Oz Interactive, among others.

Microsoft's parallel universe
Microsoft, a newcomver to SIGGRAPH, which has been the premier computer graphics conference for 23 years, proclaimed itself on its booth as "the platform for 3D graphics." The company has clearly committed itself to a number of new strategies lately, and VRML is one of them.

However, Microsoft is taking a different tack from the rest of the VRML community with its bifurcated approach. It will support VRML 2.0 and at the same time is promoting a separate 3D path it calls embedded interactive animation, or ActiveX Animation (formerly, ActiveVRML). This approach incorporates 2D with 3D graphics and other multimedia components such as audio and video.

"Immersive large 3D spaces created solely in VRML have spectacular potential for the future," explained Salim Abi Ezzi, a Microsoft engineering manager for graphics and multimedia, at a VRML panel discussion. "But, in the short term we think the opportunity is for integrated media -- embedded 2D, animation and sounds."

"Our message is 'Don't focus on Snow Crash and ignore the low hanging fruit'," said Abi Ezzi, referring to Neal Stephenson's cyberpunk cult novel, which has proven so prophetic regarding 3D worlds.

A scenario for ActiveX Animation's approach to integrated media would be a sprite interactive animation textured in 3D, combined with a simple 3D object in a 2D composition space. Mixed in would be a few "randomized and parameterized seed sounds."

Meanwhile, as a sign of its commitment to VRML 2.0, Microsoft announced at SIGGRAPH that it has licensed a portion of Dimension X's Liquid Reality. It will embed this Java-based VRML platform technology into the Internet Explorer 3.0 and upcoming versions of the Windows operating system. Microsoft will also incorporate VRML 2.0 support as a downloadable ActiveX Control for Internet Explorer 3.0.

One of the biggest players in the VRML arena is Silicon Graphics, whose Moving Worlds proposal was the basis for VRML 2.0. (To read about the initial selection of SGI's standard, see April 1996 NetscapeWorld article, The Earth Moved, and So Will Everything Else.) At SIGGRAPH, SGI announced the availability of the Cosmo Player API. Cosmo Player is a dedicated VRML 2.0 browser as well as a plug-in to Netscape Navigator that allows users to navigate 3D virtual worlds. The API allows Web developers to extend Cosmo Player's functionality by supporting new data types and additional features. They can create cross-platform Java plug-ins to add multimedia capability such as streaming media and real-time data feeds.

A week before the show in New Orleans, SGI introduced the Cosmo3D platform-independent graphics toolkit. Implemented in C++, the toolkit supports more than 30 graphic file formats, including VRML 2..0. It is accessible to Java applications through Java bindings.

At SIGGRAPH, Black Sun Interactive announced support for multi-user interaction in VRML 2.0 worlds through its CyberHub client. In a prepared statement, the company indicates that adding multi-user interaction to VRML 2.0 browsers "enables the creation of business applications such as intra-company conferencing, interactive customer services, and online sales showrooms." CyberHub is available for Netscape Navigator's Live3D and now, Silicon Graphics' Cosmo Player. Black Sun, located in San Francisco and Munich, Germany, also announced that it has acquired the software rights to Attic Graphic's avatar technology, which means that avatar construction will soon be a part of Black Sun's authoring tool.

One of the only other companies to currently offer multi-user VRML software is Oz Interactive, with headquarters in San Francisco and offices in Reykjavik. Featured in the Intel booth at SIGGRAPH, the developers from Iceland presented a show in which a singer's movements were captured through her wired body suit and Webcast live on the Internet. The company exhibited with Intel because its Oz Virtual free client software runs on a Pentium PC, and was demonstrated by Dr. Andrew S. Grove, Intel's president and CEO, at the Intel Internet Media Symposium in San Francisco in JulyIn early fall Oz extensions will run in both Microsoft and Netscape browsers.

A newly formed Silicon Valley company, Axial Systems, announced it will provide software tools based on VRML 2.0 under the product name Axial Flyer. The company claims its VRML 2.0 browser is "fast enough to play Doom scenes at Ôgame speed' without hardware acceleration."

Axial will license its Axial Flyer toolkit to developers creating viewers for multi-user chat, online games and information access applications that play VRML 2.0 files. The toolkit consists of a VRML 2.0-compliant engine, a real-time 3D scene manager and a high-speed scan line renderer with sample source code for creating Netscape Navigator plug-ins and ActiveX controls.

At an evening VRML showcase event at SIGGRAPH, ParaGraph demonstrated its new professional VRML 2.0 authoring tool, Internet3D Space Builder, which is now in beta and available from its Web site. The first PC-oriented VRML tool, ISB comes with a library of editable 3D shapes combined with interactive texture mapping. ParaGraph customers who have already used the company's consumer tool, Virtual Home Space Builder, will be able to import VHSB's <.mus> format into the new product.

North Carolina-based Virtus, a 3D tool developer, did not introduce any new products at the show, but did announce its development of the Smithsonian's 150th Anniversary Traveling Exhibit, which it calls "the largest 3D Website in the world." The site was created exclusively with Virtus WalkThrough Pro and 3-D Website Builder.

Chicago-based VREAM introduced what it calls the "first true VRML 2.0-compliant virtual reality authoring tool and Web browser product suite for Windows 95." The drag-and-drop authoring tool , VRCreator 2.0, draws on objects, colors, textures and behaviors from a library of pre-built components, but also allows customization. It contains a plug-in architecture that allows additional modules to be created by third party developers. Planned plug-in components include editors for avatars, animation, floor plans and behaviors. Meanwhile, the WIRL 2.0 VRML browser is available as a stand-alone application, a Navigator plug-in and an ActiveX control for Internet Explorer.

• a board of directors
  
• a standards and specifications committee to propose extensions, etc. It will have two working groups: networking and design integration/languages. There will also be a design review board -- already being referred to as the DRB -- which will subsume the earlier functions performed by the VRML Architecture Group (VAG), which will cease to exist. Pesce dubs the DRB as the "Supreme Court" of VRML.
  
• conformance and compliance, to work with the National Institute of Standards and Technology, which is part of the U.S. Department of Commerce. NIST is designing conformance tests for VRML.
  
• marketing and education, which will oversee the effort to create a VRML-branded trade show, the details of which have yet to be worked out.

Gzip

An ASCII compression scheme that has been used until the recent the binary file compression proposal.

Interpolators

A type of animation object defined by the VRML 2.0 spec that can be pre-defined to commence at specified times. Interpolators means that worlds can contain moving objects (such as doors that open or birds that fly), objects that change color as they move or morph into different shapes.

node/fields

A node is the most basic grammatical element of VRML scripting. David Nadeau, staff programmer at the San Diego Supercomputer Center, defines a node as a "nugget of scene information," for example, shapes or colors. Inside of nodes are fields, which describe attributes that can be changed. Nadeau was the instructor for the "Introduction to VRML" course at SIGGRAPH '96.

routes

An important addition to VRML2.0's scripting capability that refers to the ability to pass events around among nodes. These special statements are called "routes." This capability is part of what gives VRML 2.0 it's dynamic quality of movement and interactivity.

"Be aware of premature lock-in. Consider VRML an experiment in process. Don't be in too big a hurry to build the next version of the spec until you've had experience with it, especially before turning it over to a ratifying body."

-- Andries Van Dam, speaking at SIGGRAPH '96