From Multimedia Futures, a sister publication
The new standard for building interactive, moving immersive worlds is soon to be finalized and not just for the Internet. VRML, as it stands is pretty useless. The original Virtual Reality Modeling Language is only good for modeling realities where nothing changes, nothing can be touched, or moved. They are static, sterile worlds, useful for fly-throughs, and little else. The standard’s progenitors were well aware of the limitations from the start and always planned to remove them. VRML 1.0 was our way of trying to capture everyone’s attention… VRML 2 is the real deal says Rikk Carey, one of the fathers of VRML and Silicon Graphic’s director of VRML engineering.
by Chris Rose
VRML 2.0 aims to be this fix, opening the way to multi-user worlds of movement, interaction and sound. Its fans also claim it has potential away from the Internet, as a way for games designers or CD-authors to define and build interactive 3D Worlds on stand-alone machines. So is VRML 2 ready yet? Well no; in fact the specification process didn’t get underway until the informal VRML Architecture Group’s (VAG’s) request for proposals closed on February 2. There are in fact six different proposals vying for the attention of the VRML community, or at least there were; late last week Apple Computer announced that it would throw its weight behind the front-running proposal Moving Worlds. To introduce a measure of democracy the VAG has set up a Web page on which the VRML community can vote for their favorite proposal. The polls close on March 18th and on March 27th the VAG intends to make its formal proposal for a VRML 2.0 draft. The group expects the whole thing to be finalized by July 14th. They don’t believe in hanging about. To call Moving Worlds the front runner is to do an injustice to its position. It is so far out in front that you need a telescope to see it. As we went to press, Moving Worlds was in the lead with 74% of the vote. The proposal has the backing of 57 organizations, including Silicon Graphics, Netscape and Sony – in addition, many of the luminaries in the VRML architecture committee itself are intimately involved in Moving World’s construction. As Wolfgang Broll, one of the architects behind the competing German ‘Dynamic Worlds’ proposal puts it VAG and Moving Worlds are almost the same people. Broll is not terribly optimistic that many of his ideas will be included in the finished VRML 2 specification. In a nutshell the differences between the proposals can be oversimplified into three areas. One: whether VRML 2 should be an evolution of the original standard, or whether it should begin again with something new and different. Two: Ascii or binary file structure? Three: how to cope with multi-user worlds?
The standard disagreements
VRML files today are presented as ASCII text and can be written by a user with a text editor. The syntax is replete with descriptions of polygons, cylinders and cubes. Some proposal’s such as Sun’s HoloWeb suggested junking this approach and giving VRML 2.0 a more RISC-based feel, where all complex objects are built from a few very simple objects – lines, 3D dots and triangles. The resulting files would be easier for programs to parse, though virtually impossible to create manually. Sun and others further proposed that the format be saved in a more compressed binary format, rather than spacey, wasteful Ascii – it was a cry taken up by other proposals, and in particular Apple’s who’s three-dimensional MetaFormat (3DMF) binary file structure has now been adopted by Moving Worlds. When it comes to how the standard should support multi-user interactivity, there is another split. Some of the proposals suggested that multi-user behavior should be integral to VRML 2. The Moving Worlds team strongly disagreed: We believe that the multi-user aspects do not belong in VRML said Silicon Graphic’s Carey. He and his colleagues believe that external programming and separate multi- user communications protocols should
be used to attain group interactivity so as not to muddy the waters. Carey says that there are a number of initiatives already running in the labs which show how VRML2 can cope with many people. A thing called VRML+ iis one such approach.
A collaborative process
Though there are many other arcane differences between the proposals, the important thing is that the debate about VRML’s future is being conducted swiftly and in a very civilized manner. The VAG sees it as a collaborative rather than a competitive process – and says it intends that the best parts of everyone’s work are bought together. Since Apple has now folded its work in Moving Worlds, it’s 3DMF binary file standard will be sucked in. It is likely to be joined by aspects of Sun’s HoloWeb work into geometry compression and scaling. Assuming for the moment that VRML 2 will end up looking rather like Moving Worlds (and Rikk Carey has to keep correcting himself when he pre-emptively calls it VRML 2), its worth looking at what it offers developers. As with VRML 1, Moving Worlds defines its worlds in terms of nodes. A node can represent anything – it could be a thing, like a sphere, it could be a piece of script describes a certain piece of behavior (such as a rotation, or a switch node). Moving Worlds basically adds several new types of node to VRML. Starting with the basic graphics, it is now possible to create ground-and- sky backdrops to scenes, add distant mountains and clouds, and dim distant objects with fog. Currently VRML worlds stand on flat, quiet cyberspace plains, but Moving Worlds adds sound generation nodes and irregular terrain. Moving on to interaction, new sensor nodes set off events when the user moves in certain areas of a world, clicks on an objects or drags it about. Another kind of sensor keeps track of the passage of time, providing a basis for everything from alarm clocks to repetitive animations. New Collision detection properties ensure users no longer walk through walls, while terrain following lets them travel up and down steps or ramps.
Animated scripts
Animation itself is accomplished by new script nodes which can be used to give objects or creatures the semblance of intelligence. A script takes input from sensors nodes and then makes changes to other nodes as required. So, if you click on a little light switch, that click will be registered by a sensor node, which will send the event to a script node, which will send a message to the node that looks like a light bulb, changing its color. It may even go further than this triggering other script nodes that will result in the virtual dog waking up and barking. In summary, the authors believe that the language is rich enough to describe anything you will need to find in an interactive, immersive 3D world. Faced with the question of the serious commercial uses of VRML on the Internet, Carey is candid. He agrees that most VRML 1 applications today are essentially gee- whiz demo’s and admits it is not certain that VRML 2 will fare better. Though there is talk of 3D point-and-click shopping malls and virtual interactive museums, Carey says he simply doesn’t know what the final applications will be: the honest answer is that we don’t know which [application] is going to crack through the ice.
Something new for CD-ROMS
He is, however, relatively gung-ho about the applicability of VRML 2 outside of the net, among CD-ROM and games developers. He thinks it may even change the way authors think about their creations. Today, the CD-ROM market is dominated by pixel-based graphics – the user clicks on something and a picture, or animation is displayed, click on something else and something else is displayed. However the animations are all deterministic, the artist needs to think about each individual frame. Carey points out that once the objects are modeled in VRML, the rules change and the author can worry more about story-boarding than the raw frame-painting. As an example, Silicon Graphics’ favorite Moving Worlds demonstration,
called Gone Fishing throws the user into the middle of an aquarium. A sign on the wall warns visitors not to touch the fish – ensuring that they do just that. The fish busily swim around the tank, hurrying away from the intruder whenever the user gets too close. But clicking on any particular fish will trigger a characteristic behavior. One goes belly-up, for example, another attacks. The point is the general behavior is defined, not the second-by-second appearance.
Rough and ready rendering
Currently VRML-type representations look rough-and-ready, compared to detailed, pixel-based work, but as rendering and texture mapping techniques improve, it is possible that modeling rather than drawing could become the next big thing in CD-ROM design. The inclusion of Apple’s binary 3DMF file format also has some intriguing implications. Apple is obsessed with the idea of integrating 3D objects into the general desktop environments. 3DMF should/may/could open the way for the user to cut the wriggling fish from the aquarium and paste the still-twitching body into his or her word processing document. Why anyone would want to do such a thing is another of those technology-looking- for-a-problem questions that beset VRML 2. No doubt someone will think of something creative. Will we get VRML 2-based shoot-em ups? Probably not since it will be too slow and cumbersome for people who’s single goal is speed. However Carey thinks that developers may find it useful as a way of exchanging objects and designs collaboratively while the game is being designed. The bottom line is that this summer multimedia authors will be handed another powerful tool for creating content on the Internet and stand-alone machines. Not even the experts profess to know what will be done with it, but one thing is for sure – unlikeVRML 1, it should be useful, as well as pretty. *
VRML2: The Contenders
ActiveVRML Votes: 5% (author: Microsoft) URL http://198.105.232.4:80/intdev/avr/
Microsoft’s ActiveVRML is a complete departure from VRML1, introducing new syntax. Specifications are detailed, and rather than describing a geometrical language, it uses a modeling approach of reactive behaviors. It is close to a specialized programming language but has been widely ignored by the Internet community.
Dynamic Worlds Votes: 6% (author: German National IT Research Center) URL http://wintermute.gmd.de/vrml/
A consortium of German academic institutions adapting previous work of collaborative environments. Support for multi-user behavior is intrinsic to the system, scrapping the VRML nodes on a graph in favor of a system of objects, each with their own properties and interactions.
Holoweb Votes: 4% (author: Sun Microsystems) URL http://www.sunlabs.com/research/tcm/holoweb/
Another audacious departure, Sun proposed all complex objects are extraneous and should be replaced with combinations of simple shapes. Files would be easier for the browser to parse, but proper tools would be needed to build worlds. Though this aspect of Sun’s proposal will likely be rejected, others, including innovative compression techniques, may not.
Moving Worlds Votes: 74% (lead author Silicon Graphics) URL http://webspace.sgi.com/moving-worlds/
How can they lose? With the backing of 57 organizations and most of the VRML founders on board, it represents an evolution of VRML. New aesthetics and node types handling animation have been added. The authors believe multi-user capabilities lie outside the scope but they are capable of supporting them with external programming.
Out Of This World Votes: 11% (author: Apple Computer) URL http://product.info.apple.com/qd3d/VRML20/
Apple is now part of the Moving Worlds camp. Apple’s was a limited proposal: a plea to use its smaller 3D MetaFile format that underlies QuickDraw 3D, doing for VRML what RealAudio and Xing do for audio, including ‘streaming’, meaning browsers can display worlds on-the-fly rather than waiting for the whole f
ile to download.
Reactive Virtual Environment Votes: 0% (author: IBM Japan) URL no url available
Given that IBM itself is supporting Moving Worlds, the Japanese division must have had some inscrutable reason of their own for this proposal. Unfortunately they never posted the details, and consequently have received… uh, limited support. Details of voting can be found at http://vag.vrml.org. Remember, to be taken seriously as an aficionado, you must pronounce VRML as vermul.
