The costs of fully immersive VR equipment, with head-mounted displays and specialised input devices, is likely to make them prohibitive for all but the most specific applications. An interesting experimental result, pertinent to this point, comes from Satalich (1996) who looked at the claim that working within a virtual environment would teach the user to navigate naturally in that space. In fact, it was found that a group who only studied a map of the space performed better than those who had 30 minutes of prior exposure to the same space.
Satalich suggested that there may be a number of reasons for these results, such as the novelty of the technology for its experimental subjects, or the limited time during which they where immersed. These concerns are certainly worth bearing in mind, but it does suggest that until they are resolved, we need not look on desktop systems as necessarily inferior. This is especially important when one considers the expense, technical difficulties and health concerns associated with head-mounted displays (Jones 1996).
What follows will concentrate on VRML-based desktop VR, but even this section must begin with a warning: VRML is new and is in a state of rapid development. Details outlined below will change, and URLs given may not remain valid for long. However, the major Web sites, such as the VRML Repository, will almost certainly still be present to give the most up-to-date information available.
This situation of constant change is a problem, as the technologies often do not work fully, but is also very exciting, as ideas that are proposed one month appear as beta versions of software soon after. This speed of development has the consequence that, although the version 2 of the VRML standard is not yet fully defined as of the time of writing, we already need to look more at its potential rather than at what is currently possible with version 1. Doubtless, by the time many have read this report, fully functional VRML-2 tools and worlds will be available. There is, of course, some risk with this approach, but our experience of the VRML community in the last year gives us confidence. The take up and interest in the subject has grown enormously, and has been matched by the development of tools. The rapid pace of change makes the Internet - rather than one-off reports - the best source for information on the subject.
It is worth a brief digression to understand why VRML has suddenly taken off in the past year. This is no doubt partly due to the perceived inherent excitement of virtual worlds. One only needs to look at the success of the cyberpunk novels Neuromancer and Snow Crash. Although the reality of the technology is very far from the fictions, the excitement generated by them is real. Much of what is fiction will remain so, but many ideas can also be expected to come to fruition in the near future following the efforts of major players in the computer industry such as Netscape, Sony, IBM, Intel and Microsoft, now actively interested in 3D across the Web.
The importance of the term virtual reality' must be seen in a different context for VRML thanis the case for much of the advanced, high cost, virtual reality systems. The aim of an Internet-based technology is to make the three-dimensional worlds and objects as accessible to as wide a public as possible. The models created must be able to be seen on a range of machines that are readily available to the average user. This is particularly true of many HE institutions which often have many less powerful machines in situ. Speeds of network transmission for models also pose another problem. So for the immediate future it is likely that we must satisfy ourselves with simple (that is, small) models. They are unlikely to approach what anyone might call reality' for a long time. The models must, therefore, serve other purposes to be effective and useful. But as we shall see, they can convey abstract qualities, and in the near future can provide 'places' for people to meet, and these are sufficient reasons to engage in the technology.
Information on the current VRML standard are covered by another AGOCG:SIMA report, which can be obtained online (Ashdown 1996). We shall therefore concentrate on some of the issues and concerns that have arisen from our own use of VRML, and where we see its potential.
There are two main classes of browser: stand-alone, or a helper application associated with a standard Web browser such as Netscape Navigator or Mosaic. At the date of writing , in our own experience, there are various issues and problems related to browsers.
Many browsers are not fully compliant and fail to display models which contain advanced features. This situation has probably arisen due to the perception that VRML-1 is going to be a temporary measure. Its functionality is limited, and seems more a proof of concept. To overcome limitations various browsers, such as Live3D, support additional features. Equally, many browser authors now unsurprisingly seem more concerned with working on version 2-compliant browsers, than with continuing to improve those for version 1.
Secondly, there is a problem with browser authors porting software to the Macintosh. Most of the browsers run on IBM compatible machines, with some for high-end machines (such as Suns and SGIs), but the Macintosh is not well supported at present. However many university faculties, especially those in art and design, have a heavy investment in Macintosh equipment, and it can only be hoped that this situation will change. The appearance of Live3D from Netscape on the Macintosh has slightly improved matters, but the basic problem remains.
VRML-1 is simple, in that it only allows for the creation of static worlds. The interaction with these worlds consists of navigating through this space, and the equivalent of WWW hotspots on objects within the world, which allow the user to move to other worlds - or to any other Internet feature such as web pages, ftp sites etc.
VRML is based on a subset of Open Inventor from Silicon Graphics, and it defines a set of objects and functions for modelling simple 3D graphics. These are known as nodes, which are arranged in hierarchies called scene graphs. There is a top-down arrangement in which nodes that are described earlier in a scene affect later ones, but this can be limited by the use of separator nodes. This allows one to, for example, colour one part of the scene, but prevent the colouring affecting following nodes. A VRML file is an ascii file which is interpreted by the browser and converted into a 3-D display of the described world. For example, the VRML author can specify the geometric shape of objects in the world, transformations on them, surface colouring and textures, how the world is lit, and the cameras which create the final display for the user (see Ashdown 1996 for details).
Due to the need for the files to pass through the Internet, and the need for the files to be displayed on low-end machines, there are size issues for VRML which have been addressed in the language design. Through in-lining, large VRML scenes can be broken up into smaller files which are then loaded as required.
Another feature - one that has been used by games developers for many years - is levels of detail, or LODs, which assist slower machines. If we are far away from an object, the browser can display a simplified version of it. As we move in closer this can be dynamically updated to display more detailed versions of the same object.
Despite the obvious value of GUI tools, some knowledge of the VRML language has proved valuable for many current users, who have found it useful to build a few simple worlds using this method as a means to understand the general principles of the standard. It is also useful where minor errors are detected in the file output by a GUI tool, since these can be simply edited in a text-editor. Similarly, application packages often do not fully implement the standard, since they are generally a modification of an existing tool. For example, linking objects to URLs may not be supported. Therefore the ability to get into the file created by an application to edit it as required may still be necessary in the short term.
The value of this, however, is likely to diminish as both browsers and graphics packages begin to fully implement the standard.
Graphics Multimedia Virtual Environments Visualisation Contents