This report is also available as an Acrobat file.

Contents
The Design of Virtual Environments with particular reference to VRML

The uses of spatiality

Theatricalities - realism and non-realism

If it is accepted that VR can, and in many cases should, transcend the borders of physical realism and non-realism, an obvious place to seek inspiration is the theatre. As with painting, there has been a significant transition throughout this century from a view of the theatrical set as a realist pictorial view though a window, to the idea that the stage is an arena where the set-designer can do anything at all, take whatever freedoms s/he chooses, in the service of the drama. Ironically, this view of theatrical design has now itself been cited by Laurel (1991) as a model for interface design a plea to interface designers to view their task as the construction of environments suitably equipped with 'props' in which the user's actions will take place.

The construction of an alternative reality as a context for drama might have something to teach us about the construction of virtual realities for education. Both seem to be about conveying meaning by constructing spaces which the viewer/user is invited to explore and interpret.

At the beginning of the 20th Century, the art of the theatrical designer was little theorised. There was a general assumption that, in addition to providing the entrances and exits, the furniture and the props, specified by the script, it was the job of the theatre designer to represent as accurately as possible the architecture, furniture and so forth of the period described in the play. That the set should do something other than represent the surfaces of a real scene was no considered. However there was, for example in the productions of Irving, an insistence on unity in the overall conception which already might begin to guide our thoughts about VR (Bablet 1962). In particular, contemporaries praised Irving's use of electric lighting to pull together the disparate elements of the scene (and no doubt also to hide in shadow some of the mechanical aspects of the set). As VR systems become more capable of handling complex rendering including the effect of multiple light sources, we shall almost certainly want to use lighting more selectively, and perhaps under the user s control, to make the VR speak more effectively. A recent (non VR) project at the Centre for Electronic Arts made interesting use of darkness in which the user was enabled to shine a virtual torch, in an attempt to engage the user more fully in the scene.

Edward Gordon Craig (1872-1966) was a follower of Irving who took theatrical design in a direction of particular interest for this report, departing from the naturalism of his predecessors. Why did he do this? What was it about a wholly naturalistic set which he felt did not serve his purpose, and how might his thoughts guide us in our choices about virtual realities? Craig s reasons were based on a desire for what we might call media integration, concentration and consistency.

Media integration

By avoiding naturalism, the mimicry of reality, Craig found that he could better integrate all the elements of the production - in his case: actors, colour, music, movement (Bablet, 1962 p43). One of the problems of virtual environments is that we shall probably want to embed other media in them: for example, text. If we took the naturalistic route we would be obliged to put text into virtual books on virtual shelves, which seems to forfeit many of the advantages of non-physical media. To watch a movie, we would have to operate a virtual VCR. But by avoiding mimicry, we may be better able to create 'realities' which provide a seamless integration of the different media types, while at the same time keeping the strengths of each constituent medium.

Concentration

By being selective and non-naturalistic, Craig also found that he could focus attention where he wished. This raises an interesting question in relation to VR in Higher Education. On the one hand, we might be keen to create environments which, through non-realism like Craig's, focus users attention on particular features, in a way that unfiltered reality makes it hard to do. On the other, we might want to create deliberately complex environments, where the onus is on the user to discover the truly important aspects, somewhat like the approach taken by (fictional) detectives when, faced with the chaos of the world, they are able to apply their attention to just those items which turn out to be important clues.

Also of importance to Craig, and included in the opening remarks of this report, was the question: is reality enough? One of Craig's problems with naturalism (at least, with unthinking naturalism), was its inadequacy to get beyond realism. There were several aspects to this. For example, in creating a scenario for Shakespeare, a naturalistic set could undermine the intentional non-realism of the words spoken (ibid, p45). Perhaps this seems an arcane problem, but in fact it is just a special case of the tension between symbolism and naturalism, and symbolism is by no means confined to the arts. When we consider something as simple as a graphical user interface, and consider how such a symbolic environment might be concretised as a virtual environment, it is clear that not all the advantages lie with a naturalistic solution. The non-natural symbolic nature of the GUI may actually be helpful, for example in making clear that this is only one of many possible metaphors for the operation of the computer, or in its ability to take liberties with scale so that all objects however near or distant have the same size.

Another Craigian idea valuable in thinking about virtual environments is the greater ability of non-naturalistic environments to suggest the unseen (ibid p134). In current practice, VR systems deal predominantly with surface appearances, and the systems in existence tend to emphasise the visual over the behavioural aspects of what they model - depiction rather than simulation. Craig's objection to naturalism in the theatre was partly based on his fear that it created a surface layer which distracts attention from the real messages of the play. The limitations of current VR technology have tended to enforce heavily simplified and stylised worlds upon us, so that we can not currently begin to create virtual worlds containing all the rich minutiae of the real world. A Craigian reduction to elementals has been forced on us by the technology! But when the naturalistic detail of the real world becomes more achievable in VR, we shall need to decide whether this is, for some purposes, both more and less than what we want: more because details may distract us from what is being said; less because we may need powerful symbols, and symbols normally work precisely by eliminating particularities of surface appearance. A historical example might be the work of the pre-Raphaelite painters, whose work seem now to lack any clarity of meaning but which preserves in minute detail the (irrelevant) surface characteristics of what the painter observed. Recent (non VR) projects in the Centre for Electronic Arts have tried to grapple with the problem of stripping away surface excrescence from multimedia products, resolving complex functionality into simple interfaces, while commercial software development by contrast has tended to grow extra buttons to deal with extra functions, so that the user feels s/he is interfacing with a layer of control devices, rather than with information itself. Reality is full of clutter, and we do not necessarily want our VRs to be the same.

Consistency

We remarked that the theatre has been adopted by Laurel as a metaphor for the human interface. One of the points on which she and Craig concur is in emphasising that the stage, considered as a static environment, is not in itself the point, that it is a location for action. Craig was untypical of his time in emphasising 'action - movement - dance' rather than speech, literature or poetry in the theatre. Clearly, part of the question 'What can we use VR for in HE?' must include, 'What should happen in these Virtual Realities?' We shall need to choose whether our VRs include only action by a single protagonist, the user; by multiple users, probably networked; whether the VR is essentially inert, responsive to the user s actions or whether it includes objects and agents with autonomous behaviour. This in turn means considering how people interact in shared spaces (see p35, 83), as well as whether the user has a first person or third-person relationship to the environment (see p55). Strangely, Craig became increasingly fascinated by the possibilities of a purely mechanical performance without the irritating distraction of real actors! He began to feel that sophisticated marionettes would be preferable (ibid p107 passim). He would surely have loved VR.

For Craig, the emphasis on the set as a site for action led to his facilitating in his designs the process of change, often by the re-configuration of modules. He likened the set to a face, which always has the same parts laid out in a recognisable pattern, but which alters its expression with any change in one of its features (Bablet 1962, p 126). Given the potential flexibility and reconfigurability of computer models, this is something we may want to consider in making virtual environments, potentially achieving both the unity and security which a single virtual 'place' provides, at the same time as the flexibility and multiple functionality of many places. Commercially this will be important, since it is a way in which an identifiable 'brand' of place can have many uses. While real buildings and landscapes are not generally reconfigurable, virtual environments will surely want to reject adherence to this limitation of the real world.

It might be thought that this report is advocating an anti-naturalist approach to VR, such as Craig adopted in the theatre, but this is not the case. The point being made is that in deciding on the role of naturalism in Virtual Environments, some of the reasons that Craig had for (in his case) objecting to naturalism may be worth considering for our own purposes. Above all Craig was a functionalist, in that he saw the purpose of the design as to serve the idea (not the other way round).

Omission and abstraction

Craig emphasised the value of omission and abstraction at the expense of naturalism. It is interesting to look at film and see what it omits, while at the same time seeming to be a total experience, whether for purposes of dramatic fiction or factual communication. Notably, it relies on only two sensory channels vision and sound. There have been attempts to incorporate smell in cinema since at least 1906, with Smell-O-Vision and Aroma-Rama both being launched in the late 1950s. Neither system proved popular, although technically they were both successful: the idea of scenting films was abandoned (Katz 1994). This raises the question of what other economies we can make in our virtual worlds. In 1851 a Stereoscope was exhibited in the Great Exhibition, and was patronised by Albert and Victoria. A viewing device with two lenses, it enabled a pair of photographs made with two adjacent camera lenses to record stereoscopic views. By the mid 1850s, one version had sold more than a million in England alone, and by the end of the decade the ambition of the London Stereoscope Company, 'no home without a stereoscope', was almost fulfilled (Macdonald 1979). All manner of exotica were available ghost pictures, moral tableaux, freaks and oddities. There was also pornography. Today the stereograph is relegated to the status of a child s toy, not taken seriously as a medium of communication. Crucially, its ability to present the third dimension is seen as unnecessary. What might have led to its demise, while straightforward photography has continued to increase every year since its invention? Stereoscopes were available quite cheaply, and many were originally bought, but the craze passed. Perhaps a means of communication which required a mechanical device for viewing could not be assimilated into normal life? But what about the gramophone and the television, both dependent on physical machinery? In the end, as designers, we are always free to make decisions about what can and should be omitted, and it is surprising what effective use can be made of media which offer only one or two channels of address.

The following quotation may cast an additional historical light on this issue:

There seems to be little doubt that there will be colour. When colour is perfected technically, the film will be an almost perfect copy of reality, almost 'the real thing.' With the perfection of the magnascope, the screen will be made to issue into the auditorium: the audience will be enveloped in the film. The approximation to reality will be still more convincing, the opportunity of imaginative creative work still more improbable, the cheap appeals to participate in a crude and unhealthy experience still more compelling.
William Hunter 1932 Scrutiny of Cinema p11

More spaces in literary imagination

The accepted definition of virtual reality of course relates to computer-stored, primarily visual, models which achieve their aims by visual specification of the attributes of objects and environments. We have looked at analogies with architecture and the theatre. It is worth thinking about the completely contrasted approach taken by books, especially novels, which achieve their effects with a bare minimum of specification. There is much less data in a book than any but the simplest computer graphic models, yet most readers would probably agree that the paucity of data in a book is not in itself a problem - that in fact books acquire much of their effectiveness through omission.

The gamut of literature covers the whole range of relationships between representation and reality, from the highly realistic to the fantastical. Literature provides a rich resource of ideas about space and spatiality, which we suspect is largely unresearched (exceptions include Barrell 1982). The essential idea is again to avoid assuming that space just 'is what it is', an inert and essentially irrelevant location for the action, and instead to arm ourselves with ideas about spaces with meanings, spaces which facilitate and augment the action as much as do the personalities of the characters.

Barriers

We have already commented on art and literature which depicts stratified societies, and emphasises the ease or difficulty of moving from one layer to another. A similar device though used for different purposes is proposed by Lewis Carroll in Alice through the Looking Glass. Once again we have a model based on some characteristics of real spaces, but with all the important qualities of the vision arising from the tension between this realism and the elements of unreality.

Pilgrimages

Another class of interesting proto-virtual spaces in literature, is where travel along a route is put forward as an analogue of spiritual progress. Again there is a long tradition: it is only necessary to mention Pilgrim s Progress and the Yellow Brick Road! In fact these linear spaces are a special case of a whole category of constructions placing symbolic elements in spaces.

From space to place

Distinct, but not separate, from the concepts of space, is that of place - the particularities of a location or a type of location what differentiates a cathedral from a railway station, or this particular library from that library.

All really inhabited space bears the essence of the notion of home.
Bachelard 1958

One of the most influential writers in this regard is Bachelard, whose 1958 book, the Poetics of Space, includes among others chapters entitled House and Universe, Shells, Corners, and The Dialectics of Outside and Inside. The great value of the book lies in its extensive coverage not of spatiality in general, but of the characteristics of particular kinds of spaces. While spatiality may be partly perceived in an a-cultural way, as we said before, clearly the characteristics of particular kinds of spaces are heavy-laden with cultural influences. For a space to be perceived as a house, makes a highly influential difference to the way it is 'read', compared with it being read as a public institution. In VR, where we have freedom to generate new space-types which evoke any existing genres of space, plus eventually completely new ones, what a space is perceived to be matters enormously. Both the thoughts and behaviours of users, their psychology and their actions, will be heavily influenced by whether a space is perceived as of the type normally thought of as public or private, intimate or challenging, large or small.

Large and small

Another virtue of Bachelard's work is that it deals with scale, explicitly in the chapter Miniature, but also as a theme throughout the book. Most discussions of spatial constructions concentrate on spaces of a given size, such as cities, public buildings, jewel-caskets. It is unusual to find the issue of scale raised at all. Scale is an issue for VR especially of the non-immersive kind, because of VR s disembodied nature. In reality, our physicality prevents us from reaching up to things higher than ourselves and from climbing inside things smaller than us. VR knows no such prohibitions, unless they are deliberately constructed. This is yet another way in which VR cannot, by its nature, replicate the lived experience of reality, and must remain always at some level a representation.

Contents

Graphics Multimedia Virtual Environments Visualisation Contents