This report is also available as an Acrobat file.

Contents
The Design of Virtual Environments with particular reference to VRML

Virtuality

Environment into process

Laurillard (1993) has argued that hypertext is not an educational medium, in the sense that it provides only an environment in which learning may take place, but falls short of helping form the learner's actions. Likewise a library is not a course. In a sense the educator must stand in the role of narrator, leading the learner through experiences: even in intensively resource-based learning, this role for the academic survives. A similar objection could in principle be raised to Virtual Environments, and most designers of Virtual Environments assume that other media will be needed to assist the learner in getting value out of the environments and objects offered.

However, a special case of the affordances already mentioned is worth considering in this context. Some kinds of environments seem to invite the user to explore them in a particular way. When people describe well-designed gardens or buildings, it is common for them to read into these passive environments a formative, perhaps narrative, quality to the space. One example stands for many:

The resulting spatial rhythm is much smoother than that of Romanesque cathedrals or of Noyon. It is no longer split into numerous units which one has to add up mentally, as it were, to summarise the spatial totality, but concentrated in a few, in fact three, sections: west, centre, east. [...] It leads you towards the altar as forcibly as did the columns of Early Christian basilicas.
Pevsner, 1943

Clearly the architecture invited certain trajectories on the part of the user. A recent project in building a VRML environment, at the Centre for Electronic Arts, explored the idea of concealment as a motivator - users tended to look into a space which they could glimpse but not wholly see. Again we are extending concepts familiar from recent attitudes to interface design: the designer must work not on the artefact itself, without taking into account the active processes of the user: what the user will see, hear, and above all do, in the environment.

Protagonists and point-of-view a case study from film

An obvious alternative to the idea of spaces which to some extent 'explain themselves' is that of inserting either a narrator or a protagonist to assist the user. Though confined to the presentation of fiction, feature film offers some thought-provoking exemplars, among them 'The Lady in the Lake' directed by Robert Montgomery and released by MGM in 1946.

In this film some of the conventions of Hollywood film-making are overthrown, especially in relation to the viewer's relationship to the protagonist. Through our familiarity with film we have learned to 'look through' the conventional structure narrative devices so that we hardly notice them, but they are conventional and not 'natural'. For example, who is it who sees what is happening in a film? Is it the viewer or the narrator? When the director wishes to show us, not what the protagonist is doing, but what the protagonist sees, then there are well-established codes for making this clear (typically a close-up of the eyes of the protagonist, or a shot which shows the direction of view of the protagonist in relation to other actors). Despite the fact that we are watching all the characters from the 'outside', we often 'identify' (or at least empathise) with the main protagonist. This effect can be strongly reinforced by the use of a 'first-person' voice-over narration. In the Lady in the Lake, however, we hardly ever see the protagonist, for the simple reason that the whole film is shot as it were through the eyes of the main character he - only appears in mirrors.

Why is it worth looking at this film, which is after all an authorial, narrative piece of fiction? Attempting to evaluate which aspects of the technique seem to work, and which not, may help us to foresee how in a realistic, filmic Virtual Environment we could best locate the user, and what problems the user may have in relating to this 'other world'. In evaluating strengths and weaknesses however, we have to acknowledge the prejudicial effect of more familiar, established filmic codes. What we see as 'right' will probably change.

Aspects of the film which seem unsuccessful (some of which have direct counterparts in existing VR techniques) include:

A quick glance from one part of a scene to another is never used. The camera must pan cumbrously in a way quite unlike human vision. Ironically, jump-cuts (where one scene simply stops and another start) may be more analogous to how we use our eyes in the physical world.
Likewise, when our attention is summoned from behind, it is difficult to convey the swiftness with which we can turn our attention to the source.
Parts of 'our' body must intrude in the scene. For example we strike a man with our fist; we turn a door-knob with our hand. Of course this hand is unlikely to resemble the viewer's own.
Looking in a mirror reveals our 'self' which again does not resemble the viewer. Perhaps experience with avatars (representations which stand-in for a person) in shared virtual environments suggests that our problems with this other self will be overcome (see Social space and Virtual Worlds, next) ?

Some aspects of this first-person/third-person view which seem more successful include:

The sense of immediacy when we eaves-drop on another s conversation we have a sense of really being there, hearing but unheard, seeing but unseen.
When characters address us directly or respond directly to what we do (such as showing fear) the effect is more potent than in conventional film.
Similarly, we feel guilty when we seem to hurt another character's feelings!
Direct assaults on 'us' are perceived more viscerally, less intellectually than when observed from a remote standpoint. This effect of course is widely used by violent computer games, some of which allow the player to switch viewpoint at any time between being in the action and observing it from without.

One of the points of this extended discussion, apart from simply drawing attention to an aberrant attempt at virtuality within the Hollywood tradition, is to recall that the filmic conventions which we find so convincing and unremarkable in the cinema themselves took time to evolve. Again we can only wonder what new forms of expression may be latent within VR. The early cinema was confined to simple documentary sequences and technical showing-off of various kinds: no one could have predicted the major role of the cinema in our culture.

Social space and Virtual Worlds

The attention of software designers has shifted in recent years to supporting shared workspaces and collaborative working of various kinds. At the same time in some Higher Education institutions there has been increased interest in students undertaking collaborative work, which is seen both as assisting in the educational process and in improving the student s preparation for working life. A partial shift to assessment by project-based coursework has also contributed to changing attitudes.

While some effective collaborative environments may not be actually visualised at all, others are given the visual attributes of actual spaces using VR techniques. These worlds can exploit social aspects to encourage the sense of place, rather than using purely perceptual cues to suggest only space. Shared worlds are made possible by changes to VRML. The VRML-2 standard aims not to inherently provide the new means to do this - it is not part of the standard - itself but instead to provide handles for the functionality to be added by additional routines.

Issues arise for such worlds as to how users - represented by avatars - can engage in this space. Currently, most Internet-based, VRML-oriented, multi-user worlds, such as CyberGate, allow the user to be represented with a standard avatar image. More recently it has been possible to create one s own avatar with VRML. However, these inevitably look somewhat cartoon-like. 'Social problems' have occurred previously in text-based shared worlds which may emerge also in the new 3-D worlds as these become more widespread. Having been available far longer, the text-based worlds have exhibited the problems first.

In particular there is a problem in the way that both the spaces and the other avatars are perceived by users. For some they are pure fictions, based as the technology is on games, and this characteristic is likely to be exacerbated by the cartoon-like graphics of the 3-D worlds. Other users, on the other hand, see their avatar as an extension of themselves. This mismatch has lead to many conflicting situations, such as the now infamous 'rape in Cyberspace', in which one user knew the system sufficiently well to manipulate the words and actions of another's avatar.

One approach to addressing this issue is to ground the avatar more firmly with the user. This is possible for example through the use of the user s own voice, rather than the use of typed text. The avatar can be based on one s photograph, rather than a primitive VRML model. And in future, it will presumably be possible to stream a video of one s face onto the avatar.

Augmented Spaces

Another approach to alleviate such problems could be to 'de-virtualise' the world by connecting it to a real space. Recent work brings the value of complete 'virtualness' into question, and a number of projects are concerned with linking the virtual to the real world to create 'augmented realities' in which digital technologies do not replace, but supplement, that reality. An example of such ideas is the work at Lancaster University . In this world, a 3D model of the research lab is connected to the real-world lab, and when people are logged into machines, their avatars appear in the virtual world standing before a model of a terminal. It is then possible to phone them, or use some synchronous computer communications, so establishing real-time contact.

A similar project is being undertaken at the University of Nottingham, who are building an Internet Foyer. This will link a three-dimensional model, based on their web site, to the physical foyer of the building. The model is projected onto the wall of the foyer, and the foyer is visible, through a camera, to the remote users. The intention is that users visiting either the web site or the foyer are aware of each other s presence: extending the social function of the foyer-space to encompass the virtual.

It is possible that by connecting the virtual world into a real space, the participant will become successfully 'grounded' in it. That is, users will feel that they are visiting a space in which certain types of behaviour are acceptable and others not. For example, if someone feels that they are, in some sense, in a foyer within Nottingham University, then they are less likely to treat the experience as a fiction or a game, and act accordingly.

Locales

The social nature of these shared worlds also affects the way they are constructed spatially; indeed, the value of the spatial metaphor becomes more apparent. One common approach is to divide the space into areas, usually called locales, that allow for the subdivision of activity. Even non-graphical virtual worlds, such as the text-based MOOs, use this metaphor most successfully. Giddens (1984) offers us a means of understanding this with his interpretation of locales - areas that set the scene for certain actions. Within locales, the affordances of the environment, together with the nature of the tools provided, predisposes each locale to particular types of work and interaction, by analogy with the way in which in the real world we have, for example, offices, coffee rooms, and foyers. The spatial relation of such a foyer with other spaces is important: it acts as an entry point to other more private spaces.

A number of collaborative tools have drawn on this idea, such as the Worlds project (Fitzpatrick, Tolone & Kaplan 1995).

Contents

Graphics Multimedia Virtual Environments Visualisation Contents