YOU enter the theatre and take your seat in a plush executive chair. The lights dim, and on the screen, filling your field of view, you see the San Diego Convention Center in California, decked out for the Republican convention of 1996. Using fingertip controls in the arm of your seat, you explore the foyer, then move through the corridors and into the meeting halls of the vast building. The image wraps around you, drawing you in. Hi-fi rock music pounds out from every corner, and as you drive yourself forward towards the main arena you can almost hear the future crowds roar to greet the presidential candidate.
Welcome to the Reality Centre, a simulator built by Silicon Graphics the Californian graphics specialist at its British offices near Reading. According to David Hughes, who dreamed up the idea of the centre and now manages it, early simulations of the San Diego Convention Centre revealed that the auditorium鈥檚 ceiling was too low, while later ones enabled contractors to plan the building of the structure.
Goodbye to glue
Advertisement
From convention centres to city centres, computer visualisations such as this are becoming essential tools for architects and developers. The University of California at Los Angeles and local planners have used computer simulations to tell residents about their plans for rebuilding parts of the city devastated by last year鈥檚 earthquake. Officials in Kobe, Japan, are interested in using similar techniques to help reconstruct their own city. And architects planning redevelopment schemes in Berlin and Newcastle upon Tyne are using computer visualisations to show off their plans.
So can architects, planners and engineers throw away the glue, plywood and plastic trees they have traditionally used to build their models? Bill Jepson of UCLA, who is an active participant in the scheme to rebuild the city, says computer models are reaching the point where they can provide most of the information that construction professionals and their clients will need. 鈥淭he quality of the imagery is incredibly good,鈥� he says. 鈥淎s soon as people see the computer generated model their disbelief is eliminated on the spot. They know their neighbourhood and can look at what is being proposed.
Over the past couple of years there has been a dramatic improvement in the quality of the imagery available from computer simulations. No longer need they be limited to flat blocks of colour that roughly suggest the shapes of buildings. Instead Jepson and his colleagues take photographs of building facades, trees and other local landmarks near to the area they are redeveloping and use them in the computer model. Not only are photographic images more realistic, they also require half the processing power of images created from scratch, Jepson says. So where there used to be a cartoon-like animation, what you now see on screen is something like a live television picture. This has been made possible by technology developed for the military that is now available for civil uses as part of the Cold War payoff. At the same time, there has been a dramatic drop in the price of the hardware needed to run complex image processing algorithms and store large graphical databases.
Trees, please
To create his images, Jepson uses a computer system called the Onyx Reality Engine from Silicon Graphics. The Onyx system uses 24 processors running in parallel. Each one is connected by a fast link to a specialised graphics processing chip, which does the calculations needed to provide shading and shadows, and to display the image on screen. This arrangement allows the task of creating the polygons that make up the picture to be spread across all the main processors.
Jepson says that with this kind of computer system it is possible to change images while a presentation is in progress. So, for example, if one of the residents wants to see what a new car park or shopping centre will look like from their front door, the image can be presented from that viewpoint. If they then ask to see the same view as it will look in ten years鈥� time, when saplings planted by the developers have grown into mature trees, the image can be changed as they watch. 鈥淵ou can play all sorts of 鈥榳hat-if?鈥� scenarios,鈥� says Jepson. It is also possible to link the designer鈥檚 model with other data, so as to assess the impact of a new sports stadium on traffic congestion, for example, or retrieve information such as the percentage of people in a particular neighbourhood who use credit cards to buy cinema tickets.
People like the idea of seeing what architects have planned for their street, Jepson says. At present this can only happen at special showings, where residents can view the computer generated images on a large screen. In the future, Los Angeles hopes to channel computer images straight into people鈥檚 homes, using high-speed data links such as optical fibre.
This is a far cry from the days when planners shied away from showing the public their plans, because they thought their aerial views, models and drawings would not be properly understood by ordinary people. But it also puts city planners on the spot. If people have been given a realistic preview of a project, they feel let down if the end result turns out worse than it looked on screen.
So is this the end for urban planning disasters, dream homes that are a nightmare to live in, and factory buildings with doors that turn out to be too small? Not quite, say some in the building profession.
Stephen Perrella, an architect and publications editor at Columbia University鈥檚 graduate school of architecture in New York fears that the cost and novelty of the computer models may make them appear unduly important and draw attention away from the buildings they are meant to represent.
Perhaps what is needed is a stronger sense of presence in the models 鈥� perhaps a simulation that allows the viewers to trip over empty drinks cans as they stumble around a corner late at night. The imagery used in projects such as the rebuilding of Los Angeles, or the design of the San Diego conference centre is certainly realistic. And if displayed in a darkened auditorium surrounded by a floor-to-ceiling screen, some scenes can have you grasping for the arm of your chair. But critics argue that this is merely the superficial realism of an arcade game. That is why some researchers are working to build architectural models that use immersive virtual reality, complete with headsets and pressure-sensitive gloves. Mel Slater, Anthony Steed and Martin Usoh are researchers in the department of computer science and the London Parallel Applications Centre at Queen Mary and Westfield College, University of London. They believe that the credibility of virtual worlds is affected not just by the crispness and detail of their images but also by factors such as the range of ways the viewer can interact with the virtual model 鈥� things like moving around rooms and looking out of windows.
The QMWC team has developed a system of worlds within worlds for viewers navigating their virtual reality models. Viewers begin by putting on a virtual reality headset, which places them at the top level of the model 鈥� the interior of a proposed conference suite, for example. As well as the usual items you would find in a meeting room, on one of the desks they will see an image of another virtual reality headset. By making the movements as if to pick up the headset and put it on their head, the viewer enters the next level of the virtual world. The QMWC researchers suggest that this second level could be a control room, where the architectural plans of the model the viewer has just 鈥渨alked through鈥� are on show. The viewer might be able to amend these plans to adjust the lighting levels, or the position of a doorway, for example. By taking off the virtual headset, the viewer could then return to the higher level of the model to experience the changes they have just made.
For their virtual reality model, the QMWC researchers use a ProVision100 system from Division, a specialist in virtual reality hardware. This computer uses over 30 microprocessors operating in parallel to calculate and render the digital models which are displayed in the virtual reality helmet. A system suitable for a single user costs around 拢40 000.
So some architects are on the lookout for less costly techniques, which may be just as effective for their purposes. The British architect Patrick Barnes is preparing designs for the second round of a competition to redevelop the centre of Berlin, and to show them off he has created animated walkthroughs. These use 3D computer models generated on computer, which are stored on video. Barnes uses a software package called Silverscreen, with additional features written by Leonardo, a company based in Reading, Berkshire. This creates 3D models that include details right down to the dimensions of nuts and bolts.
With his models completed, Barnes sits down with Leonardo鈥檚 computer expert Adrian Narracott and plots a walkthrough of his vision of the new Berlin centre. Initially, Barnes and Narracott view the 鈥榳ireframe鈥� representation of the designs. But once they have selected the route, Narracott sets the computer to work to render the chosen series of images. When played in sequence, they will take the viewer on an animated ride through Barnes鈥檚 buildings. The end result is not interactive in the way that virtual reality representations are. But Barnes鈥檚 3D visualisations can be used to produce working drawings.
Digital mission
Other developers believe that computer visualisations can make do with even less realism. The Tyne and Wear Development Corporation鈥檚 拢69 million Quayside project in Newcastle will rebuild 25 acres of the city鈥檚 waterfront. The site includes two listed buildings as well as the Tyne Bridge, the city鈥檚 most famous landmark.
Real Time Design, a local firm that specialises in computer-based design and visualisation, has built a virtual model of the Quayside project. By keeping detail down to a minimum and making non-essential features such as trees two-dimensional rather than solid, the model has been made compact enough to run on a standard PC. But despite its relative simplicity, the model has had a useful role to play. Early in the design stage, simple block shapes with the proportions of the finished buildings were put into the model so that the impact of the buildings on the vista of the bridges and river could be assessed. This allowed the designers to go on to more detailed work, knowing that the position and shape of the buildings would not have to change.
This is a far cry from the complex interactive model used by Jepson in Los Angeles. But, perhaps the significance of his hyperreal images lies elsewhere. When you can experience a virtual city from the comfort of an executive chair, who will want to venture into the real world with all its imperfections?