Virtual Reality in Teaching Architecture.
Dr. Vassilis Bourdakis
Keywords architecture, teaching, software outline.
Aim
To help architects appreciate and utilise Computer Aided Architectural Design (CAAD) as
an everyday tool for their design work and not only for presentation purposes. In order to
achieve this aim, software designers have to be familiar with the particularities of
architectural design.
Background
It has been acknowledged that architects (both students and practitioners) use
computers as a drafting and presentation tool rather than for design. This is mainly due
to the structure of the existing software and until recently to hardware limitations. The
software available is, in its majority, designed, developed and marketed by computer
scientists engineers and as such does not adhere/follow the way architects think, design,
draw. Hardware limitations are mainly in processing power (rendering speed) and in video
graphics technology.
Furthermore, the CAAD courses available in the UK and abroad are focused on providing
the general computer software operation skills and as such are not addressing the actual
problem.
Work carried out so far
The researcher has worked extensively with AutoCAD and 3Dstudio in the PC platform and
SG as a practising architect for the last six years and more recently in teaching. Early
in the process it was necessary to develop various tools aiding in the creation of 3D
models and simple volumetrics; a quick way to produce a rough working model.
However visualisation of such models is always hindered by the need to set up
perspective views (since all modelling is done in wireframe mode) and following render
them. This way, potential problems are hidden, since it is not feasible to render a model
from all possible views. Needless to say that the views the designer selects are the ones
he has most thought off and as such are the least likely to create problems.
Real time rendered viewing help in quickly visualising designs, understanding how the
elements and components are put together as well as their relationships in 3D space and
realising errors in the earlier rather than the later stages of design. Such facilities
are provided in a few high-end workstation platforms (Silicon Graphics being the main
one). However what needs addressing is ways of editing architectural models interactively
in full rendered views (preferably on personal computers). Furthermore such software
should be priced within the reach of the users; solutions that cost £20,000 as is common
in the workstation market are not going to be of much help, especially if we want to use
them in HE.
An application that would be very beneficial to architectural teaching is a modeller
both accurate and flexible enough in modifying geometry interactively as well as
supporting Boolean operations for the creation of the complicated volumes architects are
using. This is where the existing software is failing; very often real time editing on
fully rendered perspective views is done using X, Y and Z transformation sliders that are
both un-intuitive and in certain cases impossible to use.
Some important points
The following points are identified, they do range from practical implementations to
theoretical / conceptual issues:
- Need for the establishment of new drawing database standards that will be embraced by
software engineers and used extensively (like the Data Exchange Format developed by
AutoDESK a decade ago and is now a standard for 2D and 3D drawing files). Inventor and
VRML (Virtual Reality Modelling Language) are two standards that seem to be suitable and
are gaining support and recognition world-wide. They are extensible, open, supporting
various data structures and behaviour engines, object animation and even going further to
multi-user participation (VRML+).
- Slow down (postpone) the transition to immersive VR. It limits the applicability of VR
in architecture. Universities currently struggle to provide the students with labs
suitably equipped for CAAD. They cannot possibly go immersive anytime soon especially
since there are other important issues that have to be addressed first. I believe we
should be aiming at s/w capable of running on Pentium and Power Macs as far as
architectural design is concenred. Software engineers are extremely "inventive"
in finding ways to "utilize" the power of even the fastest, newer computer;
voice recognition and immersive VR systems are two such ways.
- Input of information is another main problem of current CAAD software. Dimensions /
coordinates have often to be typed. Examining VR systems and fully interactive software on
workstations, it is apparent that movement in 3D, orientation and input is a problem that
is currently not seen much attention (having smoother movement, better lighting algorithms
is more appealing to software engineers but not to the users in the long run). This is
understandable, especially considering the research that has to be put in developing
better input systems. The 2D screens used currently and the 2D input devices (mouse,
keyboard) are responsible to a small extend. Even a 3D mouse cannot be of much use on a 2D
display. 3D glasses are a step forward, but still the software has to be redesigned.
- I'm currently working in prototyping techniques that could be used in teaching
undergraduates Computer Aided Architectural Design. So I'm testing various systems myself
on architectural competition entries and other design exercises. The current trend of
teaching how to use a particular package and assessing the students on their ability to
recreate an old project in a 3D computer model is only the starting point of CAAD
education. We have to teach students how to use a computer to design their next project.
- What is currently missing is interactivity and software that will have both the accuracy
/ detailing of current CAD programs and the VR capabilities as found currently on VRML on
the Web.
- A VRML version of an example demonstrating a few of the problems faced in architectural
design can be found in my homepage (http://fos.bath.ac.uk/VRML/).
It is an ideas competition for a membrane covered play ground. Placing a sloping
triangular pillar in relation to an orthogonal volume of a restaurant and coffee-shop
supported by a grid of conventional beams and columns highlights the unsuitability of
existing s/w tools. I usually do all my design work in a PC using AutoCAD but for this
particular problem I decided to "try" the interactive tools supplied with a
£10,000 workstation. However the user-interface of the supplied software was only
suitable for the placement of simple 3D objects within a "visually appealing"
relation to each other. Accuracy is non existent, ease of use debatable but the user works
on fully shaded views and the objects can be spanned in rendered mode easing the
identification of design problems. After a full days struggle I had to get a calculator,
do some trigonometry and figure out where the objects should be placed. Of course I had to
go back to AutoCAD for the accurate placement of these elements in the site. Then I had to
translate the model to the workstation's 3D format and check that everything was
acceptable. Hardly a reasonable and efficient design approach.
- I should point out that I'm not discussing issues like interactive shadow casting and
lighting design which are problems that should be considered. Nor talking about
photorealism and texture mapping which is always the final goal of such products.
- Architects themselves have to learn how to work with computers, appreciate the
limitations of the existing systems and realise the potential of their use.
- Two new notions introduced by VR that have to be utilized in order to achieve speed and
interactivity in modelling are:
- the need for hierarchical structure of the models, which is quite different to
the material or colour base structure used by current CADD software
- the notion of Level of Detail which is of vital importance if the aim is
reasonable speed or interaction.
- Concluding, the importance of VR in architectural teaching is threefold:
- Understanding spaces whilst in the design process
- Learning history and visualising buildings being taught (i.e. Bath and Edinburgh models)
- Database mapping (example of London Map of the Future on the web, accessible from http://www.bath.ac.uk/Centres/CASA/london/)
It is important to realise that I'm not developing s/w (although I have done a lot of
scripting / programming in the past). I'm just identifying problems on the existing s/w
available and propose new approaches on CAAD software design. However, it seems that I
will soon have to work closely with software developers as it seems that no one is really
providing software customised for architectural design. The work carried out in CASA,
University of Bath has now reached a point that we cannot progress any further without the
help of specialists.
Software developers are focusing on the construction industry looking from the project
management, drafting, surveying point of view (which is important but not the only part of
the building process that computers can/should be used and definitely not the approach
that computer education in schools of architecture should adopt).