Return to GeoComputation 99 Index

3-D city: prototyping techniques for urban design modelling

Weiso Chen
Land Use Science Group, Macaulay Land Use Research Institute, Aberdeen, AB15 8HQ, United Kingdom


The practice of urban design is a combination of art and science. It is quite difficult to foresee what effects new design ordinance imposed by the government would have on the local cityscape in the long run. It has been changed, however, due to current development of computer technology. New computer tools enable us to "visualise" what the effects might be. We present in this paper an outline of efforts aiming at using the state-of-the-art 3-D real-time modelling techniques with an option of integrating 3-D GIS to couple them together in the context of urban design. We conclude this presentation by identifying areas that can benefit from this new technology along with remarks.

1. Introduction

Visualisation, as its name suggests, conveys information directly perceptible to human brains so that detailed spatial relationships are revealed in a way that conventional planar presentations are not capable of. The term "visualisation" has been used extensively in many fields: from scientific and engineering visualisation to entertainment industry, from physics to molecular behaviour, to medical visualisation. Now it has found its way to the fields of urban design, urban planning, and landscape architecture as well. More often, however, the term is used in a static sense. That is, one uses this term to express how they perceive things in a visual manner in forms such as graphics and charts, rather than in tabular formats. In the design and geography communities, visualisation often refers to the static techniques, such as perspective drawings, photomontage, thematic mapping and graphics, though dynamic sense of visualisation is increasingly gaining attention (e.g., Sheppard, 1989; Liggett and Jepson, 1995a, 1995b). For example, Chen (1991) presented how CAD-based visual simulation techniques (AutoCAD and Truevision) can be applied to the domain of urban design, based on either existing or hypothesis design problems, to some extent to assist visual impact analysis (see Section 2 for more discussions on VIA). Similarly, Levy (1995) used ArchiCAD to visualise a water front design alternative. More examples can be found in two special issues of "Environment and Planning B: Planning and Design" (vol. 19, 1992 and vol. 22, 1995). The Cartogram promoted by Dorling (1991) is another example of how visualisation is perceived by some researchers in the urban and geography communities. In contrast to Chen (1991), Reuter (1997) presented how non-CAD techniques can be used to realise the goal of visualisation of designs, though, again, his methods produce mainly planar (2-D) and perspective views (2.5-D). Since the term visualisation per se is so attractive, the market is full of commercial software that claims the final designs can fulfil the desires of visualisation. Some of these software such as Softimage, ArchiCAD, Extreme 3-D, Model 3-D, etc., are capable of producing realistic images, though static. Sheppard (1989) and Smardon et al., (1986) demonstrated extensively how static visualisation techniques, such as photo simulation, photomontage and photo retouching, can assist designers and architects in their work.

In the computer graphics community, visualisation often refers to vivid images created and rendered using advanced techniques such as ray tracing, which can simulate the shading effects of a light source imposed on an object. The images created are static, though in many instances they can be very realistic (photo-realistic) as their real world counterparts.

The use and meanings of visualisation, e.g., static versus dynamic, are so concerned that many other researchers have had extensive discussions on this topic (e.g., several articles in the book edited by Hearnshaw and Unwin, 1994).

In this paper, however, we emphasise the dynamic sense of visualisation. By dynamic, we mean that once a 3-D geometric model of an urban scene is constructed, the user can enter the scene to experience and manipulate the environments. The experience and views the user will gain is dynamic as the user enters different parts of the model, as it is in the real world. In this paper, "dynamic," "real-time," and "interactive" are the key words, as important as visualisation itself.

The focus of this paper is to introduce modern technology that would allow one to construct a data model with which the model will be rendered dynamically as one experiences different portions of the model at a reasonable frame refresh rate. The techniques used will be very different from their static counterparts, though some of the techniques used in static visualisation, such as ray tracing and radiosity, would remain useful in a dynamic visualisation experiment. The focus of dynamics is important--an important part of designers' work is to communicate their ideas with their clients. To make the communication effective, sketch drawings alone as they are used today in most cases are not sufficient. To convey enough information for the clients to understand what a potential design will look and feel like, a rich and dynamic presentation of the ideas is indispensable.

We also would like to differentiate "urban design modelling" from "urban modelling." On the one hand, complex urban modelling often refers to the efforts of modelling urban forms and functions such as transportation and land use through economic and mathematical equations. On the other hand, however, our main interests in this paper are to model a city in a visual manner, without considering the underlying economic processes. That is, urban design modelling from an urban designer's perspective, rather than urban modelling from a regional scientist's, though the integration of both would be desirable.

The structure of this paper is organised as follows. We begin by clarifying the term visualisation. Section 2 examines why this modern technology is important in urban design and how designers can use this technique to fulfill their work requirements with examples in visual impact analysis. Section 3 focuses on how this technique can be applied to urban design modelling in details. Section 4 introduces Virtual Reality. Challenges in designing and constructing the databases are discussed in Section 5. We conclude this presentation by identifying areas that can benefit from this new technology along with remarks.

2. Visual impact analysis and visual simulation

Before we enter further discussions of real-time visualisation, we would like to point out some existing applications (though many of them may not be as dynamic as desired) already used in designers, planners and architects' practices in visual impact analysis (VIA) and visual simulation, which we believe are the closest connections between computer graphics and urban design. Visual impact analysis is often required to consider the (visual) impacts on surrounding environments posed by a proposed construction project before constructions can take place. Of various components required by VIA, visibility analysis is of particular importance. Overhead electrical transmission lines and the visual impacts of pylons on rural landscape are popular examples (e.g., Gross and Koglin, 1990; Whitelegg et al., 1981, etc.). This is an important application area because many countries now require environmental impact analysis for many large-scale construction projects. Visual impact analysis also is of particular interest to the public, for environmental amenity is of great value to their properties and community. Some well-illustrated books on this subject were given by McAulay (1988), Purdie (1983), Sheppard (1989), Evans (1984), and Smardon et al., (1986); however, as demonstrated by these books and research papers, the techniques used to analyse the visual impacts on surrounding environments are predominantly static and manual in nature. These commonly used techniques include photo simulation and retouching, though computerised techniques such as photomontage have been developed as well (e.g., Nakamae et al, 1986).

The following will describe some techniques used in VIA and examine how evolved dynamic techniques can be used to enhance the capability of VIA. Turnbull and Gourlay (1987) described a system named CAVIA, which comprises of six programs. Their system can be considered as having landscape data, object data, visibility determination and visualisation subsystems. This system is capable of integrating both digital terrain models (DTM) and man-made objects such as buildings and electricity pylons, and nature objects such as trees and shrubs, etc., and apply suitable visibility algorithms to produce an accurate perspective drawing of the visible objects in the landscape from a given viewpoint. This is important in demonstrating the visibility of the objects and their scale in the landscape. 3DOG, presented by Gross (1991), described another approach to determine the optical impression of objects on the landscape combining the standard methods of computer graphics, image analysis, physiological aspects of the human visual system, and physical conditions of the environment with example of overhead lines.

Moreover, Kaneda et al., (1989) gave an example of how terrain visualisation in 3-D can be used to assist environmental impact assessment. Chen (1991) and Tan (1997) demonstrated how visual simulation techniques could be applied to the domains of urban and natural environments, and to assist visual impact analysis. The ABACUS project created about three decades ago and still in operation is another example. Although this project was initially set up to promote the effective use of computers in architecture and building design (Purdie, 1979), an important aspect of its developments lies in its development of visualisation techniques such as photomontage, visibility analysis and walkthrough (Maver, 1975; Maver, 1987; Herbert, 1987).

In addition to visual impact analysis for urban and rural environments, the University of Washington has focused on forest areas and has produced a system named "Vantage Point" (Ulbricht, 1994; Bergen et al, 1995; Fridley et al, 1991; Bergen, 1993). What makes this project stands out is that it is a real-time visualisation software package to assist forest managers as well as the public to forecast the possible impacts of forestry harvest on neighbouring environments. This system consists of two core programs: one titled Vantage Point for maintaining, manipulating, and visualising landscape spatial and temporal data; and the other titled Tree Designer for creating the trees to be displayed in Vantage Point visualisations.

Objects such as trees are interesting, as they are vital components not only in forest environment visualisation but also in an urban setting. Modelling trees, however, is no trivial experiment. Traditional ways of creating each geometrical model consisting of hundreds or more polygons for every tree is not cost effective in terms of the processing overhead and their "contributions" to final images. Methods such as L-grammar, in conjunction with texture mapping, have been developed to facilitate tree modelling and the like where geometry of objects can be replaced by detailed textures without bringing down system performance due to hardware overhead. Other natural phenomena, such as haze and fog, among other atmosphere factors, and weather conditions providing important perception clues such as light absorption and depth cues to human vision, are difficult to model geometrically. How to model these natural phenomena has presented great challenges to the computer graphics community. Fractal geometry, among other methods, such as wavelets, can be used to efficiently model these phenomena and objects such as mountains to produce realistic images with ease.

3. Urban simulation and visualisation: the quest for dynamic visual realism

As shown previously, VIA techniques are predominantly based on both 2-D plane and 2.5-D in the forms of contour maps, perspective drawings or wire-frame. Although to some extent, these techniques can show the results of visibility analysis, they are not easily understood, due to the nature of this presentation, as human brains are mostly 3-D vision-driven. In this section, we will show how new techniques can be applied to real-time interactive visualisation while preserving as much realism as in static simulation as possible.

The creation of realistic pictures is an important goal in fields such as simulation, design, entertainment, research and education, etc. Simulation systems present images that not only are realistic, but also change dynamically. For example, a flight simulator shows the view that would be seen from the cockpit of a moving plane. To produce the effect of motion, the system generates and displays a new, slightly different view many times a second. Designers of 3-D objects such as cars, airplanes and buildings want to see how their preliminary designs look. Creating realistic computer-generated images is often an easier, cheaper, and more effective way to see preliminary results than is building models and prototypes, and also allows more alternative designs to be considered in a given time period. Computer-generated imagery also is used extensively in the entertainment and advertising industry. The most recent film totally based on computer graphics was "ToyStory," produced by Pixar for Disney in 1996. Realistic images are becoming an essential tool in research and education, too. An active area is the use of graphics in molecular modelling, where it is impossible to see their relationships physically.

Visual realism in the fields of urban design and environmental simulation is desirable. Janssens and Kuller (1986), for example, introduced the LUND simulator, a system that can be used to simulate and evaluate a planned physical environment. The representation of the simulation is on model filming, that is, filming from an eye level, which is correct with regard to the scale of the 3-D physically constructed model. This system is totally based on the physically constructed model and the movement of the camera to create an animated film that can later be played back to determine the effects of proposed objects on neighbouring environments. Although the recorded film can be very realistic, which is, depending upon the underlying model, the pitfalls of such an approach, can be identified as follows. First, since it is a physically based model, the model itself cannot be easily modified. Once the model is created, any changes to the proposed objects, such as the shapes and textures of the objects, or its location changes, will result in the recreation of the model and re-filming to reflect the changes. Second, since the camera movement and the recording of the film are pre-determined, viewers are left passively to watch the film as they watch ordinary TV with no control over the contents. They cannot change their viewpoints, for example, to see other parts of the model. This technique could be misused to avoid some areas from not to be filmed, where it may have severe impacts on the overall environments. Finally, since the filming was created indoors, lighting and other atmosphere factors are not included, thus, the resulting film may not be as vivid as desirable. Bosselmann (1998) examined the role of simulation technologies in urban planning with case studies undertaken at Berkeley. These case studies demonstrate the use of computer modelling to predict the effects of development on a host of environmental and experimental factors that affect the users of urban spaces (McNamara, 1998).

In addition, Schmitt (1996) argued that the development of computer graphics has transformed our ability to experience and to design urban space. In particular, computer graphics simulations of urban space, as he views it, have found three major applications. One of them is the reconstruction of the past, ancient cities are re-created as realistically as possible. The second application is the simulation of projects before they are actually built in order to optimise the design of new physical urban centres. The third application is the creation of a new kind of reality and urban space in the rapidly evolving information territory. These observations have been echoed by other researchers as well (e.g., Mitchell, 1995). The new urban space on the information superhighway is growing exponentially and can be experienced over the Internet. Some examples may include Superscape's Virtual World, Virtual Berlin by Art+Com, and the Virtual London project undergoing at University College of London.

These developments and examples have suggested computer generated visual photo-realism in urban simulation, and visualisation is now not only feasible but also is a trend of progress. Many of these examples are presented in the form of virtual reality. It is noteworthy mentioning that many online virtual worlds, such as those presented in Virtual WWW and Virtual Berlin, Bath Model (Bourdakis, 1998), among others, are created with virtual reality modelling language (VRML). Although VRML is not the only language that can be used to create virtual worlds, it is, however, at the time of writing and foreseeable future, the only "de facto" language to create and exchange 3-D models. Section 4 briefly introduces virtual reality and VRML.

4. Virtual reality

Virtual reality (VR), sometimes named virtual environments (VEs) or virtual worlds, can trace its root back to more than 3 decades ago. It is often recognised as contributed by Sutherland in his seminar work of the SketchPad system developed at MIT (1965), and the ground-breaking paper, "The Ultimate Display." The term "virtual reality" itself is a term coined by Jaron Lanier, founder of VPL Research of Redwood City (a former maker of data gloves, head-mounted displays and other devices for VR), according to Sun Microsystems Computer Corp. (Sun, 1994).

What is VR? This is not a trivial question to answer. A brief introduction of VR can be found in (Isdale, 1998). Mazuryk and Gervautz (1996) also have conducted a thorough review of the history, applications, technology and the future of VR. In particular, an in-depth book on VR aiming specifically at architecture and design professionals is provided by Bertol and Foell (1997).

VR can be seen as an interface technology between computers and humans, that is, a human-computer interface (HCI). As stated by Isdale, VR can be defined as "a way for humans to visualise, manipulate and interact with computers and extremely complex data" (Isdale, pp. 2, 1998). This is reflected and realised in data visualisation such as data mining and scientific visualisation (ViSc). In addition to serving as a HCI, a recently developed idea is to treat VR as an operating system for running multiple applications rather than as a means of building stand-alone demonstrations (Pettifer, 1996).

4.1 VR models on the internet (VRML)

VR models are typically created, executed, and experienced on a local standalone system or over a high-speed internal network; however, the proliferation and exploration of the Internet has led to the creation of VRML, which allows a virtual world to be accessed online from around the world. VRML, still an emerging technology and originally based on Open Inventor from Silicon Graphics, Inc., is an industry standard. Although the language is based on SGI's Open Inventor, it is defined and governed by the independent VRML architectural board (VRML, 1998; Vacca, 1998). (Note: VRML at its current state, however, is not sufficient and efficient enough to convey enough information as is expected over the Internet. Many have claimed that due to the design flaws in VRML, new protocols, which are more efficient and better suited to today's online demands, are needed. Those proposed languages include SGI's EMMA (Marrin, 1998), Java 3-D, Microsoft's Chrome, and SVR/VRT from Superscape. Nevertheless, those new protocols are still in their early infancy and, more importantly, they have not been widely accepted as VRML.).

VRML provides a common platform for different models created under different systems to be exchanged, and more importantly, to be experienced by more interested users otherwise impossible over the Internet. One significant factor that affects every Internet user is that the connection and file transfer speed is usually very slow, compared to local disk I/O access time. File sizes must be reduced significantly so that a user will not have to wait too long for files to download to his/her local machines (e.g., by using instancing instead of repetitive data info., lower texture resolution, etc.). In addition, although the VRML models will be executed locally using local processing power, end users usually only have low-end PCs that are limited in their speed and their graphic display capability. Too many level of details information, for example, may easily overwhelm end users' machines to cope with. Kofler (1996) has addressed these issues and shown how he resolved some of the difficulties with the example of a building walkthrough, modified with Berkeley's proprietary UniGrafix tools. Crispen (1998) offers general guidelines on how to fine-tune a VRML model from a practitioner's standpoint. He also maintains a FAQ of VRML online at

5. Challenges ahead

Before realising the ideas, a working model has to be created. We will review and pinpoint some issues and describe our approaches to them. Issues such as database structures and data capture will be presented.

5.1 Spatial data model

As previously stated in Section 3, synthesis urban modelling has become a reality recently due to advanced development in computer technology. Many efforts have been made to attempt to modell physical urban scenes in computerised settings. For various applications, 3-D models that can accommodate real-time interactions give rise a great challenge to the current database technology, e.g., relational databases. At the heart of real-time modelling, however, is the underlying databases that hold entities of models and geometric relationships among them. Surprisingly, not all interactive visualisation experiments have adopted a database approach. This is largely due to the real-time requirements not met by conventional relational database design. Instead, direct file access has been a favourite for many modellers (Funkhouser, 1993).

Many issues of designing virtual environment databases have been discussed by Deyo and Isaacson (1995), in which they identified the most difficult task facing a database designer--finding the right entity abstractions or models, and finding an overall database structure that will support multiple entity models and their relations--the authors could not agree more. Given existing spatial database structures and designs such as quadtree, BSP tree, as identified by Samet (1990), it is no easy task to decide which will best suite a particular area of application. Yet, while these regular space decomposition schemes have been widely applied to various model buildings, whether they will be best suited to urban applications remains to be challenged. Sillion et al (1997), for example, have argued that, given the unique properties of urban environments, e.g., huge complexity of the geometrical data and widely varying visibility conditions, and associated real-time visualisation challenges, a new database framework for such urban scene tasks is needed. Their central concept is that of a dynamic segmentation of the dataset, into a local 3-D model. They also introduce imposters to represent distant scenery. Kofler (1998) has also suggested a new LOD-R-tree scheme to address this issue. Koutamanis et al., (1995) presented an interesting book on how visual databases can be used to assist urban and architectural designs; however, their usefulness remains to be verified.

5.2 Model creations

Currently, the creation of databases is painstaking, labour demanding, time-consuming, and error-prone. It is to a very large extent still a manual procedure; therefore, it is one of our desires to suggest a procedure with which a semi-automatic, or more desirable, a fully automatic model creation procedure can be realised. Furthermore, many design projects are no more than computer graphics. We want to go a step further by linking with 3-D GIS (Raper, 1989). It is fair to say that the line between computer graphics and 3-D GIS has been a blur; however, one important distinction between the two is the adoption of direct file access (e.g., scene graph, display list) versus a spatial database. A fairly thorough discussion on these issues can be found in (Kofler, 1998). We have identified a project conducted by the Urban Simulator Team at UCLA (Liggett et al., 1998), in which a project, a link between the computer graphics model and 2-D GIS database of the model (ArcView), has been realised through the use of RPC. This can only be viewed as an intermittent solution, as a more desirable one would be to include the 3-D GIS capability inherently within the models rather than by linking to an external database.

Many building walkthrough models involve converting existing CAD models into proprietary formats for interactive visualisation, while others involve creating proprietary model formats from the ground up. Bignone et al (1995) and Graf (1995) presented a rather straightforward way to generate building models that are mainly axis-aligned, or rectilinear, to extract the shapes of roofs from airborne images, and then measure the heights of the buildings from DEM. Given this information, simple building models can be constructed. More elaborate methods have been proposed from the fields of architectural photogrammetry and computer vision to accommodate a wide variety of roof shapes and building models, e.g., Collins et al., (1995, 1996). Two Acosna workshops have reported a vast amount of work on this, too (Gruen et al, 1995). The Amobe project (Henricsson and Gruen, 1995a; Henricsson et al, 1995b; Henricsson et al, 1996) generates all aspects of buildings in a complete CAD format, including 2-D contours, 3-D segments, 3-D planes and topological relations. In addition to acquiring

3-D building models from aerial images as in Ascender and Amobe, Streilein (1996) proposed to reconstruct 3-D building models from terrestrial images, which are taken at street level, for visualisation and animation. Wang and Hanson (1992) proposed another refinement system that, given a known initial site model, could extract microstructures of buildings such as doors and windows from aerial images to greatly improve the look and feel of the model displays.

What about a large scale urban area or landscape visualisation? Unlike building reconstruction, the creation of urban models remains a great challenge, due to its unique property and complexity. We have identified some attempts, though one approach varies from another. The Urban Simulator Team at UCLA captured the street scene by manually using a camcorder, then a great deal of time has to be spent on post-processing those video images. In contrary, the MIT City Scanning project (Teller et al., 1998) has attempted to fully automate the manual procedure. Kang et al., (1995) reported an algorithm for extracting concise surface models of a scene from real 3-D sensor data that can be used for virtual reality modelling of the world. An interesting project has been reported by Quarendon (1993) aiming at creating 3-D models from existing 2-D images. How useful these approaches are requires further investigation.

6. Conclusion

We have reviewed the current technologies on real-time 3-D interactive visualisation. We have shown the progress of computer use on urban design with the particular application of visual impact analysis, and indicated how a possible link to 3-D GIS can be established, though many challenges still lie ahead. In this section we would like to point out areas of research that may benefit from this progress. Participatory planning (Sackman, 1972) has been an idea to create an environment in which planners and the public can work together. Similar efforts such as planning support systems (Landis and Zhang, 1998; Shiffer, 1995) and computer supported co-operative work (CSWC) (Rinner and Schmidt, 1998) aim to facilitate the decision making process from different directions; however, this cannot been easily achieved because of several reasons. This new technology to incorporate into a VR-based design environment can help break the barriers that lie between enabling both parties to visualise in an intuitive way what the planners have to propose and what the future effects may look like, thus, improving the decision making process. To further facilitate the communication between these two parties, an easy-to-use 3-D editing tool would be desirable.

Landscape preference modelling is another possible area that can benefit from this technology. The modelling technique has used manual or computer-generated images as base maps along with questionnaires to measure one's preferences over a set of landscape. Mainly because of the static nature of presentation and relatively small areas shown on these images, the extent of validity remains questionable (e.g., Bergen et al., 1995; Mugica and De Lucio, 1996). These pitfalls can be overcome by using 3-D and dynamic presentation of scenarios in which users can experience the environment more directly; thus, a more accurate picture of the users' preferences could be captured.


Bergen, S., 1993. Mitigating Potential Impact to Visual Quality During the Design of Forest Operational Plans. MSc Thesis. University of Washington.

Bergen, S. et al., 1995. The Validity of Computer Generated Graphic Images of Forest Landscapes, Journal of Environmental Psychology, Vol. 15: pp. 135-146.

Bertol, D. and D. Foell, Eds., 1997. Designing Digital Space: An Architect's Guide to Virtual Reality, John Wiley & Sons, New York, NY.

Bignone, F. et al., 1995. Automatic Extraction of Generic House Roofs from High Resolution Aerial Imagery. Technical Report. Communication Technology Lab, ETH, Zurich, Switzerland.

Bosselmann, P., 1998. Representation of Places: Reality and Realism in City Design. University of California Press. ISBN 0-520-20658-4.

Bourdakis, V., 1998. Virtual Reality: A Communication Tool for Urban Planning. Internal Report, Centre for Advanced Studies in Architecture, University of Bath, U.K.

Chen, S., 1991. Computer-aided Visual Simulation in an Urban Context. MSc Thesis. College of Environmental Science and Forestry, State University of New York, NY.

Collins, R. and C. Jaynes, et al., 1995. A system for automated site model acquisition. Technical Report, Dept. of Computer Science, University of Massachusetts, Amherst, MA. Also in Integrating Photogrammetric Techniques with Scene Analysis and Machine Vision II, SPIE, Vol. 7617, 1995, pp. 244-254, Orlando, FL.

Collins, R. et al., 1996. UMass Progress in 3-D Building Model Acquisition. Technical Report. Dept. of Computer Science, Lederle Graduate Research Center, Univ. of Massachusetts, Amherst, MA.

Crispen, B., 1998. VRML Post-Production: The secret of the best VRML worlds on the Web, InterActivity magazine, July 1998, pp. 20-21.

Deyo, R. and P. Isaacson, 1995. Problems in Designing Large-Scale Virtual Environment Databases. Technical Report. Evans and Sutherland Computer Corp. Also in the First Workshop on Simulation and Interaction in Virtual Environments, July 1995, University of Iowa, IA.

Dorling, D., 1991. The Visualisation of Spatial Social Structure. Ph.D. Thesis. University of Newcastle upon Tyne, U.K.

Earnshaw, R., M. Gigante and H. Jones, 1993. Virtual Reality Systems, Academic Press, London, U.K.

Evans, B., Eds., 1984. Computer-Aided Landscape Design: Principle and Practice. Proceedings of the "Greenchips" Symposium held at the University of Strathclyde. The Landscape Institute, Glasgow, U.K.

Foley, J. and A. van Dam, et al., 1996. Computer Graphics: Principles and Practices. Second edition. Addison-Wesley Publishing Company, Inc.

Fridley, J. et al., 1991. Visualising Engineering Design Alternative on Forest Landscape. Presented at the 1991 ASAE annual meeting, Paper no. 917523.

Funkhouser, T. A., 1993. Database and Display Algorithms for Interactive Visualization of Architectural Models. Ph.D. Thesis, Computer Science Division, University of California at Berkeley.

Graf, K.C., 1995. Realistic Landscape Rendering Using Remote Sensing Images, Digital Terrain Models and 3-D Objects, Ph.D. Thesis, Dept. of Geography, University of Zurich, Switzerland.

Gross, M., 1991. The Analysis of Visibility-Environmental Interactions between Computer Graphics, Physics, and Physiology, Computers and Graphics, Vol. 15: No. 3. pp. 407-415.

Gross, M. and H-J. Koglin, 1990. Computer-aided Overhead Line Design-New methods and Algorithms in Proc. of the 10th PSCC, London, U.K.

Gruber, M., M. Pasko and F. Leberl, 1995. Geometric versus Texture Detail in 3-D Models of Real World Buildings in Automatic Extraction of Man-Made Objects from Aerial and Space Images (Gruen, A., et al., Eds.), pp. 189-198. Birkhauser Verlag, Basel, Switzerland.

Gruen, A., O. Kuebler and P. Arouris, Eds., 1995. Automatic Extraction of Man-Made Objects from Aerial and Space Images. Birkhauser Verlag, Basel, Switzerland.

Hearnshaw, H. and D. Unwin, 1994. Visualization in Geographical Information Systems, John Wiley & Sons, U.K.

Henricsson, O. and A. Gruen, 1995a. Overview of Research Activities at ETH-Zurich in Automated 3-D Reconstruction of Buildings from Aerial Images. Technical Report, Institute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology, Zurich, Switzerland.

Henricsson, O. et al., 1995b. Automated 3-D Reconstruction of Buildings and Visualization of City Models, Technical Report, Institute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology, Zurich, Switzerland.

Henricsson, O. et al., 1996. Project Amobe: Strategies, Current Status, and Future Work, Technical Report, Swiss Federal Institute of Technology, Zurich, Switzerland.

Herbert, M., 1987. A Walk on the Clydeside in CADCAM International, Vol. 6: No. 4. pp. 41-42.

Isdale, J., 1998. What is Virtual Reality? A Web-Based Introduction, version 4, draft 1.

Janssens, J and R. Kuller, 1986. Utilizing an Environmental Simulation Laboratory in Sweden, Foundations for Visual Project Analysis (Smardon, R. et al., Eds.), John Wiley & Sons, pp. 265-275.

Kaneda, H. et al., 1989. Three-dimensional terrain modelling and display of environmental assessment in Proc. ACM SIGGRAPH '89 Vol. 23(3), pp. 207-214.

Kang, S., A. Johnson and R. Szeliski, 1995. Extracting concise and realistic 3-D models from real data. Technical Report 95-7, Digital Equipment Corporation Cambridge Research Lab, October 1995.

Kofler, M., 1996. The Sodahall VRML Jumpthru. MSc Project. Department of Computer Science, University of California at Berkeley, CA.

Kofler, M., 1998. R-trees for Visualizing and Organizing Large 3-D GIS Databases. Ph.D. Thesis. Technical University, Graz, Austria.

Koutamanis, A., H. Timmermans and I. Vermeulen, 1995. Visual Databases in Architecture, Avebury, England.

Landis, J. and M. Zhang, 1998. The Second Generation of the California Urban Futures Model. University of California at Berkeley. Forthcoming in Environment and Planning B: Planning and Design.

Levy, R., 1995. Visualization of Urban Alternatives in Environment and Planning B: Planning and Design, Vol. 22: pp. 343-358.

Liggett, R. and W. Jepson, 1995a. Implementing an Integrated Environment for Urban Simulation: CAD, Visualization, and GIS in Visual Databases in Architecture (Koutamanis, A., H. Timmermans, I. Vermeulen, Eds.), pp. 145-160. Avebury, Ashgate Publishing Limited, England.

Liggett, R.S. and W.H. Jepson, 1995b. An Integrated Environment for Urban Simulation, in Environment and Planning B: Planning and Design, Vol. 22, pp. 291-302.

Liggett, R., S. Friedman and W. Jepson, 1998. Interactive Design/Decision Making in a Virtual Urban World: Visual Simulation and GIS. Internal Report, School of Architecture, University of California at Los Angeles, CA.

Marrin, C., 1998. Beyond VRML.

Maver, T., 1975. Computers in design: the work of ABACUS in Architectural Design, Vol. 45: pp. 118-122.

Maver, T., 1987. Modelling the Cityscape with Geometry Engines in Computer-Aided Design Journal, Vol. 19: No. 4. pp. 193-196.

Mazuryk, T. and M. Gervautz, 1996. Virtual Reality: History, Applications, Technology, and the Future. Technical Report, Institute of Computer Graphics, Vienna University of Technology, Austria.

McAulay, I.C., 1988. Visual Descriptors: A Design Tool for Visual Impact Analysis. Ph.D. Thesis, Dept. of Architectural and Building Science, Strathclyde University, U.K.

McNamara, K., 1998. The Representation of Professionalism, Review on Bosselmann's Representation of Places for H-Urban.

Mitchell, W., 1995. City of Bit., MIT Press, MA.

Mugica, M. and J. V. DeLucio, 1996. The Role of On-Site Experience on Landscape Preferences: A Case Study at Donana National Park (Spain), Journal of Environmental Management, Vol. 47: pp. 229-239.

Nakamae, E., et al., 1986. A montage method: The overlaying of the computer generated images onto a background photograph, Proc. ACM SIGGRAPH '86, Vol. 20(4), pp. 207-214.

Pettifer, S., 1996. Beyond the Metaphysics of Virtual Reality. MSc Thesis, Dept. of Computer Science, University of Manchester, U.K.

Purdie, C., 1979. View and Bible Verification, Abacus original paper, Abacus project, University of Strathclyde, Glasgow, U.K.

Purdie, C., 1983. Computer-Aided Visual Impact Analysis, Ph.D. Thesis, University of Strathclyde, Glasgow, U.K.

Quarendon, P., 1993. Toward Three-Dimensional Models of Reality in Virtual Reality Systems, (Earnshaw, R., M. Gigante, H. Jones, Eds.), pp. 249-261, Academic Press, London, U.K.

Raper, J., Eds., 1989 Three-Dimensional Applications in Geographical Information Systems, Taylor & Francis, Hampshire, U.K.

Reuter, D., 1997. Non-CAD Computer Graphics in Design and Planning. College of Environmental Science and Forestry, State University of New York, NY.

Rinner, C. and D. Schmidt, 1998. Dimensions of DSS Applications in Urban Planning, Proceedings of Groupware in Urban Planning Workshop, Lyon.

Sackman, H., 1972. Advanced Research in Online Planning: Towards Creative Problem-Solving (Sackman, H. and R. Citrenbaum, Eds.), pp. 3-67. Prentice-Hall, Inc., Eaglewood Cliffs, NJ.

Samet, H., 1990. The Design and analysis of spatial data structures, Addison-Wesley Publishing Company.

Sanders, K., 1996. The Digital Architect. John Wiley & Sons, New York, NY.

Schmitt, G., 1996. Urban Space Simulation by Computer Graphics.

Sheppard, S., 1989. Visual Simulation: A User's Guide for Architects, Engineers, and Planners. Van Norstrand Reinhold, New York, NY.

Shiffer, M., 1995. Interactive Multimedia Planning Support: Moving from Stand-Alone Systems to the World Wide Web, Environment and Planning B: Planning and Design, Vol. 22, pp. 649-664.

Sillion, F., G. Drettakis, and B. Bodelet, 1997. Efficient Imposter Manipulation for Real-Time Visualization of Urban Scenery, Proceedings of Eurographics '97, Budapest, Hungary, Vol. 16, No. 3, pp. c-207-c-218.

Smardon, R. et al., 1986. Foundations for Visual Project Analysis, John Wiley & Sons.

Streilein, A., 1996. Utilization of CAD Models for the Object Oriented Measurement of Industrial and Architectural Objects in International Archives of Photogrammetry and Remote Sensing, Vol. XXI: No. B5. pp. 548-553.

Sun Microsystems Computer Corp., 1994. Virtual Reality from Sun.

Tan, B.H., 1997. Predicting Visual Impact of Man-made Structures in the Scottish Countryside. Ph.D. Thesis, The Robert Gordon University, Aberdeen, U.K.

Teller, S., et al., 1998. MIT City Scanning Project. Internal Report, MIT Computer Graphics Group.

Turnbull, W. and I. Gourlay, 1987. Visual impact analysis: a case study of computer-based system, Computer-Aided Design Journal, Vol. 19: No. 4. pp. 197-202.

Ulbricht, C., 1994. Vantage Point: An Interactive Landscape Simulation Tool for Engineering Design. MSc Thesis. University of Washington, WA.

Vacca, J., 1998. VRML clearly explained, Academic Press Professional.

VRML Org., 1998. Specifications of Virtual Reality Modeling Language

Wang, X. and A. Hanson, 1992. A System for Surface Texture and Microstructure Extraction from Multiple Aerial Images, Technical Report, Department of Computer Science, University of Massachusetts, Amherst, MA.

Wexelblat, A., Eds., 1993. Virtual Reality: Applications and Explorations, Academic Press, London, U.K.

Whitelegg, T., et al., 1981. An environmental assessment methodology and computer-aided design techniques for the visual analysis of transmission line routes. in Cigre Symposium, Transmission Lines and the Environment, pp. 210-211.