RESUMEN. Palabras claves: Diseño cognoscitivo, percepción, herramientas de diseño, realidad virtual.

Transcription

1 RESUMEN ABSTRACT. Executed in the C2 virtual reality space, a Virtual Architectural Design Tool (VADeT) was developed to explore factors involved in design in full scale. C2 is a synthetic CAVE environment providing a fullscale setting for image projection and perception. Applying this tool for design has four advantages over other CAD systems. First, it enables navigation performance in full scale to create a sense of immersion and reflection of the seeing-as. Second, it allows creation, modification, and editing of threedimensional objects in a virtual space. Third, designs can be modified and viewed simultaneously either inside or outside of the generated model to obtain the best design product. Fourth, the entire design process can be recorded and played back. Collectively, this tool serves the purposes of: (1) a three-dimensional sketching tool for manipulating 3- D objects, (2) a study tool for transparently displaying the design processes, and (3) a teaching tool to demonstrate the processes of how designers design. Thus, design in a full-scale representation not only is possible, but also is an unconventional mode that influences design thinking. keywords: Design cognition, perception, design tools, virtual reality. Elaborada en el espacio de realidad virtual C2, las herramientas del diseño virtual de Arquitectura fueron desarrolladas para explorar factores que involucran un diseño a escala real. La C2 es una cabina de ambiente sintético que genera imágenes a escala para percepción. Aplicando esta herramienta 0 1 de 0 diseño tiene cuatro ventajas sobre los sistemas CAD. Primero: permite la navegación en escala real creando un sentido de inmersión y reflexión de lo visto. Segundo; permite creación, modificación y edición de objetos tridimensionales en ambientes virtuales. Tercero; los diseños pueden ser modificados y vistos simultáneamente tanto de afuera como de adentro del modelo generado para obtener el mejor producto. Cuarto; el proceso de diseño puede ser grabado y revisado nuevamente. Colectivamente, esta herramienta tiene el propósito de: (1) elaborar un esbozo tridimensional para manipular el objeto en 3D, (2) desplegar de manera transparente el proceso de diseño y (3) demostrar el proceso académico de cómo los diseñadores diseñan. Por lo que, el diseño en una representación a escala real, no-solo es posible sino que es un método no convencional que influye en la actitud del diseño. Palabras claves: Diseño cognoscitivo, percepción, herramientas de diseño, realidad virtual.

2 Design in a Full-scale Immersive Environment Chiu-Shui Chan, Ph.D.Virtual Reality Applications Center / Architecture Iowa State University cschan@iastate.edu How feasible is it to design in full scale? Why would designers want to design that way? If full-scale design is worth doing, then how can it be done? Can technology be utilized to make it possible? If it is possible, then in what aspects that would affect or alter design thinking? These are interesting, bold questions for design educators and researchers to explore. This project intends to take a lead in this regard. 1. Why not design in full scale? Traditionally, architectural design begins with an idea in mind; then, either a two-dimensional concept is sketched on paper or a three-dimensional study model is constructed to visually perceive the form and to evaluate the design product. Rarely will a designer do design in full scale.

3 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA This is due to economic and logistic constraints that prevent designers from creating full-scale drawings or building full-scale models for large architectural projects. Thus, designs are typically done in small scale. However, conventions of sketching on paper and constructing miniature models are very much culture-bound. Even though the conventional methods provide a means for cultivating ideas, they set up certain limits on stimulation and inspiration. Because designers cannot physically project themselves into the space in the same way as they physically exist, the experiences of seeing-as and reflection in-action 1 are limited, and the scope of design thinking is narrowed. 2. Why design in full scale? If a full-scale model can be built and visualized, designers can discover from the model whether a doorway is too large, an access space is too limited, or a beam is not in the right position. If the full-scale model is schematic and abstract in form, it will provoke thinking and provide more clues about spatial proportions, relationships among spaces, and possibilities for sculpturing the form of the space during the primary design stage. In either of the cases, a design product represented in a full-scale model would enhance understanding of the design tremendously. If a medium exists which would provide designers with such an opportunity to interactively see design, a sense of projection into the space and the design will be created. Such a projection has the potential to yield different perceptions, enriching the mind set and diversifying the design processes. In the design profession, this diversity will stretch personal vision, enlighten mental images, and stimulate multidirectional thinking 2,3. All these are likely to improve design quality. Furthermore, diversified media applied in design will broaden thinking, change the design process, and enhance creativity. Any design medium that cultivates diversity implemented in full-scale representation could be seen as an appropriate future design tool. Conventional design methodologies involve imagining the form while conceiving design concepts threedimensionally in the mind, then revealing them through sketching or modeling. The mental image in this exercise is small in size and abstract in shape, similar to those miniatures in Alice s Wonderland. It is thus questionable whether designers, after becoming acquainted to the old convention, can open their minds to accept change and adjust their mental operations to see full-scale images. Especially, is it necessary to transform the thumbnail mental image into full scale for presenting it outwardly and then simultaneously encode it back to memory for designing? And how easy is it for designers to get the entire picture of the design (building) while they are at the moment in a remote corner of the building (e.g., in a room at the end of a long corridor)? If, however, there is a mechanism equipped with dexterous zooming and navigation capacities, designers likely would have no cognitive problems between encoding, perception, and mental operation particularly rotation 4,5. Then it would be possible and worthwhile to do design in full scale. 3. How can one design in full scale? Virtual Reality (VR), a very advanced human-computer interaction tool 6,7, provides a diversified medium for visually, aurally, and interactively experiencing activities

4 and behaviors conducted in cyberspace. Most VR facilities have the character of immersion, which is created by perceiving images through three-dimensional visual devices 8. The perception in VR is similar to projecting viewers into the environment. Applied to design, its immersive and interactive nature enables designers to perceive, grasp, and manipulate threedimensional building elements in a virtual space. By implementing design in VR and seeing the design close to reality, designers can immediately understand the spatial qualities, visualize the color and texture of materials, comprehend the major components of the heating, ventilation, and air conditioning (HVAC) system, experience the proportions of the spatial layout, and appreciate the aesthetic expression of the structural elements. Therefore, it can be beneficial to do design inside a VR environment. Advanced VR tools applying Immersive Projection Technology 9 (IPT) could display objects in large scale and would allow interaction between the user and the environment. Coping with high-speed, real-time image generation, a number of screen configurations have been developed in the VR field, such as the bench display 10 and single-wall displays 11. Furthermore, multi-wall displays, curved-screen displays, and CAVE 12 displays also can generate advanced full-scale images. Thus, combined with a design tool, it is feasible to do design in a full-scale VR environment giving a fullscale sense of immersion, which helps individuals to block out distraction and focus solely on the information with which one wants to work. Effects are similar to that of an audience and a theater production. Only when the audience is immersed in the theater will its attention focus on the actors performances. When it works, the theater has the power to engage the audience and hold its attention. This total concentration can convince, teach, and inspire. Particularly, the sense of vision overrides all other sensory modalities in an immersive virtual environment. 4. Applying virtual reality in design Theoretically, a full-scale, navigable, and interactive VR application has the potential of visualizing the design products, diversifying perception, and consequently enriching solutions. Design taught in this fashion should be effective, and projects done in VR in professional practice should also have great potential for developing concepts, communicating ideas, and presenting projects 13. However, perception in VR space is direct and prompt, and therefore the character of immersion would reduce the level of abstraction and ambiguity for perception, which might affect design thinking. Explained from a different point of view, a design product is the outcome of a series of cognitive mechanisms and activities. The full-scale virtual reality environment, when used as a design tool, is certain to change cognitive perception and have an impact on design thinking. 4.1 Perception and representation In cognitive science, perception is recognition. For instance, the moment we see a friend, his or her face triggers the image template of the face stored in our memory, and the name associated with the template is simultaneously recalled. Thus, we can address the friend s name immediately. This face template has an image pattern stored in our memory, and a rapid interpretation is activated after the cognitive in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design

5 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA mechanism of pattern matching is executed. As such, perception is to recognize, be aware of, or to understand the message revealed in the representation. Representation means to have something standing in for something else and is the means of symbolizing the things that happened in reality 14,15. When perceiving objects in virtual reality, viewers would fetch the hidden message based on their experience and prior knowledge. According to Anderson 16, our cognitive systems utilize various pattern recognizers to structure sensory input and to decode the message. Thus, it is assumed that visual perception involves the following sequences of processing information: (1) an object is first visualized, then the sensory information is encoded into shortterm memory; (2) after the sensory information is stored in short-term memory, it will be used as a sensory code; (3) the cognitive processors proceed by searching for the patterns stored in the long-term memory to find a good match with the visual sensory codes; (4) psychological attention is paid to recognize the hidden patterns in the visual sensory codes and starts the interpretation. Sometimes, the stored pattern or image has no name associated with it, or a name is saved but the associated image is not clear or is missing. In either of these cases, the pattern matching does not work and the search fails. Even though the image patterns stored in memory and the memory search methods (mainly by association) utilized for recognition affect perception, the means of representation of an image is really the key factor for perception 17. Sometimes, what the eyes have seen and what the mind believes can be two different things. Whenever the visual information is uninterpretable or misinterpreted, an illusion is produced. As there are differences between the VR representation and the conventional drawing-and-modeling representation, the illusions created from each are also different. 4.2 Illusion in conventional representation Illusion is either caused by the failure to correctly interpret information during the process of perceiving a visual object, or by the misunderstandings upon perceiving the representation. Sources that can trigger illusion are abstraction, ambiguity, and distortion that exist in the representation of information. In design, illusion sometimes can create new sources for thinking and for solution generation. Thus, it is important to explore the differences between the illusions that are generated from a conventional representation and from an immersive VR representation. Abstraction. In the fine arts, quick sketches are always simple and schematic, which would generate abstraction and provide room for interpretation (Figure 01). When a viewer sees a quick sketch, various senses of uncertainty emerge and personal interpretations are needed for comprehension. Different persons perceive different abstractions from one drawing and yield different interpretations. Fig. 01. Sketches generate abstraction.

6 Fig. 02. Louis Kahn s sketch of central Philadelphia. Ambiguity. Different types of lines utilized in a drawing suggest different levels of accuracy. A thick-line drawing is less precise and more subject to a viewer s interpretation ( Louis Kahn s sketch of central Philadelphia shown in Figure 02). On the other hand, working drawings are more accurately drawn with fine and straight lines joined precisely; the level of ambiguity is minimized and the flexibility for interpretation is limited. Distortion. Special effects applied in perspective drawing, such as punctuating colors, varying line thickness, and switching drawing focus to cheat perception, can overemphasize certain parts of the form and create a distorted and surrealistic illusion. As a result, viewers lose the sense of reality and generate misinterpretation. 4.3 Illusion in VR representation but lower levels of reality. Ambiguity. In VR, digital models do not have the luxury of allowing flexibility and ambiguity. Chances for wild guesses and room for interpretation are rare. Especially, applications requiring at-scale viewing are very demanding and require a precisely scaled presentation of the virtual object, whether it is an automobile body or a human brain. However, in the IPT display, there are cases when some objects were seen to be inappropriately large or small, or they appear at the wrong distance. Thus, there are different types of ambiguity in VR. Distortion. In VR, similar results of switching focus to create distortion might be possible by putting different details on different parts of the model. Particularly in IPT display, the three-dimensional space inside an object may appear distorted, or viewers may have difficulty with double vision or with blurred images, which might distort attentions and generate different interpretations. Therefore, distortion in VR is different in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design In VR, CAD modeling can be done on fast and imprecise sketching, even though modeling tasks usually require precision. Therefore, abstraction, ambiguity, and distortion still exist in VR representation, but have different characters and dimensions. Abstraction. In a VR environment, only three-dimensional objects exist, which are presented in some reality. Abstraction of 3-D objects is suggested by fewer details provided in the shape of the digital form. If forms are simple and colors are primitive, they will be perceived as having higher levels of abstraction In sum, conventional miniature models and drawings provide clues to understand design concepts and room for imagination. However, it is difficult to fully perceive interior spaces and their spatial proportions inside the building. In full-scale immersive VR, the digital models could be as true as the reality, which would reduce chances of generating conventional illusion from ambiguity, abstraction, and distortion. On the other hand, in comparison with the traditional representations, VR representation might create a new dimension of perception. Then, what are the new illusions? How will illusions in VR affect thinking and the implementation of CAD in full-scale VR? These

7 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA serve the purposes of exploring design in full-scale VR environment. 5. Design in full-scale VR VR, with its immersive and interactive capabilities, allows digital models to be constructed at a greatly reduced cost. However, most current VR environments 18,19,20,21 for architectural design are either primitive or require external modeling packages to get a good visual exploration. For instance, it is necessary to use computer modeling packages such as Softimage or MultiGen to complete a building model before it is imported to and displayed in the VR environment. When mistakes are found in the model while exploring it in the virtual environment, corrections cannot be made immediately. Modifications have to be done off line once the exploration is completed. This method limits the flexibility and plasticity of its application to the schematic design stage, and is good only for using VR technology as a production tool instead of as a generative tool. Moreover, it is not always feasible to compare the newly generated design with an older version simply by reloading a database. The lack of immediate interaction and feedback may cause visual perception errors while modifying the design in a more traditional modeling package. Further, the user may not recall the exact design flaw observed in the virtual setting. On the other hand, a design cannot be done in VR without having tools available. If there are building materials and modeling operators handy in the VR space, designers can simply select and build rather than having to construct the model elsewhere. Therefore, an on line computer model and methods of saving design data are needed. This is the idea of building a CAD system in the VR to help design 22,23,24. A number of researchers are working on producing a virtual walkthrough space to build 3-D shapes 25,26,27. Chu et al. had built a multi-modal VR-CAD system 28 which can be used at a Virtual Table driven by data gloves. Their efforts focused on the algorithms and techniques required for shape construction, placement, and interaction. To develop an appropriate design tool in VR, efforts must concentrate on exploring means to simulate an environment for users to do modeling with a userfriendly interface. Following this line of thought, a VR design tool is developed to: (1) enable designers to create a three-dimensional digital model and to experience 3-D design at the early, primitive design stage, (2) facilitate the evaluation of design decisions made much earlier than conventional design methods, and (3) create a more transparent tool for a virtual perception of the creative processes. This VR design tool is termed Virtual Architectural Design Tool (VADeT). 6. VADeT system VADeT was developed to explore the advantages and disadvantages of using full-scale VR for design, which is an application in C2 29. C2, a synthetic virtual reality facility, is a 12-foot by 12-foot by 9-foot cube where high-speed graphics computers render images which are back-projected onto the forward screen and two lateral sides. A projector mounted above the C2 projects the fourth image onto the floor. The back side of the C2 is open. The four-display system of the C2 uses LCD shutter glasses for the stereoscopic display and electromagnetic sensors for head and hand tracking. The C2 is controlled by two

8 Fig VADeT initial environment and the main tool palette. Silicon Graphics (SGI) 12-processor Power Onyx mainframes, each of which features two Infinite Reality graphics engines. A 3-D audio system provides localized sound. In C2, users are surrounded by three-dimensional images, projected in real time. Users can interact with the C2 by controlling a wand, which feeds back to the graphics computer. Not only an immersive and interactive setting but also a full-scale space for projection and perception, the C2 provides the facilities to implement the VADeT tool. Thus, running in C2, this VADeT system has a number of metaphorical icons symbolizing tools for selection. Tools are available for defining materials, colors, and hierarchical 3-D solid modeling operations can build up, take away, and edit an electronic building. 1 C2 is an advanced version of the CAVE facility. In the fall of, C2 is upgrading to C Tool functions VADeT utilizes the same set of the C2 input devices to enable the direct manipulation of objects and an immediate stereoscopic display of the effect of users actions. Furthermore, the VADeT system can also be used in an n-vision head-mounted display equipped with the same tracking and input devices as the C2. Whenever the system starts, it presents users with an empty space delimited by a floor, a 3-D grid (Figure 03), and a set of icons representing design tools (Figure 04). Users can select a desired icon from the menu for form generation, modify the new object to fit its real dimensions, and then place it on the right position. Five major groups of design tools are provided as the tool palette: entity 30 creation, editing, color and texture, storage and retrieval, and recording of the design processes. The hierarchy of menu selection is of standard tree structure. The five major tool groups are installed on the root level. Each associated submenu will be activated once it is selected. In the following, the five major groups are explained briefly. In this paper, the term «entity» represents geometric objects Entity creation and placement The entity creation group is marked as the «make» option in the tool palette, which allows users to interactively select a basic entity and place it in the 3- D space. Users can select and place as many entities as they desire (within system memory constraints). Numerous entities can be grouped into one entity and the modifications made will be propagated to the entire group. The main coordinate axes for the space creation are located in the scene of the C in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design

9 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA The coordinate axis arrows are attached to the wand position. The current VADeT system provides 11 predefined primitive solid entities of box, cylinder, torus, wedge, sphere, pyramid, and cone (Figure 05). Several entities represent building structural elements, for instance, box of wall and beam, torus of arch, cylinder of column, etc Entity editing After entities are created in the virtual space, designers can modify their location and size by applying the translate, rotate, and scale operators. Because of the speed of the real-time computation, accurate placement of objects in 3-D space is not an easy task, and three modes designed particularly for translation, rotation, and scale are further developed. The mode of free motion translates objects in any distance, and designers can quickly place the entity at an approximate location. The second mode of binding allows designers to put constraints on how the transformations apply to the entity. For instance, translations and rotations can be restricted to only one axis under a certain degree of values. The third mode is the «snap-to-grid-and-angle» constraint, which sets the granularity of the grid and angles for accurately revising the dimensions and orientation. These functions will fulfill the requirement of accuracy in digital modeling. Entities also can be copied, cut and pasted, and deleted in much the same way as the utilities applied in most of the CAD systems. During the operation of these functions, a clipboard is implemented, serving as a memory buffer to temporarily store the entities and their attributes Color and textures Colors and textures are unique elements in design. Decisions about particular colors and textures are determined by the designer s personal preference, design knowledge, and aesthetic appreciation. Currently, the system treats these aspects as informational issues. Thus, several tools have been developed to serve as information resources for decision making. One tool is the color palette 31 (Figure 06). This palette uses an HSV and RGB color mode to obtain the right color schemes for objects. The second tool is a texture palette (Figure 07), which has fundamental building materials for major structural components. Designers can assign and reassign any color or texture to the generated objects Model navigation, storage, and retrieval Objects existing in a full-scale model are large compared to a human scale. It is not easy for designers to move a large entity quickly within the design space 32 in VR or in reality. Thus, a navigation tool is provided to Fig. 06 Fig. 05. Shape selection palette.

10 Fig. 08. Heads-up display showing the operation of translation Fig. 07. Color palette and texture palette. let designers easily move inside or outside the building or to new locations within the virtual space. One of the key features of VADeT is the ability to save the design at any time during the interactive session. The saved design can be retrieved for future visits. The purpose is to allow the entire design process to be done within the virtual environment. Currently, VADeT stores the model and all attributes in a custom file format. Other popular model formats, i.e., VRML, DXF, and FLT, will be considered in the future to achieve compatibility for importing and exporting models between VADeT and other tools Recording the design process The series of design actions is capable of being recorded in the order the actions are performed in the system. This is similar to achieving the goal of «undo» and «redo» operations. Because design is a creative process, it is very likely that designers would change their minds as the design evolves. It also serves the goal of making VADeT not only a design tool, but also a teaching tool. For instance, saving the design actions of a master architect provides students with valuable data to learn what design methodology is used, how the different entities are combined, why a design action is creative, and what steps lead to achieve such a creative level Heads-up display During the design process, information about the actual dimensions of the selected entities or the parameters of the operators in use might be needed as visual and memory aids for perception purposes. Therefore, a heads-up display feature is implemented that displays the status, parameters, and attributes of the user s immediate actions. Figure 08 shows an example of heads-up display during a translation operation VADeT implementation and its system architecture VADeT has been designed using an object-oriented approach, focusing on supporting interactive performance, incremental development, extendibility and display interface independency. A set of C++ classes comprises the main core of the system. Three main classes are defined to correspond to the main components that constitute the system: the environment class, the palette class, and the shape class. The environment class is the top-level class, which contains all the member functions required to control the execution of a VADeT session. It includes member functions to build the model, process a user s actions, and coordinate the communication between the palette and the shape classes. The system design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design

11 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA environments in C2 as well as in the head-mounted display are fully immersive. The C2 software library abstracts the implementation and configuration of the VR devices, providing flexibility in adding new immersive devices to be used in the VADeT system. The palette class is the parent of two subclasses: the input palette and the picking palette. The input palette class is for objects that allow user input, such as the color palette. The picking palette class is used for objects where only optional selections are allowed, such as the main tool palette (Figure 03). The shape classes maintain all the different entities supported by the system. These classes keep shape attributes such as location, orientation, color, and texture, as well as the shape modification history. The shapelist classes store a «clipboard» which is used to perform cut, copy, and paste operations. The control and synchronization of the C2 and the headmounted display are handled by the C2 software library; its discussion is beyond the scope of this paper. 7. Evaluation of full-scale designing Features of the VADeT include capabilities to perform solid modeling with a variety of solid primitives and editing operations. This tool system provides several traditional and new functions. The traditional functions include the creation, placement, scaling, coloring, and texturing of design primitives. The new functions include navigation inside the model, saving the design process, and redisplaying the design process. At the present time, though this system does not have Boolean operators installed to perform high-level form manipulation, it is equipped with adequate functions for form generation. In order to test the performance ability, two design experiments were conducted for evaluation. Two subjects were asked to participate in the study by completing a simple kitchen design using the VADeT system in the C2. Design results were not required as conventional drawings, but as a full-scale digital model of a kitchen. Walls, windows, and appliances were created for completing the design problem. There was no time limitation; however, the subjects were asked to think aloud, and the entire design processes were audio- and video-recorded for data analysis. This technique is similar to applying protocol analysis as an experimental and exploration tool to study design problem solving 33 and individual style 34,35. The given design problem is the following. «The owner will provide $250,000 to build a new home with four bedrooms and a three-door garage in Ames, Iowa. The site has not been selected and decided yet, but the client would like a schematic design of the new kitchen first. There are several design constraints the client requires views and reduction of street noise as much as possible. The whole family enjoys a colorful appearance and well-made materials. Please determine the size of the kitchen, dimensions of each appliance, possible circulation, and the orientation needed for this kitchen. The north of the site is directly in front of you. Please start your design now.» Fig. 09

12 Fig Three-dimensional view of the design results completed by the subject A. Fig. 12. Diagram of the floor plan Experiment A The subject A had been working for an architectural firm in CAD for three years before this experiment was conducted and had applied VR for his diploma project. He spent a half-hour familiarizing himself with the s y s t e m operations and 75 minutes to complete the design, for which the floor plan is shown in Figure 09. The subject indicated that no particular concept or image existed in his mind before starting the design. His entire design used the primitive geometry provided by the system. The sequence of design generation was diagrammed as shown in the right-hand column of Figure 12. The alphabetical order symbolizes the chronological order of form generation. Five characteristics are found in the design processes of subject A. These characteristics suggest the influence of fullscale perception on design thinking. to any particular dimensions of the generated objects. Objects were created by using other objects as reference of scale. Therefore, after an object was generated, the subject had to make extra efforts to adjust the sizes of the object, or its adjacent objects, and the spatial relations between them. This relates to the location, size, and scale of the connected objects as a group, because the subject could perceive instantaneously the proportion of the object and its surrounding space. Since perceived visual distortion could not be tolerated, adjustment occurred immediately. Materials or textures were assigned immediately after an object was created to solidify its existence. The subject spent little time formulating the design problem. This may have been caused by the immediate psychological projection into the environment; the subject may not have had time to analyze the problem. The entire process seemed more intuitive rather than deliberate. One theory is that because the perception is full of 3-D images, the thinking process tends to be driven mainly by geometric (visual) thinking instead of conventional logical reasoning. The size of the kitchen is close to the dimensions of the C2 setting of 12' by 12' by 9' in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design The subject did not utilize or pay attention

13 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA 7.2. Experiment B The subject B was a senior architecture student at the time this experiment was conducted. He also had applied VR techniques to finish his diploma project. The total time needed to finish the design was 3 hours and 12 minutes. Results of 3-D images are shown in Figures 13, 14, and 15, and plan view is shown in Figure 16. The design solutions are to locate the kitchen facing north, to have the south side leading to the dining area, with bay windows on the facade for exposure to southern light. The counter island provides a snack bar function with views through the south (Figure 14). The chosen materials are marble finishes for the countertop, ceramic tiles for the island countertop, and stylish designs for chairs to address the requirements of well-made materials. The subject applied color and materials to all entities. The entities were mostly created on the side, then moved and placed on the right location afterwards. The subject considered the rule of the kitchen triangle 36 in the process. Fig. 14.Section perspective of the design results by subject B. Fig. 13. Three-dimensional view of the design results by subject B. The subject indicated that immediately after he read the design task, he had a solution which was in the form of a U shape with a snack bar eating area in the middle, and bay window on the south side for dining. The rest of the kitchen was generated intuitively and conceptualized in real time to find out things that worked the best for each other. Therefore, the design started from constructing the counters and cabinets to form the U shape and proceeded to the counter island in the middle. Others were thought out instantaneously. Several characteristics appeared in the process of design session B. Most of the time, the subject applied the snap function with a quarter-inch spacing interval to move entities to the right location as well as to scale objects up or down to get the correct size. No alternative design solutions were created or considered, and the whole process was linear. The subject spent a lot of attention on the proportions of each object, their spatial relationships with adjacent objects, and their location in the space that fulfilled functional links with other objects. The subject used his body as a reference scale; for instance, he utilized the height of his hip and the length of his palm to measure the height of the counter and the size of the cabinets handles. 1 Kitchen triangle is a design convention to reduce circulation conflict in the kitchen operations. Metaphorically speaking, lines joining these three elements of sink, range, and refrigerator form the known «work triangle.» According to Neufert (see Neufert, E Architects Data Crosby Fig. 15. Three-dimensional top view of the design results by subject B.

14 Fig. 16. Diagram of the floor plan. Lockweed Staples, London (1980) pp 54-56), the distance between sink and range should not exceed 1.8 meters (5'-10") and the total of the three triangles sides should be between 5.5 meters (18'-4") and 6 meters (19'-8"). 8. Observations One limitation of VADeT found from the experiments relates to the availability of needed functions or commands. For example, at the time, the system could not generate organic forms due to not being able to efficiently run Boolean operations. Another example was found on experiment B; when the experimenter asked for the rationale of placing the refrigerator next to the stove, the subject reported that he had considered moving the refrigerator to a different location but eventually left the refrigerator in the original spot. It was because the system, at the time, did not have an efficient «group» function to allow moving multiple objects simultaneously, and it was a tedious process to move objects one by one. Thus, any CAD system must provide necessary capacities to fit design purposes. The small samples of experiments yielded interesting observations about designing in a full-scale VR environment. Of course, it is risky to jump to the conclusion or to claim that the observations are really representing the design activities of all designers. However, it is worthwhile to report the findings so far. Quite often, subjects checked the function of circulation of each generated objects in the entire process. Particularly, there are no approximations in determining the fluidity of circulation on walkways and structural components of objects. The VR environment provides a real-time test for the functionality, which might limit the generation of alternative solutions and sets up a linear thinking process. Design products were generated mostly by the visualization and manipulation of the objects. Both subjects were preoccupied with completing the color and texture of the objects, and arranging and sizing them to show their existence. This suggested that the VR products contained a higher level of specification than the traditional design method. Designers not only concentrate on spatial relationship, but also on the texture, color, and material. Thus, visual illusion in VR is different, ambiguity and abstraction in VR are less significant due to the immersive nature of this design tool. Because of this, VR representation is a new instrument for understanding virtual aesthetics. A full-scale model, however, also limits immediate perception of the entire site and its adjacent context and presents problems for immediately comprehending relationships among objects macro-wide. This is similar to the fact that a designer has difficulty perceiving the scene of a city block while he or she is on the street level or inside the building. Such limitations may constrain the immediate response of reflection projected between the designer and the design status, and may block the spring of immediate intuition in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design

15 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Thus, in the schematic design stage for concept generation, the best solution is to control the scale mechanisms. In other words, a small scale should be used to build large objects and to view exteriors, and then the designer should navigate inside the structure to explore the interior in full scale after the overall form is generated. 9. Discussions VADeT provides a chance to explore design in full scale. Navigating through spaces in VADeT is quick, and the change of perspective is very mobile. Objects can be generated, copied, deleted, moved, scaled, or regenerated in real size without difficulty. The whole process can be saved, re-entered, and remodified. These unique characteristics provide a greater degree of flexibility for designing than traditional sketch-andmodel methods. Regardless of the level of sophistication and the modeling functions that have been achieved, this system can be seen as a threedimensional sketching tool. Experiments conducted in this study documented how a full-scale design tool affected design processes and representations. The flexibility to change the location, size, and attributes of each object and visualize changes immediately is very beneficial for schematic design and parti design 1. It also has great potential for teaching design, because the cause and effect of any design movement made is clearly displayed. However, both subjects in the experiments used their bodies as references to determine the size of the objects. Thus, a higher level of precision in scaling was expected when designers were in the VR environment. Precision reduces conventional ambiguity, which may occur rarely while perceiving in VR space. The new illusion in VR is generated mainly by the interaction within the environment, particularly motion. In other words, virtual illusion relates to the rapid change of composition perceived from the rapid change of perspective angles. Thus, illusions in drawing perception are one-shot-and-still illusions, but in VR they are jumping-and-continuously-generated illusions. This is a new experience for users, and it is difficult to find verbal data to prove the theory. On the other hand, in order to get a quick and imprecise sketch to yield room for thinking, precision must be sacrificed, and the effects and methods of sketching in VR, in all, is a new research frontier. This relates to what Gombrich 38 had mentioned, that the language of art is not that it enables the artist to create the illusion of reality, but to give the illusion of looking into the invisible realms of the artist s mind. Because perception in VR switches from visualizing still images to moving images, from imprecision to precision, from abstraction to reality, from image distortion to distorted attention, the definition of aesthetics in reality may need to be redefined in the future. As Read 39 wrote, «We do not always realize that the theory of perspective developed in the 15th century is a scientific convention; it is merely one way of describing space and has no absolute validity.» This applies to the immersive VR representation. It also is interesting to observe the different design processes emerging from this full-scale tool. Apparently, the sense of projection and immersion is quite convincing, which alters the design behavior. The design process is almost purely visual. The overwhelming sense of the projection and immersion diverts designers attention to local details and causes

16 them to ignore the overall scheme. Thus, the jury system for criticizing VR projects should also change. For instance, when viewers are walking through a real building, the physical experience of light, sound, smell, and kinesthetic experience are created, which would never be obtained from still images of drawings, slides, or photography. By the same token, VR provides a mimic image of the reality, stimulating some sensual experience of light, colors, material, and space. In some cases, jurors might not favor or comprehend work that is more complex in a sensual way. As a result, it could set up limitation on criticism. Thus, a jury on VR should also criticize sensual experience, judge the perception in reality, and assess the virtual illusion. 10. CONCLUSIONS Virtual reality is a new technology allowing users to step through the computer screen into a 3-D artificial world. To further design in the artificial world, users should be able to manipulate objects naturally in the virtual space with their eyes, feet, and hands. To this end, new methods of modeling architectural objects directly in VR need to be further explored especially the freedom of deformation and the Boolean operators required to make organic forms. This will make the object generation processes more sculpturally oriented. The recording and redisplay capabilities of VADeT characterize its use as a tool for observing designers thinking process to understand how a full-scale representation would affect design thinking. Given the opportunity to visualize the design process in VR, viewers will be able to understand the myth of how master architects perform their creative works. In summary, design can be done in full scale by applying VR technology. With digital models displayed in different geographical locations, designers would learn various design principles, methods, and processes inherited in various design cultures. This would transcend geographical boundaries to allow architectural design to be taught, learned, seen, and touched from different locations to improve thinking. ACKNOWLEDGMENTS This project resulted from a collaborative effort. The research was funded by the Special Research Initiation Grant and a Second Discipline Grant. The implementation was done by Lewis Hill, supported by the Research Careers for Minority Scholars Program and assisted by Professor Carolina Cruz- Neira. Special thanks should go to the subjects involved in the design experiments. Without their participation, this project would have been impossible. Parts of the thoughts were presented to CAADRIA99 and the 4th Design Thinking Research Symposium, hosted by MIT REFERENCES 1 Schon, D The Reflective Practitioner Basic Books, New York (1983) 2 Sloman, A The computer revolution in philosophy: philosophy, science, and models of mind Hamester Press, Brighton (1978) 3 Boden, M A Computer models of mind Cambridge University Press, Cambridge (1988) 4 The ability of mental rotation is regarded as a part of the skillful and challenging spatial intelligence of human beings, see Gardner, H Frames of Mind Basic Books, New York (1983) pp Shepard, R N and Metzler, J Mental rotation of three dimensional objects Science (1971) pp in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design

17 Design in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA 6 Durlach, N and Mavor, A Virtual Reality: Scientific and Technological Challenges National Research Council Report National Academic Press, Washington DC (1995) 7 Mine, M ISAAC: A virtual environment tool for the interactive construction of virtual worlds UNC Chapel Hill Computer Science Technical Report TR The University of North Carolina at Chapel Hill (1995) 8 Chan, C S Virtual Reality in Architectural Design. In Liu, Wang and Hou (eds) CAADRIA 97 Workshops Hu s Publisher Inc, Taipei (1997) pp The immersive projection technology applies large-scale visualization to interact with the environment. 10 The bench display was originated by the Responsive Workbench in 1993, see Kruger, W and Frohlich, B The Responsive Workbench (Virtual Work Environment) IEEE Computer Graphics and Applications May Vol 14 Issue 3 (1994) pp Also see the Web information at projects/rwb/newindex.html. 11 Single wall display includes ImmersaDesk which is a drafting-table format virtual prototyping device. The ImmersaDesk features a four-by-fivefoot rear-projected screen at a 45-degree angle. The size and position of the screen give a wide-angle view and the ability to look down as well as forward. 12 The Cave Automatic Virtual Environment, or CAVE facility, was originally developed at the Electronic Visualization Laboratory at the University of Illinois at Chicago. See Cruz-Neira, C, Sandin, D and DeFanti, T «Surround-Screen Projection-Based Virtual Reality: The Design and Implementation of the CAVE» Proceedings of the ACM SIGGRAPH 93 (1993) pp Biocca, F Communication within virtual reality: Creating a space for research Journal of Communication Vol 42 No 4 (1992) pp Echenique, M «Models: a discussion» In Martin and March (Eds) Urban Space and Structures Cambridge University Press, London (1972) pp Hesse, M Models and Analogies in Science University of Notre Dame Press, Indiana (1966) 16 Anderson, J R Cognitive Psychology and Its Implications W. H. Freeman, San Francisco (1980) 17 It is also true that the representation of design information is the means for displaying design concepts. 18 Ellis, S R Pictorial Communication in Virtual and Real Environments Taylor and Francis, London (1991) 19 Ellis, S R Nature and origin of virtual environments: A bibliographical essay Computer Systems in Engineering Vol 2 No 4 (1991) pp Brooks Jr, F, Airey, J and Alspaugh, J Six Generations of Building Walkthrough: Final Technical Report to the NSF UNC Chapel Hill Computer Science Technical Report TR The University of North Carolina at Chapel Hill (1992) 21 Pimentel, K and Teixeira, K Virtual Reality through the New Looking Glass Windcrest, New York (1993) 22 Sutherland, I E The ultimate display Proceedings of IFIP Vol 65 No 2 (1965) pp , pp Foley, J D, Van Dam, A, Feiner, S K and Hughes, J F Computer Graphics: Principles and Practice Addison-Wesley, Reading, MA (1990) 24 Stuart, R The Design of Virtual Environments McGraw Hill, New York (1996) 25 Butterworth, J, Davinson, A, Hench, S and Olano, T 3DM: A Three Dimensional Modeler

18 Using a Head-Mounted Display Proceedings of the 1992 ACM Symposium on Interactive 3D Graphics (1992) pp Bulterworth J, Davidson A, Hench S and Olano T M 3DM - A Three-dimensional Modeler Using a Head-mounted Display Communications of the ACM (1992) pp Halliday, S and Green, M A Geometric Modeling and Animation System for Virtual Reality Proceedings of the 1994 ACM Virtual Reality Software and Technology (1994) pp Chu, C C P, Dani, T H and Gadh, T Multisensory User-interface for a Virtual-reality-based Computer Aided Design System Computer-Aided Design Vol 29 No 10 (1997) pp C2 is an advanced version of the CAVE facility. In the fall of, C2 is upgrading to C4. 30 In this paper, the term «entity» represents geometric objects. 31 The Color selection window is a part of Daniel Heath s Virtual User Interface project developed at Virtual Reality Applications Center. 32 The virtual design space is larger than the physical size of the C2 system. 33 Chan, C S «Cognitive processes in architectural design problem solving» Design Studies Vol 11 (1990) pp Chan, C S «How an individual style is generated?» Environment and Planning B: Planning and Design Vol 20 (1993) pp Chan, C S «An examination of the forces that generate a style» Submitted for publication 36 Kitchen triangle is a design convention to reduce circulation conflict in the kitchen operations. Metaphorically speaking, lines joining these three elements of sink, range, and refrigerator form the known «work triangle.» According to Neufert (see Neufert, E Architects Data Crosby Lockweed Staples, London (1980) pp 54-56), the distance between sink and range should not exceed 1.8 meters (5'-10") and the total of the three triangles sides should be between 5.5 meters (18'-4") and 6 meters (19'-8"). 37 A parti is a proposed form or a concept of a building developed by a designer from transforming the architectural program into a design. 38 Gombrich, E H Art and illusion Princeton University Press, Princeton, NJ (1960) p Read, H E The art of sculpture Princeton University Press, Princeton, NJ (1969) p in a full-scale Immersive environment/chiu Shui Chan/ ConVEACA Design