ICSID 97 The Humane Village Congress, August 23-27, Toronto. 1997 Bringing back the human dimension : sketching in virtual space Tomás Dorta, arch. doctoral student Faculté de l'aménagement Université de Montréal Philippe Lalande, a.c.i.d. associate professor École de design industriel Université de Montréal Abstract Computer graphics technology has made powerful new tools available to designers. Designers are able to create, manipulate and control form and shape with unprecedented ease and speed. This process however, imposes a high level of coding and decoding on the designer. Virtual reality (VR) can reduce this mental load by going beyond the modeling of space to actually simulating its experience. Unfortunately, VR is mainly thought of as being exclusively immersive and as such requiring the power of high speed computers and sophisticated peripherals that put it out of reach for most designers. Researchers at the École de design industriel (EDIN) of the Université de Montréal are exploring non-immersive VR and its potential in design practice and education. The experimental inclusion of non-immersive VR in an entry level computer graphics course given to first year industrial design students suggests that this technology can be effectively used in both practice and education as a concept modeling tool. With the help and collaboration of the Virtus Corporation, two three hour non-immersive VR components were added to the course to help students develop spatial perception skills. Following each component, they were given an exercise to confirm their understanding of the subject. The first exercise aimed at understanding VR space and acquiring navigation skills. By exploring a "forest" of cubes arranged in a regular 3D grid pattern, three hidden «treasures» were to be discovered and tagged. The second involved modeling the computer laboratory. A virtual mannequin was provided as the starting point around which the student modeled a chair, table and computer. The scene was progressively built up from the size of the human figure much the same way an artist intuitively sketches a scene by proportionally scaling objects in relation to a given reference. The initial results were judged very positive by both investigators and students. Further experiments are under way to scientifically measure the comparative effectiveness of non-immersive VR as a modeling tool in the initial stages of concept development. Designers can call on a wide range of traditional digital and physical tools for the representation of 3D objects. The traditional digital tools are CAD (Computer Aided Design) programs. Among these are 2D drawing, image processing, rendering and animation packages. An important aspect of these tools is their interface which has evolved greatly over the years, becoming gradually more and more friendly. Among the non-computer tools traditionally used by designers are plans, perspectives, physical models and freehand sketching. This last item constitutes the most popular and easiest to use. It has the potential to form a direct link between the designer s intentions and their graphic representation without the encumbrance of the abstract commands of a software interface and requiring minimally a pencil and paper. Through a series of almost instantaneous movements and traces, the designer positions objects, removes walls, extrudes or extends forms just by moving and erasing lines. As quoted by Lebahar (LEBAHAR, J. C.; 1983) «this practice of graphic simulation gives the designers the power to transform the environment in his mind before it reaches physical reality». The problems with traditional modeling tools The problem with most CAD applications stems from the way they have been adapted to the digital realm by simply imitating the same tasks done previously by hand and reproducing them by computer. Initially, CAD applications were referred to as Computer Aided Drawing or Drafting, leaving aside the treatment of 3D 1
objects. Three dimensional modeling programs, once they appeared, required manipulation of the same representational abstractions that hampered traditional hand drawings: 2D projections (plans), perspectives and physical models. Plans: 2D representations are a great resource for designers. They allow the three dimensional aspects of an object to be displayed in a flat two dimensional medium. They play an important role in project execution. Their main handicap is the spatial abstraction that is necessary to understand an visualize a three dimensional object represented on 2D media. Often, initial concept development takes form through 2D drawings in order to facilitate the transition to final execution drawings; this expediency directly affects the design outcome and hides errors of abstraction due to the coding a decoding sequences required to implement the standardized rules of 2D representation of 3D objects. Perspectives: In spite of not requiring the spatial abstraction previously mentioned, perspectives possess two important shortcomings: their static nature and their complexity. Hand drawn perspectives require the application of lengthy and cumbersome construction techniques which abound with opportunities to deliberately or unconsciously introduce distortions and thus affect the perception of the object being represented. Computer generated renderings apply sophisticated algorithms to automate the elaborate and error prone process of perspective construction. The production of a single view is nonetheless lengthy and each viewpoint modification requires a complete regeneration. Mock-ups: the chief drawback with physical models or mock-ups occurs when representing large objects at reduced scale. The observer becomes a giant evaluating a very small object. Scaling affects the viewing parameters and thus distorts the perception of the object s real proportions. Sketches: hand drawn sketches despite the advantages discussed earlier, pose two important problems: the difficulty of transforming the representation at will as discussed in the case of perspectives and the cognitive processing of information, in the mind, where spatial abstraction errors can easily produce distorted mental models. Decisions based on these faulty models are passed on, the errors perhaps only to be detected much later in the design process. The affected cognitive aspects The mental load (SPERANDIO, J. C.; 1980): the spatial abstraction required to perceive complex 3D objects through 2D representations requires coding and decoding of information. This process imposes an additional mental load to that already required by the task of creation. Mental models (SPERANDIO, J. C.; 1984): experienced designers acquire the ability to mentally visualize three dimensional objects and the modifications that design alternatives will bring about. In the case of less experienced designers as well as for experienced designers faced with complex objects, this mental image may be flawed, adding a new set of errors to affect the perception and comprehension of the object s characteristics. In design teams, it is important that professionals agree on using the identical mental models in order to rapidly arrive at design solutions without wasting time continuously correcting errors. Virtual reality VR is the results of improvements in user interfaces. It is the final, principally visual, intuitive interface between the digital real and the designer. VR can be defined as the systems and techniques that provide the feeling of presence in a three dimensional computer generated world. This sensation of presence is the results of the user s ability to interact directly and in real time with the virtual environment (RHEINGOLD, Howard; 1992); (ZELTZER, D.; 1992). 2
Immersive VR RV aims at placing the user inside a 3D environment that he can directly manipulate. The objective is that users forget they are interacting with a computer and feel they are intuitively controlling a 3D environment from within. To achieve this, users must don sophisticated and often cumbersome computer peripherals such as head mounted display 1 and special gloves in order to experience total visual immersion and six degrees of liberty 2 in manipulating the environment. Non-immersive VR There exists a more accessible alternative: non-immersive VR (ROBERTSON, George G. et al. «Xerox- PARC»; 1993). The user is placed in a 3D world which is directly manipulated using a conventional computer graphics work station consisting of screen, keyboard and mouse. The scene is shown with the same representation of depth (3D) as immersive VR: perspective views, hidden line removal, color, texture, lighting and shadows. Also as in immersive VR, navigation an simulation are controlled directly by the responses of the user. In non-immersive VR, the computer screen becomes a window on the virtual world. The human dimension Due to the inherent problems of traditional modeling tools and immersive VR s dependence on sophisticated equipment, we have directed our work towards exploring the potential of non-immersive VR. It is not proposed as the designer s ultimate solution but rather as another tool that can greatly assist professionals in their design work. As an application of this interaction with non-immersive VR, we have explore the notion of sketching in virtual space. Sketching in virtual space The notion of sketching in virtual space is used here as an extrapolation of freehand sketching, i.e. a 3D drawing in virtual space. It involves intuitive modeling of 3D objects based on a known object or a virtual mannequin. The mannequin serves as a reference against which to verify the form and its dimensional relationship to the human body, just as the artist constructs a scene or the tailor cuts his pattern. The richness of sketching in virtual space comes from the intuitive nature of its execution. There is no need here for dimensions as in a CAD program or standards as codes for representation. Moving through virtual space it is possible to create objects from an infinity of viewpoints, even from within. Our capacity for spatial abstractions is no longer a limiting factor in our ability to transform environments. Moreover, cognitive limits are often the cause of errors due to mis-representation or mis-interpretation. Virtual reality allow us to directly express changes and accurately visualize complex 3D proposals with greatly reduced mental load. Advantages Sketching in virtual space provides many advantages for both design practice and education. 1 Apparatus permitting the stereoscopic visualization of digital images and equipped with movement sensors which locate the position of the operator in respect to the 3D virtual environment. 2 X, Y and Z as well as the rotation of each of these axes. 3
In practice: Less spatial abstraction required in coding and decoding information from 2D representations. Rapid feedback on changes made to 3D objects. Intuitive control of navigation and modeling functions. Direct interaction with 3D virtual object generates better quality mental images. Fewer errors caused by mis-representations and mis-interpretations. In computer graphics education: Better and easier control over modeled objects. Simpler and more intuitive operational commands. Real time model manipulation and navigation avoids long delays between modifications and screen output. In design education: Better grasp of basic design principles through interaction with 3D interface. Real time manipulation of objects allows more freedom for evaluating design alternatives while reducing errors caused by mis-representation or mis-interpretation. No distortions due to scaling. Faster proposal/correction cycle due to rapid feed back. The experiment While the introduction of computer graphics in design curricula has implied major investments in time, resources and training, it has not profoundly modified the established schema of teaching methods. Computer graphics tools have simply been substituted for the previous set of hand generated tools and techniques. Because of the commitment required to master this new technology, the introduction of computer graphics imposes a very heavy load on the course structure, threatening in some cases to shift the center of gravity away from the teaching of sciences and humanities towards a more technical training. The entry level computer graphics course at the École de design industriel of the Université de Montréal confronts students, many of whom are totally unfamiliar with any but the must primitive of graphics applications, with a wide variety of software ranging from spreadsheet, to page layout, 2D vector and paint programs and 3D surface and solid modelers. The objective, far from specialization, is to introduce students to the broad spectrum of computers graphics potential and open as many doors as possible to current applications which the students are then encouraged to explore in more detail on their own according to their needs and inclinations. This is the course which served as a testing ground for the experiment in virtual reality and its application in the creation and manipulation of 3D space. The introduction of a VR component in this class was part of a scientific experiment designed to measure the comparative effectiveness of VR as a modeling tool in relation to traditional computer technologies. This 4
experiment, carried out with the collaboration and support of the Virtus Corporation 1, was the catalyst for the inclusion of a VR module consisting of two three-hour sessions within the existing structure of the class. The VR module Each of the two sessions consisted of a classroom presentation dealing with theoretical aspects of VR, demonstrations of VR applications and techniques and a practice period allowing students to explore a VR environment and attempt modeling with it. The subjects covered in the two sessions were as follows: Session 1 Session 2 Introduction to VR : terminology and definitions The historical development of VR Navigation and orientation within 3D virtual space Simple modeling techniques, scaling and positioning of objects in space The application of VR within design methodology Object libraries and file transfers Texture mapping Intermediate modeling techniques After each session, an assignment was given, to be completed individually for the following week. The assignment was designed to encourage the students to work at mastering the techniques reviewed in class as well as procure a sense of satisfaction at having achieved a non trivial result. Following is a description of the two assignments, corresponding to the two sessions on VR. Assignment 1 Exploring 3D space Objectives: Description: to become familiar with 3D virtual space to gain assurance with navigation and orientation with 3D space to master some simple modeling skills. Each student was given a file of a 3D environment consisting of an array of identical hollow cubes arranged in a 3D grid pattern, with instructions to explore this cube forest in order to discover the three that contained a hidden «treasure». Once identified, the student was required to model a chair according to given specifications and to tag the special cubes by placing a copy of their chair inside. 1 Virtus Walkthrough Pro v. 2.6 from Virtus Corporation was installed on 15 Power Macintosh 7100 computers (16/500) from Apple Computer for the purpose of the experiments. 5
Illustration A - The 3D space filled with cubes that had to be explored in order to discover and tag the objects hidden inside three of the cubes. Illustration B - The inside view of a typical cube. Illustration C - An inside view of a special cube with its «hidden treasure». Illustration D - The special cube tagged with the deposition of a modeled chair. 6
Assignment 2 3D sketching Objectives: to acquire the ability to Description: model complex objects using a limited selection of modeling techniques and an appropriate level of detail build up complex objects progressively by scaling new components in relation to certain base or reference elements add realism to objects modeled in VR space through the use of texture mapping. Starting from a 3D mannequin, the student was to model the computer lab in which the course is given including furniture, and equipment. No dimensional information was given; instead of proceeding with nominal measurements, the mannequin was used as a starting point for modeling the chair. From the chair and the seated mannequin, the tables and equipment were added, and so on until the entire lab was generated. As the work progressed, the student navigated around the scene, evaluated the proportions and made corrections in order to insure that the final result was as good a representation of reality as possible. 7
Illustration E The initial modeling stage: the chair is fitted to the mannequin. Illustration F - The table sections added to form a basic work station for two computers. Illustration G - The completed work station with furniture and equipment. Illustration H A completed lab modeled by Egon deroth. 8
The results Initial reactions to non-immersive VR on the students part was varied. Some had received college level training in 2D CAD programs - principally AutoCAD - expressed skepticism as to the ability of non-immersive VR to satisfy their perceived need for modeling precision and functionality. Novice students reactions tended to be more accepting of non-immersive VR although many expressed doubts as to their ability to produce adequate results. The general level of acceptance and the quality of work produced after only six hours of introduction to the technology and two assignments was very impressive. The second assignment in particular generated a very high level of student involvement and satisfaction as even the least experienced computer graphics users were able to demonstrate their ability to create credible relatively complex 3D environments. These easily acquired skills enabled the students to create, manipulate and modify groups of 3D objects in a form that is accurately perceived by them and easily communicated to others. They are thus able to develop ideas rapidly and easily test various design alternatives while reducing mistakes due to coding and decoding errors. The greatest potential impact for this work relates to the theme «Education and Consumption». VR and the notion of sketching in virtual space are proposed here as better learning tools for design. On the one hand, they allow faster acquisition of modeling skills; on the other, they allow a more effective mastery of the basic principles of design and their application to 3D objects. While most 3D computer programs are cumbersome to use and posses abstract interfaces, VR and particularly non-immersive VR is easily accessible through a relatively transparent and ergonomic interface. The reduction of our society s consumable by-products can be attained in part by insuring the quality of those physical objects that are produced. The application of VR technology to the design process can contribute to this goal. It can also be used in the production of products never intended for the physical realm, those that remain in the virtual domain (CAMPBELL, Dace; 1996) only accessible through cyberspace (BENEDIKT, Michael; 1991). Moving from the physical to the virtual implies dematerialization. Non-immersive VR dematerializes the design process while augmenting its potential. The physical tools of design and the commitment they require in terms of time and resources are no longer acceptable. VR allows us to transcend these limits and let creativity explore a wider range of possibilities without incurring the penalties of traditional modeling tools. The VR representation of an object is a very energy-efficient vehicle for the transfer of knowledge which can be effectively translated into physical being by technologies such as rapid prototyping. VR can be considered as a medium for 3D communication, easily and effortlessly transferable through cyberspace in the form of bytes. An interesting challenge for VR would be to model the Global Village as a means of evaluating various alternatives before translating its most Humane and error free variant into reality. References BENEDIKT, Michael (1991). Cyberspace: First Steps. MIT Press, Cambridge CAMPBELL, Dace (1996). Community and Environmental Design and Simulation, the CEDeS Lac. at The University of Washington, pp. 201-224, in BERTOL, Daniela (Ed.) Designing Digital Space: An Architec s Guide to Virtual Reality. Wiley. New York. LEBAHAR, J. C. (1984). Le dessin d architecte, Simulation graphique et reduction d incertitude. Éditions Parenthèses, Roquevaire. RHEINGOLD, Howard (1992). Virtual Reality. Touchstone. New York. 9
ROBERTSON, George G., CARD, Stuart K., MACKINLAY, Jock D. (Xerox PARC) (1993). Nonimmersive virtual reality, Hot Topics, pp. 79-83, in Ronals D. Williams (Ed.) Computer 2-93. SPERANDIO, J. C. (1980). L analyse de la charge de travail mental (chap. 5), pp. 195-219, in La Psychologie en ergonomie. P.U.F., Paris. SPERANDIO, J. C. (1984). La representation mental (chap. 5), pp. 77-90, in L ergonomie du travail mental. Masson. Paris. ZELTZER, D. (1992). Autonomy, intercation and presence, in Teleoperators and Virtual Environments 1 (1). Winter, MIT. 10