/ The Yingzao fashi
project / Teaching materials / Coursework
/ Reaction / Papers
/ Who's who / Related
/ About the materials / Design and construction / Examples / Animations / Interactive models / Student work / Glossary / Readings and bibliography /
by Andrew I-kang Li and Jin-Yeu Tsou
In this paper we discuss an experimental computer-based teaching tool for traditional Chinese wood construction, specifically the construction system of the Song dynasty (960-1127), as set out in the manual Yingzao fashi (published in 1103). We recently used this tool for the first time in a classroom setting. The experience supported the usefulness of virtual modelling for this purpose and reminded us of its limitations. Students were enthusiastic about the assignment; we speculate that it affirms their sense of identity in some way.
The long-term goal of the present project is to develop a new approach to teaching Song wood construction, based on virtual modelling. There are two main reasons to use virtual models to teach Song construction. The first is that the Song construction system and virtual modelling share a compelling conceptual commonality, namely the distinction between primitive and instance. The Song construction system defines a small number of what we may call component-types, which are nothing less than primitives: they are instantiated individually and in groups throughout the building frame. In addition, component-types do not have absolute dimensions; their dimensions are given in fen, a modular unit which ranges in value from 19.2 to 9.6 mm. Thus, for example, the cap block (lu dou) as component-type is a hypothetical form, 32 fen square in plan. (See ill. 1, p. 8, above.) Only when it becomes part of a building does it acquire absolute dimensions, ranging from about 614 to 307 mm square.
The second reason is what we may call virtuality itself, one aspect of which is the ease of duplication and modification of objects. Virtual model-making should therefore reinforce the conceptual basis of the Song construction system, which involves large-scale repetition of a small number of component-types. Real model-making, e.g., with cardboard, is too labour-intensive to be practical.
The assignment, four weeks long, was to design and construct a virtual model of the structural frame of an official Song building, according to the rules of the Yingzao fashi. Students used a computer-based teaching tool which we have been developing over the past year. This consists of a kit of virtual model parts and supporting materials.
The assignment was given in the required introductory CAD course and served that course as an exercise in three-dimensional modelling. However, we were equally interested in the tool's effectiveness in helping students learn about the Song construction system. There were 49 students, all in the first term of the second year of a university architecture program. They had had no courses in Chinese architecture, but were taking 20th-century architecture concurrently.
Our goals for the students were:
We divided the assignment into two stages. In the first stage, about one and a half weeks long, each student was to build one bay of a sample model, which used, unaltered, the parts provided in the model kit. (See ill. 2, p. 8, below.) This was to give the students practice in manipulating the components and to introduce them to the parameters governing the overall dimensions of the building, principally:
In the second stage of the assignment, lasting about two and a half weeks, the students worked in groups of three or four. Each group proposed and, after our approval, constructed a complete model of a variant building and prepared a report. Most variations centred on one or more of the parameters listed above; a few groups investigated parameters that we had simplified or eliminated, such as the sources of curvature. Students were to conform with the rules governing the values and interrelationships of the parameters.
In order to carry out this assignment, we used the following materials.
The outcome of this assignment supports our starting assumptions about virtual (as opposed to real) modelling. The first assumption was that virtual modelling makes it feasible to build complete models. This proved true for our class: 49 students with no background in Chinese architecture constructed 16 models in four weeks. Since the models had as many as 900 individual pieces, this feat clearly would have been impossible with real models. Taken as a group, the 16 models emphasize the systematic nature of Song construction by illustrating some of the varied outcomes of the interaction of a small set of rules and a small set of building parts.
The second assumption was that virtual modelling makes comparison studies possible. One group took the initiative to do a small set of comparison models, showing that they recognized this possibility and thought it useful. They constructed three models which differed only in bay width (and number of bracket sets per bay). (See ill. 3, p. 9.)
Notwithstanding the evident advantages mentioned above, there were equally evident disadvantages. The virtual models were less correct than real ones would have been - assuming, of course, that the students had been able to complete real ones - because the mistakes were possible only with virtual models; they would simply be impossible with real ones. In other words, the mistakes are attributable to the virtuality of the model. This should be a caution that the virtual model does not replace the real model; in fact, the virtual model opens up new possibilities not only for learning, but also for error.
Students made two kinds of mistakes. The first appeared as impossible connections among structural members. This is a serious mistake, as the interlocking of members is a key feature of Song construction. Our cardboard model makes this type of error physically impossible, but our virtual model lacks this kinesthetic feedback. This mistake is especially easy to make with members which are otherwise easy to confuse. Clearly, the cardboard model is still indispensable, as one group of students wrote:
The second kind of mistake stemmed from the lack of gravity (which one could say is actually another form of intangibility). (See ill. 4, p. 10.) As some students wrote,
Students found it difficult to manipulate the elements of the model; they found it especially difficult to select individual points, lines, and other elements. This results from the inconsistency between the three-dimensional virtual model and the two-dimensional representation of that model on the screen. We, the instructors, tended to think in terms solely of the three-dimensional model, but in fact what the students saw and manipulated was its two-dimensional representation: the interface mechanism between the user and the virtual model consists of a two-dimensional monitor display and a two-dimensional pointing device.
We propose two approaches to investigate this question of the interface of the virtual model. The first approach is to redesign the existing interface to support the operations that this teaching tool requires. Such an interface would provide features like self-aligning functions, a graphical menu template, and object constraints.
The second, more radical, approach is to transfer the model kit to a virtual reality (VR) simulation environment; we are currently developing the technology for this environment in another project. In the VR environment, students will be able, by means of a head-mounted stereographic display, to enter the three-dimensional virtual space containing the model kit and to manipulate the model parts directly. This would eliminate the problem of manipulating two-dimensional representations of three-dimensional objects. According to our proposed schedule, this approach will be completed by September 1995.
It appears that the students were reacting to the philosophy of the modelling environment, which, by attributes like its characterization of three-dimensional space and the commands it provides, makes some tasks easy and others difficult.
Most students chose variants which could be constructed efficiently and avoided those that could not. Students constructed the "efficient" variants by first forming a single bay and then applying commands like "copy", "array", and "mirror" to form the complete structural frame. These models were generally large and repetitive. (See ill. 5, p. 11.)
Only a few students opted for the "inefficient" variants, which required intensive manipulation of individual components. These models were small, but involved subtle issues, like the structural frame's deviations from orthogonality, for which the Yingzao fashi has explicit provisions. (See ill. 6, p. 12.) (We had omitted these provisions from the assignment for the sake of simplicity.)
Students reacted enthusiastically to this assignment, much more so than we had expected, and were keen to learn more about Chinese architecture. Some of this enthusiasm appeared to be related strictly to the intellectual appeal of the construction system itself.
But the assignment also seems to have affirmed our studentsí identity as Chinese. They were proud and excited to have discovered the logic and beauty of this uniquely Chinese construction system.
We speculate that our students reacted positively to this assignment because of their particular background: ethnically, culturally, and psychically, they are Chinese, but intellectually they are more Western. They were born and brought up in Hong Kong, a city with a population that is 98% Chinese and surrounded by more than a billion Chinese. But at the same time, they have had an essentially Western education, with little specifically Chinese content. They are surprised and pleased to discover (not merely to be told) that Chinese architecture can be understood in ways that stand up to Western analysis.
Our experience with this first in-class use of the teaching tool suggests that it is a double-edged sword. On the one hand, it makes it possible for students to master quickly the conceptual basis of Song construction. On the other hand, it has limitations that we must understand in order to be able to anticipate and avoid error. We plan to continue this project by investigating the question of the representation of the virtual model. We will also expand the model kit to encompass a larger part of the Song construction system.
We are pleased to thank our colleague, Mr. Jeff Kan Wai Tak; our teaching assistant, Mr. Shrinath Tandur; our research assistants, Mr. August Ma Ho Wai and Mr. Andy Chiu Chun Kit; and, of course, the students in Introduction to computer-aided design, fall 1994-95. This project was funded by a UPGC Direct Grant for Research 1993/94 (small project 220 201 810) from the Chinese University of Hong Kong, which we acknowledge with thanks.
/ Top of papers / Top of reading /