[ dmg . research . ArchiDNA ] edit
A design style may be described in terms of the common components of a design and how they are arranged in combination. A design style can be studied in the form of computer algorithms that manipulate the design components. Thus, a computational design tool can be developed to enable the designer to systematically develop a new design style or modify an existing style. This paper concerns the computer generation of various spatial layouts using a set of rules that define a certain design style. It discusses how generative CAD software can be developed that embodies a style and how this software can serve in the architectural design process as a computational design tool. We created a prototype software application, ArchiDNA, to demonstrate the use of rules to specify a design style. ArchiDNA employs a set of rules to follow Peter Eisenman's style in Biocentrum building plan in Frankfurt, Germany. Analyzing Eisenman's 2D and 3D drawings, we implemented a shape generative process in which one shape remains fixed while a second shape is attached to its edges. We call the fixed shape the base-shape and the second (attached) shape the applier-shape. Designers can use any shape as applier-shape by simply selecting it and then applying to the base-shape. The computer duplicates the applier-shape, transforms it, and attaches each duplicates to an edge of the base-shape. ArchiDNA generates 2D and 3D drawings similar to Eisenman's plans for the Biocentrum building. ArchiDNA supports designers in defining rules and generating configurations. Designers can enter and edit their own shapes for applier-shapes and base-shapes and also change the combination of primitive shape operations (translation, rotation, scale, and repetition) that make up the process of applying one shape to another. Thus, designers can define rules and explore their consequences, using ArchiDNA to generate shape configurations that are consistent with a given set of rules of style.
researchers: Doo Young Kwon, Mark Gross, Ellen Do
Last updated 10.19.2009 by Brian R Johnson