Subdivided Pavilions (2005)
Initial 2D and 3D subdivision studies
While much contemporary discussion in algorithmic architecture and generative design centers on agent-based models and complex systems, this project intentionally pivots to a simpler, highly controllable generative approach to produce intricate, heterogeneous forms. This simpler methodology offers enhanced control and predictability, enabling precise refinement through iterative parameter adjustments.
This project marks an early exploration into 3-dimensional subdivision processes. Traditionally used in computer graphics to create smooth forms, these algorithms are here adapted with additional weights. This modification allows the processes to generate forms with fundamentally different attributes, influencing not only their curvature but also their internal structure and surface characteristics..
This project formalizes modifications to these processes, and it applies them to generate a series of architectural pavilions. Each pavilion is conceptually based on two interlinked cubic frames, reminiscent of a tesseract. Critically, the underlying generative process remains identical across all pavilions; only specific parameters, primarily their division weights, are varied. This approach demonstrates the diversity of form achievable from a consistent, simple rule-set.
Initial subdivision tests are based on two-dimensional processes. Several types of subdivision masks are tested, some of which are based on the work of William Floyd. These masks divide each quad into three, four, or six further quads. Quads are note necessarily contiguous.
Minimal Parameters
In every instance, the number of parameters is intentionally minimal, restricted to between three and six division weights that govern the placement of the quads' vertices. The values of these weights is held constant throughout multiple subdivision iterations. There is no use of conditional or boolean logic, nor are random numbers used. The processes thus remain entirely deterministic.
Transition Between Values
The figures on the right show a linear interpolation between two distinct sets of weights. Effectively the final iteration of a form is drawn hundreds of times using transparent strokes as the weights gradually transition from one value to another. It is astounding to see that these extremely simple processes can produce such a large variety of forms.