Interacting with thousands a parametric-space exploration method in generative design

Abstract

Although generative and parametric design methods open possibilities for working with a large number of solutions, there is almost no computational support for designers to directly manage, sort, filter, and select the generated designs. In this study, we propose an approach that presents a similarity-based design exploration relying on similarity indices that aims to reduce and collapse design space into manageable scales. The similarity indices are calculated either before generating solutions by evaluating the input parameter tuples and functions that produce aspects of designs, or after by looking at the solution features. Our goal is to enable designers to identify the design space of interest by incrementally and iteratively studying the solutions in sets, creating subsets through filtering and sorting. We also demonstrate a visualization of a similarity table that presents different interaction opportunities for system design. The approach requires further integration with existing parametric design tools and development of new interfaces.