A Phenotype-Based Representation that Quantifies Aesthetic Variables

Design search should provide designers with diverse yet high-quality options. However, current methods to represent parametric options are mainly genotype- or performance-based. Qualitative values such as composition or aggregate shape are often not considered. Thus, the exploration of options is constrained, as the search algorithm is limited to genotype or performance descriptors. The aim of this paper is to improve existing methods to quantify aesthetic variables, thereby leading to more intelligent forms of design search. The paper proposes a phenotype-based representation that captures key features such as geometrical proportions, the framing of negative spaces and shape compositional relationships. A design space that consists of three different parametric models is introduced to demonstrate how the Zernike moments provide a uniform and consistent method for phenotype representation. The paper also proposes ways to measure phenotype similarity between options. Lastly, strategies to navigate, organize, and bring design logic to the design space are also presented.