Creative Evolutionary Design System

A hold on creativity

The aim of this project is to implement a Creative Evolutionary Design System, as described by P. J. Bentley and U. O’Reilly in their paper Ten steps to make a perfect creative evolutionary design system (2001). The program seeks to develop a workflow in line with the one proposed in this paper, nevertheless spiced by personal interpretation and design intuition. For this reason, the main line of action runs parallel to the ten suggested steps, and the most relevant ones will be quoted and further explained.

This program is about creativity. It deliberately excludes any kind of objective hardwired fitness measures. It is conceived as a generic experiment to explore the vast search space defined by the project’s degrees of freedom and design constraints. For this reason, an abstract topic is chosen, a one that serves no particular function. The generated Individuals respond properly to subjective evaluation, but are not entirely open to objective measurements. The program can be understood as a Master explorer and a squadron of rangers, each one of them travelling in different directions and coming back with new geographical information, unknwon materials and exotic species. For the next exploration, the master can choose to concentrate a greater number of explorers in  a specific area, exploring it in more detail and thus drawing his map with greater accuracy.

2.jpg

The possibility of evaluation and optimization of certain features is nonetheless not ruled out. A graphical measure of curvature, solar incidence or area is available to the user in an intuitive and visual way. That means that any of these parameters are potentially subject to optimization by a ‘classical’ GA that could be run at convenience at some point in the design process. Both categories are not mutually exclusive but complementary. 

A number of mathematical transformations are applied to the periodic functions of sinus and cosinus. The parametric definition of a sphere is the starting point. 

x = sin(v)*cos(u)  |   y = sin(v)*sin(u)  |  z = cos(v) 

O.png

Every transformation is well described in Choma, J. 2015: Morphing. A Guide to Mathematical Transformations for Architects and Designers. Drawing from these, a ‘surface genome’ is designed which encompasses a pair of chromosomes, each one composed of 10 genes that represent 14 mathematical transformations, which in turn result in 14 phenotypical features. This diploid organismresults into greater phenotypical complexity. By providing two sets of information that can be decoded the same way, two sets of features can be implemented and combined. In this case, the first chromosome defines the basis geometry and the second provides a three dimensional texture mapping for the skin, applied onto the surface using the previously calculated normals and solar incidence (in turn calculated with the normlas and an estimation of changes in concavity) as an attractor.