Visual Character Analysis Within Algorithmic Design, Quantifying Aesthetics Relative To Structural And Geometric Design Criteria
Buildings are responsible for 40% of world C02 emissions and 40% of the world’s raw material consumption. Designing buildings with a reduced material volume is essential to securing a post-carbon built environment and supports a more affordable, accessible architecture. Architecture’s material efficiency is correlated to structural efficiency however, buildings are seldom optimal structures. Architects must resolve several conflicting design criteria that can take precedence over structural concerns, while material-optimization is also impacted from limited means to quantitatively assess aesthetic decisions. Flexible design methods are required that can adapt to diverse constraints and generate filagree material arrangements, currently infeasible to explicitly model. A novel approach to generative topological design is proposed employing a custom multi-agent method that is adaptive to diverse structural conditions and incorporates quantitative analysis of visual formal character. Computer vision methods Gabor filtering, Canny Contouring and others are utilized to evaluate the visual appearance of designs and encode these within quantitative metrics. A matrix of design outcomes for a pavilion are developed to test adaptation to different spatial arrangements. Results are evaluated against visual character, structural, and geometric methods of analysis and demonstrate a limited set of aesthetic design criteria can be correlated with structural and geometric data in a quantitative metric.
Keywords: Generative/Algorithmic Design, Computer Vision, Environmental Performance, Multi-Agent Systems, Visual Character Analysis, Sdg10, Sdg11, Sdg9, Sdg12, Sdg13