This paper presents a new software tool for visibility-based analysis of architectural (interior) spaces. It combines two- and three-dimensional geometric, topological, and image-based methods and computes a large number of quantitative measures to investigate the relationship between geometric shape, the visual appearance of space, and human experience and behavior.

Author: Christoph Opperer

6FMA (6th International Symposium Formal Methods in Architecture)
24.05.2022 – 27.05.2022
A Coruña, Spain

The relationship between visuospatial properties, psychological responses, and behaviors is among the most important considerations in architectural design. Visibility-based analysis methods, such as isovist and visibility graph analysis, are one way to explore and understand these relationships.

This paper presents a new software tool for visibility-based analysis of architectural (interior) spaces. It combines two- and three-dimensional geometric, topological, and image-based methods and computes a large number of quantitative measures to investigate the relationship between geometric shape, the visual appearance of space, and human experience and behavior. Important visual surface properties such as light, surface color, texture, etc., are taken into account by using custom-coded raytracing techniques, lightmap and texture-baking techniques, as well as saliency-detection models from the field of computer vision.

The software was developed as a framework of tools allowing the computation of axial lines, spatial partitions such as s- and e-partitions, sufficient sets of points and isovists, movement patterns based on agent-based models, isovists, visibility graphs, and spherical panoramic images. The reasoning behind this was that it should be (1) simple and straightforward to use in architectural practice and design, and (2) provide as many features as possible to be used extensively in research. One key aspect of the development was that the software can be used in the early design phase to automatically generate, analyze, and optimize a large number of different design variants which is possible due to its open framework of tools and its computational efficiency.

Instead of developing an entirely new program, it was implemented in SideFX Houdini, which is known for its procedural geometry generation, data manipulation, and simulation capabilities. As a result, most algorithms are fully multi-threaded and are written either in VEX, Houdini’s internal high-performance scripting language that runs on the CPU, or in OpenCL to take full advantage of today’s GPU processing power. To ensure maximum flexibility and customizability, only a few algorithms were written in C++ using the Houdini Development Kit (HDK) for performance reasons. Researchers who wish to implement custom algorithms have full access to all underlying geometric and numerical data, such as isovists, visibility graphs, occluding/solid edges, etc. Those who do not want to deal with the inner workings can simply select the measures to be calculated in the user interface.