In an often quoted sentence of his 1976 book “The Architecture of Form”, Lionel March drew a clear distinction between science, interested in extant forms, and design, which initiates novel forms. The theories, methods, measures and diagrams of space syntax have often developed following this first more scientific scheme, and they have been concerned with the analysis of existing or projected buildings and cities.This emphasis on analysis is evident in current software, algorithms and measures. But is it possible to think of a space syntax not only as a way of analysing existing situations or validating future designs, but as a form of actually generating architecture?
In our work we have used space syntax at the early stages of the design process, not so much as a form of analysis, but as a sort of architectural diagram. The shift of space syntax into a generative role has demanded a set of conceptual and technical adjustments: from the emphasis on graphic language and visualisation to the need for fast feedback and interaction.
In this paper we present an example from our work, and the framework (technical and methodological) necessary to produce it. The digital diagram we have created deals with the design of a new hospital ward. It represents some basic problems we have encountered in the relation of patients, staff and architecture, which are incorporated into the software through 3 basic interactive entities: isovists (from patients positions), the circulation paths of hospital staff (with a calculation of their visibility relations to the patients), and the arrangement of walls to form rooms. All these 3 entity types are interdependent: isovists depend of walls and positions and the visualisation of staff paths depends on the patients isovists. They are also editable in real time, that is, walls, isovists and paths can be added,deleted, or moved, and the effects of any of these actions visualised at once.
This fast interaction and feedback require efficient algorithms and data structures. In particular we have implemented an algorithm for the calculation of isovists or visibility polygons with a complexity dependent of the size (in terms of visible vertices) of the visibility polygon, rather than being a function of the size of the boundary. This allow us to calculate visibility polygons in real time irrespective of the size of the boundary, may this be a building or a whole city. Our method implements an idea by Åsmund Izaki for the calculation of isovists and visibility graphs, based in the use of an underlying triangulation data structure for the search of all visible vertices from a point. Besides the general interest of our approach to the use of space syntax in a generative rather than in an analytical way, we believe that the algorithms for the calculation of visibility polygons or isovists can find application also into existing space syntax software, improving its performance, and in some cases opening the possibility for an extension of its role from forms of analysis to generative ones.
Our software has been developed using the C++ programming language, and it makes extensive use of Open Source libraries such as CGAL, Dime, Qt and Boost.
Seoul: Sejong University Press , 2013. 062: 1-062: 13 p.
Isovist, Visibility Polygon, Low Complexity, Algorithm, Interactive Software, Design Application, Paths.