Refining a computational fractal method of analysis: testing Bovill's architectural data

In 1996 Bovill applied Mandelbrot's fractal method for calculating the approximate' visual complexity of images to architecture. This method is one of only a limited number of quantifiable approaches to provide a. measure bf the relative complexity of an architectural fonil. However, the method has rarely been tested despite many scholars uncritically repeating Bovill's conclusions. While Bovill's original work was calculated manmilly, a software program, Archimage, is presently being developed by the authors as a tool to assist architectural designers and researchers to understand the visual complexity of building designs. The present research returns to Bovill's original architectural data (elevations of famous buildings) and re-calculates the results published therein using Archimage and the commercial software Benoit. These results are then compvared with those produced by Bovill (1996) and Lorenz (2003), to determine if any consistency can be found between the sets. The level of consistency will assist in determining the validity of Bovill's method and provide important data in the ongoing process to refine the Arch image software and the analytical method.