Orthogonal Composition Generator in GH
Continuing my line of research into automated morphogenesis, I decided to create a new “form generator” in Grasshopper, this time limiting myself to purely orthogonal geometries. This self-imposed restriction appealed to me for a number of reasons.
Firstly, a technical one: limiting my geometries to boxes means that grasshopper can handle more operations without failing – there are no trims or complex curves to worry about. I can also do things like thin and thicken volumes that I couldn’t do without significant distortion if I were working with more complex geometry.
Secondly, I’m interested in the concept of an “aesthetically restricted medium,” as discussed by Douglas Hofstadter in “Le Ton Beau de Marot,” a marvelous book on poetry, translation, and creativity that has had a large impact on me. Hofstadter holds that the imposition of a restricted medium – a particular meter, register, or language in the case of poetry, for instance – is often a more productive territory for creative thought than the wide open field, free of restrictions.
Thirdly, there is a long heritage of architectural composition that is limited to the XYZ planes. From Van Doesburg to Rietveld to Eisenman (to name just a few) architects have explored the expressive possibility of planes, volumes, and lines oriented to the cartesian spatial grid. Even today a large percentage of the so-called “dwell modernism” is limited to more or less orthogonal geometries, and I will admit a certain visceral partiality to this (easily reproducible) aesthetic, even while I appreciate the more flexible and unique geometries of Thom Mayne, Zaha Hadid, Neil Denari, Daniel Liebeskind, and so on.
Technically speaking, the large algorithm that generated all of these forms is an (arbitrary) sequence of formal operations operating on BReps. Each operation (create random plane surfaces, thicken all planes uniformly, thicken a random subset of planes into volumes, split volumes by planes, to give a few examples) takes BReps as both input and output, allowing me to link them in an arbitrary sequence. The final step separates the generated volumes into four groups that constitute varying proportions of the total set, and applies four different surface conditions to those pieces. It’s essentially a brute-force attack on the problem of 3-dimensional composition, allowing a designer work in an evaluative rather than generative mode.
The surface conditions were modeled after a particular axon of Eisenman’s House X, which was perhaps slightly tongue-in-cheek, given Eisenman’s stated disdain for parametric methods. While the algorithm I produced was by no means reproducing the process of his early “syntactic” explorations of form, there is a certain kinship in the application of rigorous geometric processes, independent of function and semantics.
It’s important to me that parametric process be recognized as independent from any particular aesthetic or style. Because novel modeling tools enable complex, curvaceous geometries, mass customization, voronoi, progressive differentiation, and every other techno-fetish you can think of, it’s easy for the critics of computation to dismiss it on the grounds of this very fetishization.
See all 200 variations produced by the first run of the algorithm
Filed under: Architecture, Grasshopper | 11 Comments