Saturday, September 6, 2008

Algorithms for Architectonics

More examination of the generation of architecture through algorithms.

The most convincing algorithmic processes--meaning, that they generate shapes that most resemble architectural forms--with which I have yet experimented seem to be composed of sheets of a plywood-like material along with various rods and columns that seem to be proportionally similar to wood framing or steel-tube frames. But of course this resemblance to architectural form, or even to real physical objects, breaks down on close examination. It's easy to find bits of the generated model that behave in ways that building components or even real objects cannot: at best they defy the possibility of gravity, while at other times they project through or intersect with neighboring forms in a fashion that solid matter (and building components) cannot.

Now, with careful tweaking of the set of operations that constitute these algorithms, I could conceivably create a situation where those flaws did not occur, and where the generated shapes even included inhabitable space. Of course, it is possible that by so doing I am adding an entirely pointless iteration to the design process: I no longer design the architecture itself, but instead I design a set of formulas which themselves design the architecture the way I would have without the mechanism of the generative algorithm. If there is a difference between the architecture I would design and what the algorithm designs, the difference is there because either I made mistakes in designing the algorithm, or else the limitations of the system employing the algorithm are such that it cannot yet completely simulate my design process.

Of course, the preceding assumes that the design of architecture is in fact a procedural process (i.e., a mathematical algorithm) itself. I'm relatively certain that I do not agree with that assumption.

It has occurred to me that my concerns with the algorithmic generation of architecture (1-the result is not realizable without extreme effort; 2) the process includes a pointless iteration; 3) the process assumes
architectural design is equivalent to math) could be rendered moot if I used the technology in a completely different manner, as already suggested in a previous post.

I am proposing that this generative technique would be far more useful as part of an architectural cognate for
decalcomania or similar techniques for creating suggestive forms that, following elaboration or subtractive editing, would not otherwise be conceivable. These painting techniques--popularized by such Surrealist artists as Max Ernst--were belated rediscoveries or implementations of an inventive device originally recommended by Leonardo da Vinci in the treatise A Way of Stimulating and Arousing the Mind to Various Inventions, where he advised that one should “look at a wall spotted with stains, or with a mixture of different kinds of stones; if you have to invent some scene, you may discover a similarity with different kinds of landscapes, embellished with mountains, rivers, rocks, trees, plains, wide valleys and hills in a varied arrangement: or, again, you may see battles and figures in action or strange faces and costumes, and an endless variety of objects which you could reduce to complete and well-drawn forms.” I believe that the key is to begin with a complex field of visual information of limited organization and a restricted range of components; the human mind will begin to edit this limited chaos in such a way to percieve a meaningful shape pattern that is/was not there before that act of perception.*

It seems to me that algorithmic generation of three-dimensional forms could satisfy that requirement for that complex-field-of-many-objects-of-a-limited-range-of-types in a way particularly suitable for the creation of architectural forms if the components generated by the algorithm are reminiscent of the typical forms that compose architecture. In other words: 1) I create a formula that generates shapes in a unusual pattern (an act of design or chance); 2) I observe the generated assemblage and allow my meaning-seeking mind to find something there other than mathematically-prescribed alignments; 3) I elaborate, delete, simplify and otherwise edit the assemblage until the meaningful shape that was described by that vision (or delusion) is all that remains of the original math-generated assemblage.

*I should admit that I have only found quotations from this work of Da Vinci's in other essays and books. I do intend to find a full copy as soon as I can.