W:Blut

I’ve had the good for­tune of being invited by Corneel Cannaerts to give a pre­sen­ta­tion on the first Share && Tell hosted by dorkbot and timelab.

Not really know­ing what to talk about, I decided to take a gam­ble. What bet­ter way to dis­cuss cre­ative cod­ing than using Processing to build the entire pre­sen­ta­tion. This holds the risk of a crash-and-burn-with-audience. On the other hand, the alter­na­tive would’ve been Powerpoint. For a crowd of design-savvy indi­vid­u­als, a burn­ing wreck is prob­a­bly preferable.

For what it’s worth on its own, I’ve put it online. It’s a demon­stra­tion of the HE_Mesh library built around the theme of sur­face divi­sion and space par­ti­tion­ing. The title Division, The wrong way to draw a grid was inspired by Matt Pearson’s upcom­ing book Generative Art: A Practical Guide Using Processing.

The core idea of the pre­sen­ta­tion is that the tra­di­tional, right way to define a grid, a col­lec­tion of points and explicit con­nec­tiv­ity, doesn’t lend itself well to gen­er­a­tive explo­ration. Modifying a grid while main­tain­ing struc­tural integrity requires exten­sive rule­sets or hefty con­straints. In my expe­ri­ence the code will turn out to be rigid and pre­dictable. In another con­text this might be desir­able. But gen­er­a­tive and para­met­ric algo­rithms dif­fer from tra­di­tional art techniques.They are not meant to exe­cute an idea, rather they should be tools to inspire and give new ideas.

I present two alter­na­tive approaches to a grid. Slides 4 to 10 han­dle the grid as a sur­face or vol­ume divided by pla­nar slices. While we lose some gen­er­al­ity, the advan­tage of this dual rep­re­sen­ta­tion is that it doesn’t break down under per­tur­ba­tion. The grid con­nec­tiv­ity is implicitely defined by the slic­ing oper­a­tions. We are com­pletely free to devise rule­sets with­out hav­ing to worry about integrity issues. The iter­a­tive con­struc­tion tech­nique, the sur­face or vol­ume is divided into a grow­ing col­lec­tion of sub­sur­faces or sub­vol­umes, opens new venues for fur­ther explo­ration by mod­i­fy­ing the impact of each step.

Slides 11 to 16 expose a rather dif­fer­ent approach. Constructs made with more or less global slic­ing oper­a­tions are obvi­ously geo­met­ric and hold lit­tle of the organic. Nature works on a local scale, bottom-up. The Voronoi con­struc­tion of a grid reflects this local­ity. In a sense it’s another dual rep­re­sen­ta­tion, this time defined by a col­lec­tion of con­trol points. The points them­selves can be gen­er­ated by any of a huge num­ber of con­ceiv­able methods:pure math, phys­i­cal par­ti­cle sim­u­la­tion, autonomous agents, exter­nal data … The con­nec­tiv­ity is implicitely defined by the process that gen­er­ates the cells and han­dles noise in a robust way. Voronoi struc­tures can also be eas­ily nested, first gen­er­a­tion cells fur­ther refined by local point con­fig­u­ra­tions, adding another layer of depth. We humans have devel­oped an acute sense of bal­ance between the geo­met­ric and the organic. Voronoi divi­sions come pretty close to the sweet spot.

The image of the Giant’s Causeway in slide 1 was taken from this site. Unfortunately it’s not cred­ited so I can’t prop­erly acknowl­edge the pho­tog­ra­pher. The Causeway is a beau­ti­ful exam­ple of struc­ture cre­ated by divi­sion rather than growth. Philip Ball gives a mar­velous exposé on its ori­gins in the third part, Branches , of his tril­ogy Nature’ Patterns. I heartily rec­om­mend get­ting the entire trilogy.

The pre­sen­ta­tion:
The image takes you to the Processing sketch itself. Use up and down to move from slide to slide. The slides typ­i­cally respond to a mouse-click and/or the space­bar. Imagine some more or less coher­ent bab­ble, invite some good friends over and you can relive the Share && Tell expe­ri­ence right there and then in your own liv­ing room!