Breed – Driessens & Verstappen: Evolutional Diffusion Lattices
Tuesday, 20 April 2010
Breed – Driessens & Verstappen
If the recent history of digital fabbing were overlayed onto the geological era of Earth, Erwin Driessens, and Maria Verstappen ‘s Breed sculpture may have evolved somewhere in the Triassic Period. This piece, an elegant crystalline cubic sculpture, explored in a variety of materials, appears in its initial incarnation as early as 1995 in manually constructed plywood. It was subsequently produced in 2001, in nylon, using the Selected Laser Sintering technique.
Using evolutionary algorithms to simulate the process of cell division, Erwin and Maria’s program recursively removes material from an original block at progressively greater levels of detail. The process involved in determining which cells are left in the structure and which ones removed is not unlike the process of a simple Cellular Automata, albeit in three-dimensions:
‘One initial cell, a cube, engenders throughout successive stages of cell division a complex, multi-cellular “body”. Morphogenetic rules determine how the division of a cell occurs, dependent on its situation between the cells surrounding it. Every potential situation has a separate rule, so a cell surrounded on all sides by other cells may divide differently from a cell that only has a neighbour on the left and underneath, or a cell with nothing at all in the vicinity, etc’
An interesting post procedural aspect to the creation of the sculptural lattices is the ability to search for shape “coherence” in order to find “fitter” models, i.e. models that would not collapse under the effect of gravity in the real world:
‘The simplest method (two-membered evolution strategy) is already effective: take a randomly composed genotype as base, generate the phenotype and test it for fitness; mutate the base genotype, generate the phenotype and test it for fitness; compare both results with each other, and take the result with the highest fitness as the new base. Repeat the mutations until the result satisfies the stated requirements’
The most recent versions of the piece, Breed 1.2, were produced in 2007 in printed metal. Looking at these cuboid multi-forms its not difficult to recognise the all to familiar frozen geometric patterns of Cellular Automata and the self-organising structures that emerge out of Activator-Inhibitor systems. One specific comparison that should not go unnoticed is the obvious similarity between these models and real world Hopper crystals such as those of Bismuth. Bismuth crystals grow in this shape because the edges of growing structure generate a higher electrical charge activating crystal growth to a higher degree than in the centre of the sides. The program behind Breed essentially works in a similar way reversed – activating removal of cells depending on their mutual arrangement and proximity within a system defined by simple rules.