Saturday, November 15, 2008

Sandbox: Cylinders

Friday, November 14, 2008

Wednesday, November 12, 2008

Sandbox: More Voronoi Excercises

If You Give Enough Monkeys with Typewriters Enough Time...

...one of them would eventually type Hamlet, the infinite monkey theorem says.

I would add that if you mess with genetic-algorithms-powered logo turtle long enough, sooner or later you'll see the shapes which resemble organic forms, like spiders or ants. :-)

Monday, November 10, 2008

Sandbox: When Hell Freezes Over

Sandbox: Dante's Inferno

Sandbox: Spirals, Always Great Fun

It doesn't take long, if you mess around with the logo turtle, to start thinking about ammonites.

Really Amazed

I was really amazed when I observed various visual patterns emerging in front of my eyes, all developed from extremely simple math formulas. I unintentionally visited the world of fractals yet again.

Sunday, November 9, 2008

Sandbox: Captain Nemo

Yeah, it IS yellow.

Saturday, November 8, 2008

Logo Turtle Introduced

To answer the problem of trivial morphonegesis because of the direct connection between the genotype and the phenotype, I came up with the idea of using the Logo programming language, or a "logo turtle".

The genotype would consist of logo instructions for the turtle. The recombination of subjects would lead to new and modified programs. The phenotype would be generated when the logo program is run and the poins are depolyed by a logo turtle.

I explored this possibility in the 2D world first. Later I will use this approach to deploy Voronoi input points.

Wednesday, November 5, 2008

Big problem

I came across a big problem in my genetic algorithms. The traits of the subjects (phenotype) are not passed on very well to the next generation.

Each generation consists of numerous subjects. After I select some of them and the recombination occurs, the next generation has subjects which are totally different from their ancestors.

The recombination of two subjects leads to a completely different subject, which is not the desired behaviour.

I studied the problem and came to the conclusion that this is because the connection between the genotype and the phenotype is much too direct. In my current design, the genotype is defined by input points. The phenotype is a very simple function of the genotype. The recombination of genotypes produces a very trivial combination of phenotypes; like tearing two buildings apart and combining their parts without any grace.

All possible architecture properties such as rhythm, proportions, emphasis... are lost when the next generation is produced.

Tuesday, October 28, 2008

To Err is Human

As I've already found out, inducing some error when distributing the input Voronoi points adds a little paprika to the design.

Saturday, October 25, 2008

Sandbox: Back to Basics

Let's deepen the understanding of how the distribution of the Voronoi input points affects the object's morphology.

Thursday, October 16, 2008

Inside

Tuesday, October 7, 2008

Sandbox: Dubious Furniture

Furniture, chairs esp, has always been architecture's pet project.

Thursday, October 2, 2008

Flat or Curved?