Another Visual Experiment
The Evolutionary Art of Process-Painting
In my last piece, I went through a simple design process that I use to create interesting images out of pure noise. This time I went even further. I got an image of 256 pixels by 256 pixels of random black and white pixels from the ANU Quantum Random Numbers Server. This is statistically the “purest” noise that I could find. Here is what I got:
Next, I did the same thing as before. I copied the noise as a second layer, did a pointillize operation on that, then merged the layers, adjusting the transparency:
What I did next was add another layer of the original pure noise, and did a pixelate operation on that, merged the two layers with some transparency:
The trick is that this time I use the “difference” function on the new pixelated noise layer. That’s what made the interesting shapes, it’s the negative mutual exclusion of points and pixels (pointillize vs. pixelate).
That’s the point of all of this. The idea is to begin with pure noise and in a short number of steps, of operations affected ON the pure procedural noise, to come up with interesting patterns.
The idea behind this method is an Evolutionary Art System. By iterating in this way, going back and forth, adding layers and merging them, doing primitive operations on procedural noise, what I am in effect doing is functioning as a “Generative” system. I am producing art in an iterated manner, advancing through experimentation.
It’s in fact the way that I go about doing my Sound Design, it’s just been applied to the production of digital images. I know that I’m not a master graphic artist, I’m a visual artist, a painter, but not a digital artist, but still I think that this design methodology is valid. Not only that, I think it can be done entirely by numerical computation alone. I will soon be proving that.
All that I needed to do was prove the concept, for now. In three easy steps, I went from pure procedural noise to something “interesting” to the eye. That’s my proof of concept. If you find the final image even remotely interesting, then understand that I did it purely by following a procedural method that could be automated by a computer program.
The idea is that the pure procedural noise is as “random” as possible. You can’t really get more random than that. It’s a Quantum Random Numbers Server, after all. The random noise, or random matrix (because that’s what it is), is not all that “interesting”. It’s just random. It’s in a state, however, that cannot be compressed. The shortest description possible is the specification of every last pixel in the image. It has a high Kolmogorov complexity, high minimum description length (descriptive complexity).
What I have done is add some “order” but not in a top-down manner. I let the order “emerge” from the pure noise. I felt that this was ideal, much better than me trying to “decide” what the final picture should look like. I let the noise decide for me. I merely “modulated” the original noise signal.
This is all part of a greater project which is the Noise Field Theory that I have been developing for the greater part of the last year. What I want to move towards is a signal-based understanding of the universe, a signal-based perspective on things. I believe that what I call “noise fields” are ubiquitous and pervasive in the physical universe. All that we have to do is “see” them, become aware of them. Everything essentially is a “modulated noise field”. I hope to prove that in future posts on this site.
Edit/Note: Thanks everyone for your appreciation, recommendations etc. It really means a great deal.