This project grew out of a series of experiments made for the 24th Generative Art Conference held in Cagliari in 2021. It is intended to bring generative art working as closely as possible to the actual pixels that are shown on a screen or printed, using computer vision techniques to manipulate them. It was inspired by an exhibition, at the Tate Gallery in 2021, of Turner's works, which allowed me to focus closely on the way Turner researched and developed his techniques. I asked myself what was the equivalent of this research, into colours and papers and brushes, for a generative artist who works with pixels, file formats, and algorithms like those in the OpenCV library, and the Python Imaging Library.
The result is a large and growing toolkit. I have built a 'front end' to identify starting points (usually other images) and then to help me select and apply the different techniques, sometimes in a succession of ways, and to collect and archive the results.
Some of these results are shown on this page. My intention is to offer the higher-quality prints for sale as I develop more confidence in them. At the moment I am building up a portfolio of the ones that I like, ready to start this process.
This project was also inspired by Dubuffet and particularly his 'Texturologies' series, in which he mixed paint with soil from his garden in Vence, where he was painting. This allowed him to create links between the abstract and figurative paintings and the solid material of the scenes he was painting. My equivalent is to use photographic images as starting points, thereby I hope grounding the images in my own life and also in SW London and other areas that matter to me. All these images began as parts of my life, and I have taken pixels from them, signs which I have recombined and nuanced with other similar signs.
A pixel exists in several forms, which is another way of saying that it has no definitve existence. It is not as material as a piece of soil: at its most basic, it is just a mathematical expression; embodied as a temporarily configured block in computer memory, or as a group of illuminated cells flickering on a screen; or as pigment left by a printing process on paper.
Pixels can also access human memory, when a scene evokes recollection of a place or an object. For a moment the memory lives again, perhaps, assembling all the emotions or sounds or smells or intellectual content of an experience in your brain: but this recollection is now no more concrete that the pixels that evoke it. The same image, say of a London landmark, will evoke many different associations for many different people. You and I may each have a memory stirred by the same image, but the wealth of detail and personal vividness that comes with the memory will not be the same.
Playing with image masks opens another exciting dimension: like surreal collages, you can bring fragmentary meanings and associations together to create new shapes, as well as new colours.
As Vera Molnar said, "randomness... will show you billions of possibilities, which you, with your limited imagination, couldn't have thought of. So it enriches the senses. Therefore randomness has a lot of importance to me, but not in the way of dadaism. Its not to say anything can be art. On the contrary, it helps me to better find what I like. Because when you work with intuition, you do ten, twelve, fourteen tests. At the twentieth, you're tired and stop. With computers, you can first open the entire spectrum, and say, 'this is the part that interests me, and not the rest'. So you place the focus, and develop all possibilities within. Afterward, you'll find the interesting part is over here. So you get closer. Its a paradox, but the people who argued at the beginning that using computers dehumanises art, the opposite is true. Because it's thanks to all this technology that we can get very close to what we have imagined, that we might not have found otherwise.
I've now got round to letting the system choose which original image(s) and mask it wants to work from. The first group it came up with I would never have dreamed of putting together, partly because they look so light-hearted.
When I limited it to images of trees, it came up with this. The combination of very similar textures gives an eerie effect that looks almost realistic, but there's an uncanny valley effect as well.
I'm starting to play with shapes as well as colours; this image convolutes with a Sobel matrix filter to accentuate vertical lines, so that an image of a London office building becomes a view of Rouen Cathedral by Monet.
Now exploring the morphological methods on OpenCV. These allow a surprising range of effects, much more interesting than those described in the text books.
One of my pixologics series has just been placed sixth in the 2022 'American Art Awards' section on 'Manipulated Photography: Landscape or Still Life'.
