Boring to beautiful
Images are just numbers. To be precise, I standardise my square images at 3000 pixels in each direction. That's 9 million individual pixels. Each pixel consists of three numbers, which in turn control three tiny little light emitting cells. These, as most people know, are red, green and blue (R,G,B). By turning these up or down, you can create virtually any colour you like.
Our eyes, of course, also have receptors which specialise in seeing red, green, or blue. You can't actually 'see' any other shades: that beautiful primrose yellow or vivid ochre that you think you see in front of you have been carefully (and instantly) calculated by your brain from the R,G,B values your eyes are feeding back. By supplying your eyes with those same RGB data, a computer screen presents what you think you see in just the same way as direct vision of the original object does.
So the most beautiful colour image in the world, or the ugliest, is just a mass of numbers. 27 million in the case of my images. (9 million cells, each with three numbers.) Imagine, being handed 27 million numbers, and making sense of them instantly and with no effort!
In order to manipulate these numbers so rapidly, the Python language uses an 'add on' library called Numpy. Numpy is a lower level language than Python. It works far more closely at the level of the computer, and is homogeneously typed and more rigid to programme. There are even faster alternatives now, but for my purposes Numpy is adequate.
Numpy manuals are written using strange terms like'broadcasting', designed to confuse struggling artists, and they deal with difficult concepts to visualise, such as slicing and set theory. I hope they will not mind me saying so, but the authors of the Numpy library are not great at explaining it.
Yet after a lot of reading and trial and error I started last week to make something useful from it: a number manipulation algorithm that I have never seen before and which, magically, provides new images out of chaos. The process seems similar to 'magical realism'. Clearly there is photographic realism: I took the underlying image last year during a summer evening walk with a local pagan group, through a riverside area known locally as Arcadia, and full of interesting artistic associations. But then I applied my new Numpy algorithm, and the blind processes of mathematics produced something with new depth and interest.
Playing with these algorithms is a constant source of joy. Many times the results "don't work" - meaning I don't like them. But surprisingly often something comes out that makes me sit up and smile, and add it to my list of images.
