Signal Science
Published in

Signal Science

Noise Field Theory Redux

Visual snow. Source: Unknown.

Noise Field Theory is something I developed over a good 30 years of working with noise in various forms, visual and auditory. I’ve been studying noise for almost 30 years. It started with electric guitars and tape recorders and then continued with computers, from computer-asssisted sound design to digital synthesis methods in the visual mode. I studied noise using digital photography and painting. Noise Field Theory is just the outcome of all those years of experimentation with noise in all its forms.

The idea is that everything is fundamentally composed of noise, of what I call “noise fields”. That is, at the quantum level, you have noise fields, which are like random matrices. They are “probabilistic”, meaning that they take on probabilistic values, or values probabilistically if you will. Nothing is set in stone until you do the measuring. At the quantum level, I see the vacuum similar to the image of “visual snow” that I posted above. It’s a noise field.

Random Black & White Pixels. 512 px by 512 px.

A noise field is just a fancy way to say a random matrix, or a m by n matrix with randomly chosen values. Essentially, they are noisy distributions. There are many kinds of noisy distributions. There is Gaussian white noise, there are Poisson distributions, etc. My idea was that everything was made up of superpositions of various noise fields. Anything physical was at the source, at its most fundamental level, made up of noise fields, of superpositions of various vacuum states. Similar to how complex waveforms can be decomposed into n number of pure sine waves, I believed that any complex “form” could be composed of n number of noise fields. That’s how I came up with Noise Field Theory, in a science I called “Signal Science”.

So everything was merely some form of noisy distribution, however complex, with superpositions and so on. Everything is just a “modulated noise field”. That’s it, noise fields get modulated, and they create other distributions, from simple to complex. The idea was to come up with a “Universal Synthesis Engine” (“U.S.E.”) in which you could just pump in various distributions and get various forms, physical (in 3D through 3D printing), visual, auditory, etc.

“Signal Science” was just about the idea of a signal-based view of everything in the universe. The noise field, in other words, is just a “signal”, it just happens to be a noisy signal, a random signal, but it’s a signal nonetheless.

In essence, anything can be composed of superpositions of noise fields, which is what I call “modulated noise fields”. I have been making “noise fields” in the form of visual digital art, through noise synthesis, for many years. What I do is “modulate the noise field”. I start with a noisy distribution, usually Gaussian white noise, and I “modulate” it, that is, I use various basic functions in image processing, to slowly change the form of the noise field, to try to get it to take on a quality I call “visual interestingness” or “VI”.

Part of this has to do with randomness and compressibility. They say that pure white noise, a purely random signal, cannot be compressed. Like in the animated GIF posted at the beginning of this article, noise fields can evolve “over time” as well. Part of my inspiration for Noise Field Theory has to do with what I called “historionic processes”. They are processes-with-a-history, any process with a history, and are modeled on crack propagation or the wear and tear of the “weathering of the elements”, like the “patina” that forms on old objects, or that naturally forms on bronze and so forth. It’s an ageing process, and to me it involves a form of randomness, like the propagation of cracks in a vase or whatnot. These are random, and pure randomness, in any given noise field or other signal, is incompressible.

Noise field as sequence of 2D matrices over time. A.G. © 2022.

You could think of noise fields as a sequence of 2D “random” matrices over time. That’s how I believe things are composed in the universe, by the modulation of noise fields over time, starting with “seed” noise, just like when I compose images of “modulated” noise fields through image synthesis. The noise “signal” is just random fluctuations over time. In sound, any sound is really just variations in air pressure. Noise just happens to be random fluctuations in air pressure. The distribution of the fluctuations over time are random. That’s all there really is to it. It’s not any more complicated than that.

Historionic process evolving over time. A.G. © 2022. All Rights Reserved.

In essence, a noise field, as a process, starts with what I call a “noisebit” or “n-bit”. In essence, the simple formula is “x = n-bit (process starts) + n-bit propagates as historionic process (line).” As you can see in the image, there is a point marked “x” where the line starts, the “process”. That is a noisebit, and a noisebit is just like an “error” a random “error” in spacetime. I don’t know how else to describe it. You have to think of the universe in a sense as discretized, as a mathematical universe, as a digtal simulation, and a noisebit is just a random bit, a random 0 or 1, and bits can also “flip” randomly.

A historionic process (historion) is like an n-path. In my writings, I often use terms like “Noisons” or “Nixels”, “Noxels”. A 1-bit noisebit can start a process that can be disastrous. It can cause a great, and grave, perturbation. All it takes is a 1-bit noisebit, the point “x” where the “crack” begins propagating. All things in the universe, I believe, are founded on random processes of this kind. Even though things take on many recognizable shapes, in essence, at the foundational level, they are just errors, they are temporary “aberrations” in the grand scheme of things, at the universal scale.

All lines lead to here. A.G. © 2022. All Rights Reserved.

All lines lead to here. All previous processes have led to this point. Historiomics is the science of population histories. Historiomics studies all processes that have histories. When I generate my noise fields (images, modulated noise fields), I generate populations of images, and I choose the most visually interesting ones using a fitness function. Whatever images “pass” through the fitness function move onto the next “generation”.

The World — Machine printing Historiomes. A.G. © 2015–2022. All Rights Reserved.

A historiome is like a print-out of a given history, at a given moment in time, the various population histories that make up a historical landscape. It is like a 3D surface evolving over time. The historiome is made up of noise fields. N-bits evolve into historionic processes that make up the given Historiome. Historiomics is the study of historiomes, like genome or epigenomes, but of historical processes, or processes-with-a-history.

ΔW(x); Or, delta-workspace. A.G. © 2022. All Rights Reserved.

Imagine a workspace evolving over time, as a 4-dimensional surface, a 3D surface evolving over time. It’s a 4D signal. “The work W done by a constant force of magnitude F on a point that moves a displacement s in a straight line in the direction of the force is the product: W=Fs”. Work transfers energy from one place to another, or one form to another. Every space is a workspace. The delta-workspace is just the changes-in-the-workspace-over-time.

This all comes back down to the fundamental units, the noise field and the n-bit or noisebit. All processes are derived from it, all processes have histories, all processes are signals, and all signals are probability distributions. It comes down to the same thing: everything is a noise field, first and foremost.

The Historiotopia. A.G. © 2022. All Rights Reserved.

The delta-workspace is a historiotope or historiotopia. It is a 3D surface evolving over time. Again, all comes down to the fundamentals, the noise fields that generate all matter and vacuum space, at the quantum level. It’s the evolution of a random process, developing a probability distribution. Everything is merely a given probability distribution in its simplest form. Regularities in a signal make it compressible. Randomness makes it incompressible. It’s about something’s information content. The noise field at the heart of everything has maximum entropy. Entropy is what forces evolution in a given direction, the evolution of any given process. The 1-bit noise-bit or n-bit is what sets off the process, the crack propagation of sorts.

All lines lead to here. Everything is a signal, everything is a noise field. A signal is any variation of a medium that conveys information. A signal may also be defined as an observable change in a quality such as quantity. Every signal is a noise field, a given noise distribution, or superposition thereof. Imagine noise fields as the clay that make up objects. It is the basic historiotype.

Nota bene:
My noise fields are more or less conceptualized as scalar fields, I must add. Here is an example of a scalar field:

A scalar field such as temperature or pressure, where intensity of the field is represented by different hues of colors.

The scalar field, in 2D in this case, has a magnitude or quantity at every point in space (a “scalar value”). My noise fields are similar, they have a pixel value at every point, which are randomly chosen, so the function could be written as: N(x,y). A noisebit at every point in space, in the x and y axes.

A.G. © 2022. All Rights Reserved.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store