Casey Reas — Code as Canvas, Software as Art
Casey Reas co-created Processing and p5.js — the open-source tools that put creative coding in the hands of millions. But Reas is equally an artist whose conceptual, software-driven practice asks a deceptively simple question: what happens when you write instructions for a machine to make images?
From Troy, Ohio to MIT Media Lab
Casey Edwin Barker Reas was born in 1972 in Troy, Ohio. He studied design at the University of Cincinnati, where he was a member of the band nancy alongside Scott Devendorf and Matt Berninger — the pair who would go on to found The National. After graduating, Reas spent two years writing software and building electronics as a purely artistic exploration before returning to academia.
In 2001 he earned a Master of Science in Media Arts and Sciences from MIT's Media Lab, where he was part of the Aesthetics and Computation Group led by John Maeda. It was here, alongside fellow student Ben Fry, that the idea for Processing was born.
In 2003 Reas moved to Los Angeles, where he is a Professor in the Department of Design Media Arts at UCLA. Students in his programme have included Refik Anadol — whose own practice of AI-driven, data-scale installations is directly rooted in the software thinking Reas teaches.
Processing — A Language Born in the Sketchbook
In 2001 Reas and Ben Fry released the first version of Processing, a free, open-source programming language and environment built on Java. The explicit goal was to give visual artists and designers a way to learn programming without the ceremony of enterprise Java — and to give programmers a direct path into visual output.
Processing introduced the concept of the sketch: a lightweight, self-contained program that could be written, run, and iterated in minutes. The two core functions — setup() and draw() — remain familiar to any creative coder today. setup() runs once on start; draw() loops continuously, creating the animation loop.
// A minimal Processing sketch — organic line growth
int x, y;
void setup() {
size(800, 600);
background(10);
stroke(255, 80);
x = width / 2;
y = height / 2;
}
void draw() {
x += int(random(-5, 5));
y += int(random(-5, 5));
x = constrain(x, 0, width);
y = constrain(y, 0, height);
point(x, y);
}The language spread rapidly through art schools and design programmes worldwide. By the mid-2000s it had become the lingua franca of creative coding. Today Processing is used in thousands of classrooms and has spawned implementations in Python (Processing.py), Android (Processing for Android), and web environments.
p5.js — Processing for the Web
In 2013, Lauren Lee McCarthy began developing p5.js with support from Reas and the Processing Foundation. p5.js is a JavaScript library that re-implements Processing's core philosophy natively in the browser — no Java, no plugin, no install. The library runs wherever JavaScript runs, which means the web, Node.js, and any environment with a Canvas or WebGL context.
The API deliberately mirrors Processing. setup() and draw() work the same way. Colour functions, geometry primitives, and the mouse / keyboard event model are all familiar. What changes is the runtime: p5.js sketches can be embedded in any webpage, shared as a URL, and remixed live in editors like the p5.js Web Editor.
// A minimal p5.js sketch — the same organic line in JavaScript
let x, y;
function setup() {
createCanvas(800, 600);
background(10);
stroke(255, 80);
x = width / 2;
y = height / 2;
}
function draw() {
x += random(-5, 5);
y += random(-5, 5);
x = constrain(x, 0, width);
y = constrain(y, 0, height);
point(x, y);
}p5.js also extends beyond the canvas. The library provides helpers for working with HTML elements, audio (p5.sound), video, and WebGL 3D — making it a complete creative environment, not just a drawing library. The Processing Foundation officially adopted p5.js as the web-facing version of Processing in 2014.
Artistic Practice — Writing Instructions for Images
Reas is equally a practising artist as he is a tool-builder. His artistic work centres on a single, enduring question: what does it mean to write a software instruction that generates a visual form? Each project is an exploration of how simple rules, when executed computationally, give rise to complex, often unpredictable behaviour — connecting his practice to traditions of conceptual art, systems art, and process art.
His works are typically presented in multiple forms simultaneously: natural-language descriptions of a process, machine-code implementations, and printed or screen-based outputs. The same underlying instruction can produce a static print, an animated software work, and a large-format installation — each rendering illuminating a different facet of the same process.
"Software is not a tool. Software is a medium — like oil paint or marble. When I write code, I'm not making a tool to make an image; I'm making the image."
Since 2012 Reas has also incorporated broadcast images — footage from television, social media, and the news cycle — into his work, algorithmically distorting them until they retain only traces of their representational origin. This strand of the practice interrogates how images circulate, accumulate, and lose legibility in a media-saturated culture.
In 2020 Reas co-founded fxhash — wait, that is Genie Szabo. Reas instead co-founded Feral File, a platform for exhibiting and collecting internet and digital art as NFTs, with curated exhibitions presented at the intersection of artistic integrity and digital ownership.
Selected Works
Software Structures (2004 — Whitney Museum artport, New York)
Commissioned by the Whitney Museum of American Art's online gallery artport, Software Structures is one of Reas' most discussed early works. It takes Sol LeWitt's wall-drawing instructions — written in plain English and realised by others — and reimplements them as generative software. The project makes explicit the connection between conceptual art's use of instruction-based practice and the computational logic of a computer program.
Each structure is defined as a written process; the software then executes that process continuously, producing an ever-changing but always rule-governed image. The work is freely accessible online and remains one of the clearest demonstrations of Reas' central idea: that software and conceptual art share the same DNA.
Process Compendium (2004–2010)
The Process series — collected and published in the Process Compendium (REAS Studio, 2010) — is the definitive statement of Reas' early artistic practice. Each work begins with a short written description of a behaviour, often involving simulated biological entities: neurons, organisms, surfaces. That description is then implemented in software and translated into a range of media including print, screen, and large-scale installation.
Process 13, for instance, is held in the permanent collection of the Centre Pompidou in Paris. Process 18 (Software 3) is in the collection of the Victoria & Albert Museum in London. The series established Reas internationally and demonstrated that generative software could produce work of museum-quality ambition.
CENTURY (2012)
CENTURY situates generative software within the history of twentieth-century abstraction. Each work in the series takes the structural concerns of modernist painters — composition, colour relationships, the grid — and re-animates them within computational logic. The results feel simultaneously historical and alive: colour fields that shift according to rule-based systems, compositions that evolve within strict structural constraints.
The series is a direct argument that generative art is not a break from the Western painting tradition but a continuation of its most systematic impulses — from Albers' colour studies to Sol LeWitt's instructions.
Ultraconcentrated (2012–ongoing)
Ultraconcentrated transforms source material from television broadcasts, newspapers, and social networks by compressing and layering them algorithmically until legibility collapses into dense, abstract fields of colour and texture. The work explores what happens to an image when its information is retained but its form is destroyed — what remains when meaning is compressed past the point of reading.
The series coincided with Reas' growing interest in broadcast culture and the sheer volume of images produced and discarded daily by contemporary media. By rendering those images as dense, indecipherable matter, the work makes their accumulation visible as a physical weight.
In Silico (2018)
In Silico marks Reas' turn toward machine learning as an artistic medium. Using generative adversarial networks (GANs) trained on botanical imagery, the series produces images that sit at the threshold between plant and not-plant — organic forms that are simultaneously recognisable and fundamentally synthetic. The title (Latin for "in silicon") positions the computer not as a tool but as a site: a place where new kinds of matter can exist.
The work was a significant shift from writing code line by line to working with a learned system whose behaviour must be guided, shaped, and curated rather than explicitly programmed.
Compressed Cinema (2020)
Compressed Cinema applies machine learning to feature films, compressing entire movies into compact, shifting audiovisual fields in which all frames are present simultaneously. The result is neither cinema nor painting but something between: a temporal image in which the film's entire visual history coexists, layered and animated, in a single frame.
Reas has described the approach as a form of "violent compression" — an act that reveals what a film is made of by destroying its sequence. Alongside In Silico, the series represents a decisive move toward machine learning as a primary medium rather than a post-process.
Atomism (ongoing)
Atomism returns to the pixel — the irreducible unit of digital imagery — as both subject and material. The works deconstruct existing images by isolating and recombining their smallest components: individual pixels and signal values. The result explores what a digital image fundamentally is: not a picture, but a grid of numeric values that can be read, shuffled, and re-rendered according to any rule.
The title references the ancient philosophical doctrine that matter is composed of discrete, indivisible particles. Reas positions the pixel as the digital analogue of the atom — the point at which images become numerical data, and data becomes image.
Caesuras (2021–ongoing)
Caesuras operates as an alter ego — an anonymous identity through which Reas develops works centred on fields of colour and noise in relation to experimental sound. Released under a pseudonym, the series allowed him to explore a more purely aesthetic, less conceptually burdened space, free from the weight of the Processing legacy and the institutional framing of his primary practice.
The name caesura — a pause or break in a line of verse — signals both the works' relationship to rhythm and the deliberate rupture they represent in Reas' public artistic identity.
Technology — From Processing to GANs
Reas' practice spans three distinct technological eras, each of which opened new artistic possibilities:
- Processing (Java) — 2001 to present. The foundational medium. Reas still writes Processing sketches; the language remains the most direct expression of his core artistic logic. Its setup/draw loop, coordinate system, and immediate visual feedback define a mode of thinking-through-code that underpins everything else.
- p5.js (JavaScript) — 2013 to present. The browser-native version of Processing. Reas' role in establishing the Processing Foundation and supporting Lauren Lee McCarthy's development of p5.js extended the ecosystem into the web, making generative art native to the most widely distributed platform in history.
- Generative Adversarial Networks — 2018 to present. From In Silico onward, Reas has incorporated GANs and diffusion models into his practice. The shift represents a fundamental change in authorship: instead of writing explicit rules for image-making, Reas curates training data, shapes model behaviour, and edits outputs — a practice closer to directing than programming.
Across all three eras, the constant is Reas' commitment to process as both subject and method: the software is not a means to an end but the work itself, made visible.
Collections & Exhibitions
Reas' work is held in permanent collections at:
- The Victoria and Albert Museum, London
- The Los Angeles County Museum of Art (LACMA)
- The Centre Pompidou, Paris
- The Carl and Marilynn Thoma Art Foundation
- The Mary and Leigh Block Museum of Art, Northwestern University
He has exhibited at the Whitney Museum of American Art's artport, Ars Electronica (Austria), ZKM (Germany), Transmediale (Berlin), the Microwave International Media Art Festival (Hong Kong), and the Sónar Festival (Barcelona), among many others.
Books
- Processing: A Programming Handbook for Visual Designers and Artists (MIT Press, 2007)Co-authored with Ben Fry. The canonical reference for the Processing language — equally a programming manual and a conceptual manifesto for creative coding. Still in print and widely assigned in art and design schools worldwide.MIT Press
- Process Compendium 2004–2010 (REAS Studio, 2010)Documents the complete Process series. Each work is presented with its natural-language instruction, its code, and images of its outputs across different media. An essential document of the generative-art practice of the 2000s.
- Making Pictures with Generative Adversarial Networks (Anteism Books, 2019)A compact, practical guide to using GANs for image-making. Written from the perspective of an artist rather than a machine-learning engineer, the book was among the first to address GAN-based art-making as a serious creative practice.
Philosophy — Software as Medium
Several ideas run consistently through Reas' practice across more than two decades:
- Software as medium, not tool. The distinction is fundamental to Reas' thinking. A tool is instrumental — it serves a purpose outside itself. A medium is the substance of the work. When Reas writes code, he is not making a tool to produce an image; he is making the image in the only form it can take.
- Process over product. Reas is consistently more interested in the act of defining a process than in any specific output the process produces. His written instructions — in natural language and code — are as much the work as the images they generate.
- Connection to conceptual art. Reas situates himself explicitly in the tradition of Sol LeWitt, whose wall-drawing instructions are software programmes in all but name: descriptions of a process to be executed by others. The computational realisation of that tradition is, for Reas, not a metaphor but a literal continuation.
- Openness and education. Processing and p5.js are free and open-source. The Processing Foundation's mission is explicitly educational. Reas has consistently prioritised access — the tools he builds are designed to lower the barrier between having an idea and realising it in code.
"I'm interested in the point where a simple idea becomes complex behaviour — where a short instruction, given to a machine, produces something I couldn't have predicted or drawn by hand."