On Generative Pixel Art

Since Larva Labs released their seminal project, CryptoPunks, back in 2017 pixel art and generative art have been bound together. Fired by the NFT as a means of marketizing digital art, pixels have resurfaced in recent years as a visual code celebrated by artists and collectors alike. As the essential ingredients of digital images, pixels define an information economy in which humans are increasingly enmeshed. They are also strongly suggestive of technostalgia, born of a longing for video games past and the perceived innocence of early personal computers. Yet they are also a visual means by which artists working with code can regenerate our shared digital universe. 

Back in 2012, Kim Asendorf introduced “pixel sorting” as a new open source code by which generative artists could modify existing images. Its value as a tool of visual disruption has been to unite pixel, generative, and glitch art according to a logic of wilful malfunction. Our new exhibition with Unit London, “The Pixel Generation,” includes a number of generative artists who are currently exploring pixels as a way of challenging the operations of digital culture. Here they speak with Alex Estorick about what pixel art means to them.

Alex Estorick: We feel lucky to have such a range of artists participating in “The Pixel Generation.” What can you tell us about your work for the exhibition?

‍Fay Carsons: I focus on aesthetics at the edge of human experience — 3D objects that, while appearing severe and foreign, are to me familiar and human, and the result of a mind wrestling with inputs it cannot properly process. I think of divine madness, psychedelia, psychosis, and the Book of Ezekiel if it took place in the present or future. I like to call these objects, “angels,” regarding them as beings that visit during intense transcendent states. Regardless of their medium, my works are often heavily inspired by William Burroughs and Georges Bataille. Given the theme of this exhibition, it only made sense to meld that interest with the aesthetics of failure and haywire tech, which is a pairing that I will continue to explore. 

‍Elsif: My work for this show, Last Rain of April, applies pixel sorting techniques to an abstract generative landscape. 

Every pixel is created entirely from JavaScript code, and then displaced by digital “rain.” With this piece I wanted to evoke a feeling of quiet contemplation.‍

Most of my generative works are created in JavaScript, using a combination of the Canvas API, p5.js, and Three.js libraries. For this exhibition, I developed a new program to draw, paint, and then sort the pixels, using many off-screen canvases in the process. The drawing step uses a few solid colors and precise lines; the painting step then blends the line colors and paints millions of small ellipses. I also used these techniques in my last long-form collection, For Algernon (2023). The sorting step is a new addition that takes place at the end, iterating over all pixels individually multiple times to sort them based on their perceived brightness. The end result is a soft painterly glitch effect.

‍Galo: My work for this show is part of a series called COMPUTATION. The artworks are generated algorithmically using a modified version of an existing JavaScript program I developed last year for a long-form project on fxhash. I used this program to generate new outputs that I could then selectively curate into a new series. Instead of real-time artworks, this series comprises looping animated GIFs. COMPUTATION references many different types of imagery from early punch-card programs to low-res video game graphics.

‍Aleksandra Jovanić: I am honored to participate in “The Pixel Generation” with my new generative artwork, Finity Pool, where I delve into the intricacies of dithering techniques, weaving infinite animations with soft movements, fake depth, and smooth color changes. My creative process for this work began with an exploration of dithering in knitting patterns. This involved similar restrictions to pixel art, while simultaneously drawing parallels with the constraints, rules, and limitations inherent to generative art at the algorithmic level.

Initially, the constraints of display technologies imposed limitations — all digitization processes result in the loss of data and a shift away from infinite resolution to finite, lowered resolution. But such constraints are not only inherent to digital technologies. 

Weaving is often regarded as a predecessor to programming, while numerous crafts that predate digital computing implement a form of pixelation. Cross-stitching, knitting, patchwork, and mosaic all rely on a reduction in resolution and stylization according to square units of color. ‍

The dissemination of patterns for such crafts and the adherence to instructions are very close to the practices we employ in generative art.

‍Kamau Kamau: For this show, I wanted to present my culture according to the theme of pixel art. The works I’m presenting are inspired by the procedural repetition that is quite common to a lot of African fabrics. Askew borrows the principle of simple shapes and bold colors to create patterns. In this case, lines of different thicknesses (weights) are drawn recursively throughout the canvas with segmented breaks in between. As the title suggests, the lines are rotated along the canvas to give greater presence and to generate organized chaos via the asymmetry and subtle nuances that can be spotted up close.

‍Loackme: My work for this exhibition, Limbo, draws inspiration from the countless isometric and platform games that I have played since childhood. As an homage to “Game Over” screens, the artwork depicts the liminal and almost meditative state that players enter after failure and before trying again. At 160 x 144 pixels (before upscaling) the resolution of the artwork pays tribute to one of my favorite gaming consoles growing up, the Game Boy.

‍Sarah Ridgley: My work, Pennyland, is a small series that focuses on individual pixels as the building blocks of a unified whole. The imagery was inspired by aerial photography and the natural pixelation that appears when viewing landscapes from above. Many years ago, I bought a giant roll of large aerial photographs from a law firm that was closing down, proceeding to wallpaper my entire print studio with them. I’ve always wanted to explore this subject in a generative work.

The title comes from the historical practice of a pennyland, where a farmer would pay a penny for each plot they cultivated. I like the idea of pennies and pixels as small units of measurement, and how these individual plots of land come together to create an intricate landscape. ‍

‍AE: An increasing number of artists are using code to generate pixel art right now. What makes pixels a natural aesthetic and symbolic vocabulary for you? 

‍E: Pixels are native to the world behind screens and the fundamental building blocks of all digital images. Working with pixels directly gives artists unparalleled control over the output. 

Making pixels visible as a part of the aesthetic of generative art is as natural to me as impressionist painters embracing brushstrokes. In a way, it’s like letting the viewers in on the secret behind the art, giving them a glimpse of the creative journey.

‍SR: I use different versions of pixel sorting in almost all my work, and those individual pixels are extremely important to the final images. My work takes advantage of the greater number of pixels available to artists today, so it doesn’t really reflect the traditional pixelated style. But everything is built on that basic unit of the pixel. 

I tend to create a lot of texture and depth in my work as a way to incorporate my human self into this purely digital medium. The vast number of pixels available in today’s digital canvases allow me to do that. As screens allow greater depth and density of pixels, I think it’s fascinating to explore what can be created using current technology. However, even though I have so many pixels at my disposal, I’m still limited by processing power and my own coding skills. I’m sure there are techniques I don’t know of that would allow me to utilize even more pixels in my work without stalling out the browser. I’m still constantly learning and seeking ways to make my code more efficient so that I can increase the detail of my work even more.

‍L: Pixels are everywhere in digital art, since virtually any piece of visual digital art is either made of or transcribed into pixels at some point. I find it really interesting to give prominence to this basic unit of digital art in my work. It’s the essence of digital art.

I also appreciate the suggestive aspect of pixel art. Fine details are suggested by chunky pixels and then require the viewer’s imagination to fully emerge. Every piece of pixel art is a bit collaborative, in a sense, and I think that’s part of what makes pixel art engaging.

‍AJ: Pixelated forms are hastily associated with computer art, video games and everything computerized. Early in my career I would get frustrated when clients asked for enlarged squares, mimicking pixels to emphasize that something was digital or made on a computer. At the time, I regarded that approach as overly narrow until I started exploring the aesthetics of retro video games — Pong, Pac-Man, and my favorite, Space Invaders. Self-imposed, rather than technologically imposed, limits became a new field of creative exploration for me.

‍G: The pixel is a discrete addressable unit, which means that it exists in something of a dormant state until it can be “activated.” It is also arguably the most important material in my approach to screen-based art. I like to reveal the nuances and materiality of pixels as a way of shifting the focus from the object of an image to the process of image-making. 

In the age when we (and AI) can produce essentially any kind of image content, examining the structure of images is a much more interesting intention when it comes to digital work. I’m thinking in particular of Hito Steyerl’s concept of the “poor image” where the crumply, corrupted, awkwardly file-named image says much more than a glossy, high-resolution, “perfect” one.

‍FC: There are plenty of connections that one can draw to nostalgia, video games, histories of digital art, and computation, but, for me, working at the level of the pixel is the only approach that gives me the level of control I desire. Seeing images as matrices and the value of each cell as a function of its position, while being able to decouple that data and use it in any arbitrary way, works for me on an intuitive level. Getting these motifs that reference old technologies or provoke nostalgia “for free,” so to speak, is simply an added bonus. 

‍KK: At the core of every language is individual letters that are systematically matched and paired together to create a vocabulary that is then used to convey endless emotions and actions. In the same way, individual pixels are the core of everything on our digital screens, arranged to display an infinite amount of content. 

Pixels, whether immediately noticeable or hidden in detail, speak to anyone who interacts with a piece of digital content, making it a universal language.

‍AE: Are there any particular generative techniques that you swear by: flow fields or shaders, for example? How do you alter your approach when pixelating the motif?

‍E: I’m not married to any particular technique, but I do love a good noise function as it is an important building block of many techniques, including flow fields and stippling. I sometimes convert my code into shaders to make it run faster, but I prefer to program most things in JavaScript. I’ve been a fan of Kim Asendorf’s ASDFPixelSort technique for a long time, so for this project I implemented a custom version of it to run over the generated image. In order for it to work well I took extra care to pick a palette that blends well, and did a lot of color blending with code to create pixels that look good next to each other when transposed.

G: I often say that I do not make pixel art, but rather low-resolution image art. The former, I think, is about deliberately placing individual pixels to construct compositions, while the latter is much more about image processing.

I often use shaders or limited resolution canvases as they allow me to manipulate the entire pixel grid of an image at the same time. Some recurring techniques I employ are image feedback, quantization, advection, physics simulation, and good old noise algorithms. I spend much of my time designing ways to filter low-resolution images into something else. Sometimes a project starts with a simple algorithm like Bayer dithering or Sobel filtering and then, through a back and forth process of breaking things and putting them back together, something else emerges.

‍FC: I have been going very deep into shaders for a while now. Honestly, I don’t know how people do it any other way. They afford so much versatility, and a real wealth of information and techniques spanning from the early demo scene to the present. Fine-grained control like that is a must for me.

‍SR: I love that working with code always involves learning, but I do have certain techniques that I pretty much always use, such as circle packing and pixel sorting. Most of my work involves creating graphic layers where I lay out shapes and colors, and then pixel sorting those layers into arrays that I render on a screen.

I use a lot of tiny pixel blocks in my work, but there are still limitations on what I can render and my work reflects what the computer is capable of at this current moment. As computers become more and more advanced, the appearance of generative art will change. Maybe eventually we will be able to recognize an image as belonging to this time, just as we can recognize an 8-bit image belonging to a certain era.

‍L: It’s not generative per se, as there is no randomness or autonomous process involved, but one of my favorite set of algorithms is dithering. Of course, it’s a fundamental tool of any pixel artist, but I really find the algorithms themselves so interesting. Most of these algorithms are relatively simple in their functioning and yet they allow nuanced and complex patterns to emerge.

Most of the time, I don’t think in terms of pixelating a motif or pre-existing composition. The pixelated aspect is part of the process from the very beginning. That being said, it obviously influences the way I approach contrast and detail level. As I’ve been working with dithering algorithms for a while now, it gets easier and easier to get a sense of how these algorithms interact with their inputs.

‍KK: Although I’m still exploring new techniques every day, I fell in love with flow fields at the beginning of my journey into creative coding and I still find new ways to incorporate them every now and then. They are extremely versatile and aesthetic making it a perfect tool for exploring different styles. As of now, I haven't merged pixel art into this but it is something I am excited to do in the future.

‍AE: While generative art and pixel art have somewhat different histories, both have tended to exist outside the mainstream art world due to their associations with science, technology, and video games. What do you think lends cultural importance to the pixelated image at this moment in time? Is it mere technostalgia or something more?

‍L: The association of generative art and pixel art with wider society may be the root of their cultural importance. Science, technology, and video games are ubiquitous, so it makes sense that artistic practices rooted in these resonate with people and can be seen as a broader mirror of culture. 

Technostalgia may play a role but I think a key aspect of pixel art now is that it’s an artistic choice and no longer a product of technical constraints. Even if that choice is purely aesthetic, I believe it has intrinsic meaning.

‍FC: I think that generative art’s explosion in popularity is a natural consequence of the continuing reach of technology into our lives. Everything from how we communicate to public transit to our household appliances is managed by computers or, more specifically, by the code on which they run. With more and more people entering tech, of course some of them are going to be artists, plenty of whom have personal associations with the aesthetics of retro or failing tech, as well as scientific illustration. 

‍G: Nostalgia certainly has something to do with it. In my book, Digital Fabrications: Designer Stories for a Software-Based Planet (2019), I have an essay that discusses the “post-digital” as a nostalgic trend. In short, it argues that the high-tech aesthetics of the 2000s yielded a longing for material qualities reminiscent of older media such as typewriters, film cameras, etc. While one might regard this as a “hipster” moment, scholars like Florian Cramer see it as indicative of specific desires that we have for our media, such as tactility and texture. This is evident in the skeuomorphic simulation of paper textures or watercolor effects in generative art.

Through my work, I’m making the argument that pixels have a tactility and texture of their own that is seriously underexplored and perhaps underappreciated because they are made of fluttering light. But as those of us who grew up with Game Boys and Tamagotchis can attest, screens and pixels are very materially interesting.

‍KK: Like a good writer who expresses a lot with few words, pixel art is a timeless act. We all love a good dose of technostalgia.

‍E: Working with pixels is like letting viewers in on the secret behind the art, giving them a glimpse of the creative journey. 

Pixelation highlights how digital art is a collaboration between an artist and a machine. In a world where AI is making machines more human, pixelated images are a reminder of how machines natively perceive the world when they are not pretending to be human.

‍AJ: On a personal level, pixelated images appeal to my own technostalgia, as they evoke memories of my childhood and initial experience of video games. However, on a broader cultural level, the importance of this style and technique extends beyond mere nostalgia, sparking creative exploration within the same constraints — limited color palettes and low resolutions — as early computing technology.

‍SR: I think a lot of the misunderstandings about digital art that existed in the beginning came from a lack of familiarity with the media involved. But many people living today have grown up with digital technology seeping into every aspect of their lives. They can therefore appreciate the skill and intention required for digital art, including pixel art. I don’t think there has been mainstream appreciation of digital art until now, but it has been propelled by the rise of NFTs and digital ownership. Artists are able to sustain their practices in a way that simply wasn’t possible before.