Raster art stores images as grids of pixels. Learn its history from Russell Kirsch's 1957 scan to AI upscaling and Texel Splatting for 3D pixel art in 2026.
The first digital image was made in 1957 by Russell Kirsch at the United States National Bureau of Standards. Kirsch and his team built a scanning drum that could digitize a photograph. They scanned a picture of Kirsch's three-month-old son, Walden, into a computer as a 176-by-176 grid of pixels. Each pixel was either black or white, because the computer could not store grayscale values. The image was stored as a binary grid: 1 for black, 0 for white, or vice versa. That 30,976-pixel grid is the ancestor of every digital photograph, every digital painting, every screenshot, and every frame of video you have ever seen. The technology has changed beyond recognition. The principle has not. A raster image is still a grid of pixels, each storing a color value, and the grid is still the fundamental structure of digital visual representation.
Raster art is digital graphics stored as a grid of pixels, where each pixel contains color information. The word "raster" comes from the Latin "rastrum," meaning a rake, and refers to the scanning pattern of horizontal lines used in cathode-ray tube displays. A raster image is defined by its resolution (the number of pixels in the grid) and its bit depth (the amount of color information per pixel). A 12-megapixel photograph is a raster image with approximately 12 million pixels. A 24-bit color image can display 16.7 million colors. Unlike vector art, which scales infinitely without quality loss, raster art loses sharpness when scaled up because the pixels must be interpolated, inventing new pixels from existing ones.
This entry covers the history of raster art from Kirsch's 1957 scan to contemporary AI upscaling, the artists who have used pixel grids as a fine art medium, and how 2026 developments in 3D pixel art and AI-assisted raster tools are extending the medium.
A raster image is a matrix of color values. Each cell in the matrix is a pixel, and each pixel has a color defined by numbers. In an RGB image, each pixel has three numbers representing red, green, and blue intensity, typically ranging from 0 to 255 in an 8-bit image. A pure red pixel is (255, 0, 0). A pure white pixel is (255, 255, 255). A pure black pixel is (0, 0, 0). The image file stores these numbers in a grid, and the display reads them and lights up pixels accordingly.
The resolution of a raster image is fixed. A 1920-by-1080 image has 2,073,600 pixels. If you display it at twice the size, the software must create new pixels to fill the gaps. This process is called interpolation. Bilinear interpolation averages the colors of neighboring pixels. Bicubic interpolation uses a more complex formula that produces smoother results. But no interpolation method can add detail that was not in the original image. Scaling up a raster image always produces some softness or blur. This is the fundamental limitation of raster art, and it is the problem that AI upscaling and super-resolution techniques attempt to solve.
Common raster file formats include JPEG (lossy compression, designed for photographs), PNG (lossless compression with alpha channel support, designed for web graphics), TIFF (uncompressed or losslessly compressed, standard for print production), and BMP (uncompressed Windows bitmap, largely obsolete). Each format trades file size against quality and capability. JPEG files are small but discard color information that PNG preserves. TIFF files preserve everything but are large.
Kirsch's 1957 scan of his son established that a photograph could be represented as a grid of numbers. The idea was simple, but the implementation was difficult. Computers in the 1950s had tiny memories and no display hardware. The scanned image existed as data on magnetic tape, not as a visible picture. It was not until the 1960s that computers could display raster images on screen, and not until the 1970s that raster displays became common.
The transition from vector displays to raster displays happened in the mid-1970s. Vector displays, which steered an electron beam to draw lines, were used in CAD systems and early video games like Atari's Asteroids (1979). Raster displays, which scanned the beam across a pixel grid like a television, could display filled areas, gradients, and photographic content. The Xerox Alto, developed at Xerox PARC in 1973, was one of the first computers to use a bitmapped raster display. Its screen had a resolution of 606 by 808 pixels, each one bit (black or white). The bitmapped display made it possible to show what we now call a graphical user interface: windows, icons, menus, and text rendered as pixels.
The first raster-based painting programs appeared in the early 1980s. Apple's MacPaint, released in 1984 with the original Macintosh, was a 1-bit (black and white) raster painting tool. It stored images as bitmaps at the Mac's native resolution of 512 by 344 pixels. Adobe Photoshop, released in 1990, brought raster image editing to professional photographers and designers. Photoshop supported grayscale and color images, layers, filters, and a wide range of editing tools. It became the dominant raster image editor and remains so in 2026.
Digital photography accelerated the dominance of raster art. The first digital camera, the Kodak DCS 100 (1990), captured a 1.3-megapixel raster image. By the 2000s, digital cameras had largely replaced film cameras for both amateur and professional photography. Every digital photograph is a raster image, and the proliferation of digital cameras made raster art the most common form of visual representation in human history.
Artists began using raster-based tools as fine art media in the 2000s. David Hockney's iPad drawings, begun in 2010 and exhibited internationally, use the Brushes app to create raster paintings on a touchscreen. Hockney's iPad works are raster art: they are stored as pixel grids, and they cannot be scaled up without interpolation. But Hockney does not see this as a limitation. He sees the iPad as a new kind of canvas, and the pixel grid as a new kind of surface.
Pascal Dombis uses algorithmic processes to generate raster images that explore the relationship between code and visual excess. His Irrational Geometrics series (2010s) uses algorithms to generate thousands of curved lines, rendered as high-resolution raster prints. The images are so dense that they appear as textures rather than as drawings, and the raster grid becomes part of the aesthetic: the pixelation is visible at large print sizes, making the digital nature of the image explicit.
Kirsch led the team at the National Bureau of Standards that created the first digital image in 1957. His scanning drum digitized a photograph of his infant son into a 176-by-176 pixel grid. The image is in the collection of the Smithsonian National Museum of American History. Kirsch's work established that any image could be represented as a grid of numbers, which is the foundational principle of all raster art.
Hockney's iPad drawings, begun in 2010, are among the most recognized examples of raster art in a fine art context. Using the Brushes app, Hockney creates paintings that are stored as pixel grids. He has exhibited iPad works at the Royal Academy in London, the de Young Museum in San Francisco, and other major institutions. His work demonstrates that raster art is not limited to photography or screen graphics. It can be a painting medium, with its own properties and its own aesthetic.
Dombis uses algorithms to generate raster images that push beyond what a human could draw by hand. His works are rendered as large-format prints where the density of algorithmically generated lines creates visual fields that oscillate between structure and noise. Dombis's work connects raster art to generative art, showing that the pixel grid can be a substrate for algorithmic expression.
While Miyamoto is a game designer rather than a gallery artist, his work at Nintendo defined the visual language of raster art for a generation. The sprite designs for Super Mario Bros. (1985) and The Legend of Zelda (1986) were drawn pixel by pixel on graph paper and then encoded as raster data. These images are among the most recognized raster artworks in the world. They demonstrate that raster art's constraints (limited resolution, limited palette) can produce a distinctive and enduring aesthetic. Read more in our entry on pixel art.
The fundamental limitation of raster art, that it cannot be scaled up without quality loss, is being addressed by AI upscaling. Modern super-resolution models use neural networks to infer high-frequency detail that was not present in the original image. These models are trained on pairs of low-resolution and high-resolution images, learning to predict what missing detail should look like. In 2026, AI upscaling is built into consumer tools like Adobe Photoshop and open-source tools like Stable Diffusion's upscaling models.
In the game development world, 2026 has brought new tools for raster art production. Pixelorama v1.2, released in 2026 by Orama Interactive, introduced a keyframe-based timeline with non-destructive animated layer effects, autotiling support for tilemap layers, and a rebuilt 3D layer system that lets artists draw directly on 3D models with the pencil tool. These features blur the line between 2D raster art and 3D modeling. Read more at Orama Interactive.
A significant technical development in 2026 is Texel Splatting, a rendering technique created by Dylan Ebert that produces perspective-stable 3D pixel art. Traditional 3D pixel art suffers from shimmer and flicker when the camera moves, because individual texels are sampled at different rates depending on viewing angle. Texel Splatting renders cubemaps from fixed world-space origins and splats them to screen as world-space quads, decoupling pixel resolution from screen-space sampling. The result is pixel art that remains crisp and stable under any camera rotation. The technique was published as an arXiv preprint (2603.14587) and has a live WebGPU demo. Read more at BrightCoding.
Raster art is the counterpart to vector art, which stores graphics as mathematical paths rather than pixel grids. The two represent the fundamental division in digital graphics. Pixel art is a subset of raster art where the pixel grid is small enough that individual pixels are part of the aesthetic. Raster art connects to digital photography, since every digital photograph is a raster image. It relates to digital art as the most common digital art format. For more on digital art history, read our post on the history of digital art from pixel to prompt, or explore our guide to digital art.
Raster art is the most common form of visual representation in the world. Every photograph on your phone, every frame of video on YouTube, every screenshot, every digital painting is a raster image. To see raster art treated as fine art, look at David Hockney's iPad drawings at the Royal Academy of Arts in London or the de Young Museum in San Francisco. Pascal Dombis's algorithmic raster prints are exhibited internationally at galleries specializing in digital art.
For more on the broader context, read our entries on vector art, pixel art, and digital art, or explore our post on the history of digital art.
Discover other art terms you might be interested in
3D modeling is the process of creating three-dimensional objects in digital space. Learn its history from the 1975 Utah teapot to Meta's SAM 3D and Roblox's AI Build in 2026.
Concept art is visual development art for games, film, and animation. Learn its history from Disney's Snow White to AI-assisted workflows in 2026.
Creative coding is programming for artistic expression, not utility. Learn its history from Processing to p5.js, key artists, and how code became a creative medium.
Read more about this topic on our blog
How video games became a serious visual art form: from the painterly worlds of Journey and Shadow of the Colossus to Elden Ring's gothic mythology. A guide to game art direction, concept art, and what makes interactive visuals different.
Discover how visual artists use code as a creative tool. A practical guide to p5.js, Processing, and openFrameworks with real examples and first steps.
David Hockney at 88 is still one of the most original painters alive. From Bradford to LA pools to iPad paintings in Normandy, here is his story.