In the late 1960s, the mathematician Georg Nees plotted a grid of identical squares and gave the machine a simple rule: as you move down the page, let each square rotate and drift a little more. The result, Schotter, German for "gravel", begins in calm, rigid order and gradually tumbles into scattered disarray. It remains one of the defining works of early computer art, and it is built entirely from repetition.

That is the paradox every generative artist faces. The loop is the most fundamental structure in code: do this, then do it again. It is also the fastest route to lifeless, mechanical output. Repeat a shape a thousand times without thought and you get wallpaper. Repeat it with intent and you get something that seems to breathe.

The difference lies not in the loop itself but in three decisions surrounding it: what the artist allows to change, what is allowed to accumulate, and what the repetition is meant to signify. Looking at how skilled practitioners handle each one shows the line between monotony and meaning.

Vera Molnár, one of the pioneers of algorithmic art, began working with computers in 1968 and spent decades exploring a single idea: the power of a small amount of disorder. In works like (Des)Ordres, she drew grids of concentric squares, then let the computer nudge their corners slightly off true. The structure remains legible, but every square becomes an individual. She described the goal as introducing roughly one percent of disorder, just enough to unsettle the eye without destroying the system.

That instinct sits at the heart of meaningful repetition. Perfect uniformity reads as machine output because nothing in the physical world repeats exactly. A hand-drawn line wavers. Leaves on a branch share a form but differ in every detail. When code introduces measured deviation, the viewer recognises the quality of something made, or grown, rather than stamped.

The craft lies in choosing the right kind of variation. Uniform randomness, where every value is equally likely, tends to look like static. Gaussian distributions cluster most results near a centre and allow occasional outliers, which feels more organic. Coherent noise functions, such as Perlin noise, make neighbouring values related, so variation flows across the canvas like wind through grass rather than flickering chaotically from one element to the next.

Skilled artists also layer variation hierarchically. The overall composition might be fixed, regional properties like colour temperature might shift gently, and the finest details, such as stroke weight or a vertex position, might vary freely. Seeded random generators make this controllable and reproducible, so the artist can explore thousands of outputs and return to the ones that work. The repetition provides the rhythm. The variation provides the melody.

Takeaway

Variation only becomes expressive when it is constrained. The artist's real decision is not whether to add randomness, but how much, of what kind, and at which scale.

One of the first lessons in creative coding environments like Processing is that the draw loop usually clears the screen every frame. Remove that single line and something remarkable happens. Each frame paints over the last without erasing it, and motion turns into trace. A single moving point becomes a line, then a tangle, then a dense field of marks. Many artists describe this discovery as the moment code stopped feeling like animation and started feeling like drawing.

Accumulation turns repetition into material. A stroke drawn at five percent opacity is nearly invisible, but a thousand of them layered in the same region build a rich, smoky tone that no single mark could produce. This mirrors traditional techniques like cross-hatching or charcoal, where value emerges from the density of many light gestures. The computer simply extends the scale, laying down millions of marks with a patience no hand could sustain.

Jared Tarbell's Substrate shows how accumulation can produce emergent form. Lines grow across the canvas and, when they meet an existing line, new ones branch off at right angles. No instruction describes a city, yet the result resembles an aerial view of urban blocks, cracked mud, or crystalline fracture. Flow field works, including Tyler Hobbs's widely discussed Fidenza series, operate similarly: countless curves follow an invisible vector field, and the field becomes visible only through their collective paths.

What makes this approach compelling is that the form lives nowhere in the code as an explicit shape. It exists only in the relationships between repeated elements. The artist designs conditions rather than images, then watches density, collision, and avoidance sculpt the outcome. Density itself becomes a compositional tool, with areas of heavy accumulation drawing the eye and sparse regions offering rest.

Takeaway

In accumulation-based work, the image is never drawn directly. It is the residue of a process, which means designing the rules is designing the picture.

In 1965, the painter Roman Opałka began painting numbers in sequence, starting at one, in small white figures across a dark canvas. He continued the project, numbering every subsequent canvas in the series, until his death in 2011. The work is not about any single number. It is about time made visible through a repeated act, and it anticipates a mode of thinking that algorithmic artists have embraced.

John F. Simon Jr.'s Every Icon, from 1997, takes this idea into code. A small black-and-white grid systematically works through every possible combination of filled and empty cells. The first rows cycle quickly, but the full sequence would take a span of time so vast that it exceeds any meaningful human scale. The piece will almost certainly never produce anything recognisable, yet it contains every image the grid could ever display. Repetition here becomes a meditation on infinity, possibility, and the limits of attention.

These works sit within a much older lineage. Islamic geometric patterns use repeating tessellations to suggest an order extending beyond the visible surface. Agnes Martin's hand-ruled grids invite slow, contemplative looking. Brian Eno's Music for Airports layered tape loops of unequal lengths so that they drifted in and out of alignment, creating a piece that never quite repeats despite being made entirely of repetition.

For many creative coders, the practice itself takes on this quality. Writing a loop, adjusting a parameter, and watching the result unfold becomes a form of attention. Zach Lieberman's long-running habit of daily sketches treats iteration not as a means to a finished product but as a way of thinking. The repetition happens in the artist's routine as much as in the artwork.

Takeaway

A loop can be a vessel for duration. When repetition is treated as a way of experiencing time rather than producing output, the process can matter as much as the result.

Repetition in generative art is never neutral. The same loop that produces tedious wallpaper can, with careful choices, produce texture, emergence, or contemplation. What separates them is intention applied at every level of the system.

Variation gives repeated elements individuality without dissolving structure. Accumulation turns countless small marks into forms no single gesture could create. Conceptual framing gives the act of repetition weight, connecting code to traditions that have long used iteration to approach time, infinity, and attention.

Perhaps the deeper lesson is that computation does not remove the artist from the work. It relocates them. The creative act shifts from making each mark to designing the conditions under which marks repeat. In that shift, the humble loop becomes one of the most expressive tools a digital artist has.