FOUNDATIONS · 10 MIN READ
What Is Algorithmic Art?
Algorithmic art is art whose form is determined, in meaningful part, by an algorithm: a defined sequence of rules or operations. Instead of placing every line or pixel directly, the artist describes a procedure capable of making those decisions when it runs.
Most code-based algorithmic art sits within the wider field of generative art. The terms emphasize different things. “Algorithmic” identifies the kind of system being used; “generative” describes the system’s active role in producing the work.
IN THIS GUIDE
- An algorithm is a procedure, not necessarily a computer program.
- Algorithmic art usually becomes generative when the procedure helps determine the finished work.
- The same algorithm can be exact and repeatable while still producing many outputs from different seeds.
- At 256ART, artist-authored code, a token seed and on-chain data combine to produce each live artwork.
01
An algorithm is a set of instructions
In computing, an algorithm is a sequence of operations that transforms an input into an output. An artistic algorithm might repeat a shape, subdivide a plane, move simulated particles, grow a branching structure or assign colors according to neighboring forms.
The idea does not depend on a screen. Instructions can be followed by a person, a plotter, a robot or another physical process. Vera Molnár described a “machine imaginaire”—an imaginary machine through which she applied systematic transformations by hand—before she gained access to a computer in 1968.
function draw(seed) {
const random = seededRandom(seed);
for (let i = 0; i < 128; i++) {
const x = random() * width;
const y = random() * height;
const size = 4 + random() * 36;
circle(x, y, size);
}
}02
Where algorithmic art fits within generative art
Generative art is the broader idea of giving a system some functional autonomy in the making of a work. That system could be algorithmic, but it could also use biological growth, chemical reactions, mechanical behavior or written instructions interpreted by people.
Algorithmic art names work organized through an explicit procedure. It overlaps most clearly with generative art when executing that procedure determines features the artist has not composed one by one. A completely predetermined pattern can still be algorithmic; variation and chance are not requirements.
There is no single taxonomy accepted by every artist and historian. In contemporary creative coding, however, “algorithmic generative art” is a useful description for work in which artist-defined computational rules generate one or many outputs.
04
Determinism, seeds and apparent chance
An algorithm can be deterministic: given the same input and environment, it returns the same output. That does not mean every output from the system must look alike. A seed can initialize a pseudorandom number generator, producing a different but repeatable sequence for each input.
This combination is especially valuable in collected digital art. The output can feel surprising at mint while remaining independently reproducible afterward. True unpredictability may help assign a seed, but permanent reconstruction depends on the artist’s script interpreting that seed consistently.
Not every algorithmic artwork uses randomness. Some explore all states of a logical system, transform live data or animate a fixed sequence. Randomness is one artistic material among many.
05
A short history of algorithmic art
Procedural image-making predates electronic computers, but access to programmable machines made the algorithm visible as a modern artistic medium. During the 1960s, Vera Molnár, Georg Nees, Frieder Nake, A. Michael Noll and Manfred Mohr were among the artists and researchers who used code and pen plotters to turn mathematical instructions into drawings.
The V&A describes early computer art as the product of collaborations with laboratories, universities and corporations because computers were not yet personal tools. Plotter drawings also make an important point: algorithmic art was never restricted to images displayed on a monitor. Code could direct ink, paint, movement, sound and physical machinery.
Personal computers, Processing, p5.js, shaders and browser-based JavaScript later opened the field to more artists. Blockchains added durable seeds, provenance and programmable distribution, but algorithmic art does not require a blockchain—and putting an image on-chain does not make it algorithmic.
06
Algorithmic art is not synonymous with AI art
Machine-learning systems are built from algorithms, so AI art can be discussed within a broad algorithmic history. The working methods are nevertheless different enough to name clearly. In traditional algorithmic art, an artist commonly writes explicit rules that map inputs to forms. In machine learning, a model learns statistical relationships from training data and generates results through those learned parameters.
The boundary is not absolute. Artists can train their own models, combine neural networks with hand-written systems or use AI as one stage of a larger algorithm. The useful questions are practical: who designed the operative system, what data and dependencies does it require, and can the relationship between process and output be examined?
07
How algorithmic art works on 256ART
256ART is built for artists who treat code as the artwork’s operative medium. For modern releases, the collection preserves the artist’s compressed script in contract bytecode. Each token is associated with a seed and on-chain traits that become inputs to that script.
The contract’s tokenHTML function assembles a browser-readable HTML document from this material, while the standard ERC-721 tokenURI returns the token metadata. The result is not only a record that an image existed. The artist’s algorithm and the token-specific inputs needed to execute the live work remain retrievable from the blockchain.
This architecture does not make the artistic decisions. It preserves where those decisions live: in the artist-authored procedure. A 256ART project can use geometry, simulation, animation, interaction or blockchain state, but its visual identity still comes from the rules the artist chooses.
08
How to read an algorithmic artwork
A single output can be appreciated on its own, but algorithmic work becomes richer when viewed as evidence of a system. Compare several outputs and look for invariants, ranges and relationships. Ask which qualities belong to every work and which emerge only under particular inputs.
- What remains constant across outputs?
- Which features vary continuously, and which switch between discrete states?
- Does complexity come from many rules or interactions among a few simple rules?
- Is the process deterministic, responsive or time-based?
- Is the executable system preserved with the output?