Tie-in of art and tech
Actually, there is one expression method in particular I would like to discuss
What is it?
It's usually referred to as a "Generative design"

Generative Design

The process of creating media with the help of scripts and purpose built programs.

Purposes of Generative Design

Just art

First and foremost, it’s most commonly presented just as a form of art. It can be used to create visually appealing images, animations, or even music.

To visualize

However, generative art can be utilized also for other means. For example to visually represent the significance of data to the viewer. As visual content is processed much faster and easier than text or a spreadsheet full of data it’s advantageous to let a computer generate visual representations for them.

Or simulate

Another related field is simulation and modeling. Its aim is to copy physical properties and visually understand complex systems by building their approximations - models.

Media used by Generative Design

Image and video

The most commonly used media in generative design are without a doubt images and video.

Block animation (2016)

</> Browse source code

To show this output format I have chosen a very simple animation generated by a computer.

Generative artProcessingAnimation

3D objects

Rendering tools became very powerful over the years. And in combination with generative design they can produce visually appealing and complex scenes.

Golden cards (2021)

Here, I made a collection of procedurally generated 3D models. I used modeling software Blender and a script which builds the scene and then renders it. Each card here represents a topic that people talked about at a given day.

3D modelingProcedural generationBlenderData visualization

Games and virtual worlds

The most well known example of this approach is Minecraft. I can provide you with a procedurally generated world for your game. Or it can be used as content to fill the background of your scenes. For example a generated cityscape.

Infinite procedural worlds (2015)

</> Browse source code

The next example is in an old procedurally generated world of mine. It's built using a recursion of galaxies, planets, their terrains, and trees. User interface allows you to view the world on any zoom level and it's infinite in both x and y directions.

Try it out for yourself

Procedural generationPerlin noisePythonMap API

Generative Design concepts

Parameterization

I think that the most important concept of generative design is parameterization. A script can use input parameters (for example width of the image, number of fish) and using them calculate location, shape, color, and movement of objects on the screen. You should be then able to tweak parameters to produce many different art pieces.

“Ponyo” animation (2016)

</> Browse source code

If you watch this example closely you can notice that it's only using a single math function. sin(x). But when it's applied to basic properties of objects like position, rotation, and color it can still produce complex-looking results.

exp4jProcessingJavaMaven

Repeatability

I think the second most important concept of procedural design is repeatability. The idea of this content is really straightforward: Given a script (source code) and inputs users can always generate the same artwork. So this means you do not have to store generated media as it can be always reproduced. Or you can still choose to produce new artworks by altering its code or inputs.

Fern profile picture (2022)

</> Browse source code

I decided to demonstrate this concept on this simple “profile picture generator”. Given a name it will produce an image composed of fern leaves to represent the name.

Try it out for yourself

ReproducibilitySassReactCreate React Appd3

Randomness

The most important concept of generative art may be randomness. It can be used to alter input parameters of the generative process to produce a varied collection of artworks or assets.

You can use this concept to produce many unique but visually similar items. For example a lot of different rocks for a game, a collection of card decks where each package is slightly different, or now-so-popular NFT collections.

A tri-tree (2016)

</> Browse source code

In this example I am using a stream of random numbers to place new leaves forming tree-like structures in a process.

ProcessingRandomness

Transformation

Another approach to procedural art is not to fully generate the media by yourself but instead take some existing content as an input and only transform it in a way. You can take things like images, videos, text, numbers, datasets and use a deterministic script to produce new content. For example by doing geometric transformations, color modifications, abstraction, or simplification.

Album mood (2018)

</> Browse source code

To demonstrate this concept I chose a project that turns a series of photos (for example a gallery or video) into an abstract image containing a color palette that best represents the original images. This example uses statistical methods and other math magic to extract the relevant information.

Machine learningImage processingPythont-SNENumPyscikit-learnMatplotlib

Visualization

A specific case of transformations are data visualizations. We can treat real world data as an input into a procedural design script. But in this case its purpose isn't an artistic expression, but a representation of information.

Geocache density (2019)

</> Browse source code

For example here I used a heatmap view, and various plots to describe distribution of caches and their visits for a game played in the real world, Geocaching. I am sure you have encountered many visualizations in your life, so I don’t have to explain their capabilities.

Data visualizationVegaPandas

Simulation

And the last concept used in generative design I can think of is simulation. This can include physics simulation, flow of fluids, or discrete agents interacting with each other.

Sheep herding simulation (2017)

I have prepared this simulation using very few simple rules which jointly produce a complex behavior. In this case we have three basic rules:

  • sheep alter their heading to match other sheep nearby
  • sheep run away from a dog
  • dog tries to preserve a fixed distance from the closest sheep

And their interaction forces sheep to create a herd every single time even though it is not directly specified anywhere.

SimulationAgent-based modelingNetLogo

Concept mixing

Only once you add all these concepts together you can start to see the full potential of generative design. It can be used to create art, visualize data, simulate complex systems, or even generate new content based on existing one. The most powerful use cases are still in game development. But I can imagine understanding individual concepts can help you create visuals of any kind.

Made by

DeeDominik Gmiterko

References

Github
m.me/dee.gmiterko
LinkedIn
@dee.gmiterko

Isn't it fascinating how simple components can create such complex behavior, right?
Absolutely!
And as opposed to the traditional top-down design, generative design builds up from the bottom, creating virtually unbounded possibilities.
The biggest issue then is keeping it varied and interesting
And for it to have a point

Contact

You can start your very own conversation with me...
Feel free to reach out to me at your preferred platform 😊
Dominik Gmiterko
Email
me@deegmiterko.com
Messenger
m.me/dee.gmiterko
Instagram
@dee.gmiterko
LinkedIn
dominik-gmiterko
Discord
dee.gmiterko
© 2026 Dominik Gmiterko