Edition #8 Generative art - same same but different; algorithms as co-creators
Art movements like minimalism find a new life in the digital realm
🖇 Generative art - same same but different
When I first started learning about generative art, it felt new - glitzy visuals on the backbone of mathematical/algorithmic systems, at times futuristic.
Then I realized it’s rooted in many antecedent art movements such as op art, suprematism, and minimalism, which started in the 1900s and often focused on geometric abstractions.
Many of the commonly used algorithms in generative art were developed much earlier too :
curl field - first used more than 2 centuries ago
quadtree - invented 5 decades ago
Perlin noise - developed for CGI around 4 decades ago
Back in the time computers don’t have screens, to view the output it requires a mechanical plotter - a drawing robot that holds a pen and move across the imaginary coordinates of a piece of paper.
Built on top of an accumulation of all these earlier movements, generative art moved beyond computer labs and made its way into artist studios circa 1970s. It keeps acquiring its own aesthetics, which sometimes takes the form of oddly satisfying fluidic animation, other times composition of neo-psychedelic colors.
A unique trait of it compare to other art genres is its capability of quickly generating and regenerating forms that look different.
Unlike traditional art which depicts humans, landscapes, or events, The subject matter itself is either abstract or serves as an abstraction layer overlaid on physical objects.
Its complexity can also be higher than other modern arts since humans don’t need to carry out the laborious sketching. At the same time, the algorithms used embued the work with a certain rhythm.
If interested in the backstory of generative art, here’s some additional reference:
A brief history of generative art
📚 Recent Reads
This is a section I share about the books, papers, or blogs on data visualization, data science, or others.
Make Time: How to Focus on What Matters Every Day by Jake Knapp & John Zeratsky
One of my resolutions this year is how to be more productive. Given the constant new developments in data to learn, it’s imperative to make the most out of limited time.
A technique this book illustrates is to make sure every day has a highlight. Be laser-focused on what must be done. At the end of the day, reflect on which highlight brought the most satisfaction.
This approach is slightly different from the usual urgency/importance quadrants used by ruthless prioritizers. It adds in a third factor, which is satisfaction & joy. I thought it is a nice addition, as it links back to the ultimate motivation.
You Look Like A Thing and I Love You by Janelle Shane
The author writes an ML humor blog about the weird things it does. The book illustrates cases of unfortunate & unexpected shortcuts AI makes to ‘solve’ problems. Also doubles as a communication guide in explaining things clearly.
🧉 Worth viewing
This is an occasional section on open learning resources.
Deep Learning for Art, Aesthetics, and Creativity
This video series features AI art practitioners going over their explorations, including work such as using quantitative methods to study art history, perceptions & art and ML-based creator tools.
For example, one of the applications is DoodlerGAN. Unlike quickdraw, the composition of the sketches relies on human-machine collaboration, where the user can choose specific parts (e.g. the eyes and beak of a bird) to sketch, and let the algorithm add wings and body etc. The architecture consists of a StyleGAN2-based generator and a discriminator which distinguishes both the appearance and location of the specified body parts.
Such a co-composition and co-creation makes me think about the role of ML algorithms in art.
Often they were used as the wayward executor of commands – unexpectedly aesthetic outcomes are cherished, out-of-whack unaesthetic results get discarded.
Alternatively, if we only partially leverage the algorithm instead of having them complete the whole work, it may strike a middle ground between novelty and familiarity - where algos add wings to human imagination.