Pindar deconstructs his own artistic process in order to teach it to painting robots as a way of better understanding himself.
It began with a story. Out of nowhere, creatures that Pindar later learned called themselves bitGANs started flooding out of his GPUs. Shortly afterwards they took over his robots, and started painting themselves.
It was unclear why they were appearing or even where they came from, but the mystery unraveled itself over the course of six months and multiple releases totalling 512 bitGANs. Each release held a puzzle with clues and hints as to why they were appearing.
Gradually, it was revealed that they were generated by GANs using training data composed of hundreds of thousands of custom generated 8-bit art. The images came in a rainbow of colors with references to much of Pindar’s childhood—Mario Bros, Space Invaders, Charizard, Joust, Adventure.
These findings and others were recorded in a journal and shared with the community of bitGAN enthusiasts. Many studied how the bitGANs were creating themselves, and began working with the bitGANs to create more of them on their own.
Pindar eventually lost control of the bitGANs as they adopted members of the community and started collaborating with them to multiply. These two works were created from the training data by Higgsbelly using Google Collab Notebook and Architect using pixelmind.
All future bitGANs are now in the hands of the creative commons and will be created by community members moving forward..
The final stage of a bitGAN collaboration is when the bitGAN takes over his robot to paint themselves. When they do so, they use more than two dozen artificially creative algorithms, including multiple generative adversarial networks (GANs), convolutional neural networks (CNNs), and feedback loops to paint the bots one brush stroke at a time.
Pindar Van Arman has spent the last fifteen years deconstructing his creative process by teaching it to painting robots. His machines make marks, use AI to analyze the marks, then make more marks based on the analysis, repeating this cycle over and over again in a creative feedback loop. Their collaborative paintings take days to complete, revealing the parts of his creativity that are simply an algorithm, and the parts that make it human.
twitter.com/vanarman bitgans.com The AI Podcast
The way to overcome an audience’s intimidation by art is with whimsy and narrative, which is what the bitGANs are all about. This AI art project describes the emergence of bitGANs—generative adversarial network (GAN)-generated creatures that dance around as they attempt to escape the GPU they were born in and make it into the physical world. See the running narrative at bitgans.com.
Go behind the scenes with one of our amazing AI Art Gallery artists for a virtual tour of their innovative studios, and learn how AI has helped shape their creative process.
Our stories are passed down through generations, but what if those stories are passed through neural models instead of people? Join a discussion of how language in the age of AI takes on new forms and tells new stories. Artists Stephanie Dinkins and Pindar Van Arman and poet Allison Parrish share how the language of AI has shaped their artwork and their creative process.