Cave Paintings

2021
Machine Learning, Computer Vision, JPEG files, prints

Cave Paintings are produced by a neural network system that uses Computer Vision (CLIP by OpenAI) to detect and reinforce visuals that emerge out of imagined caves. Carefully chosen text prompts, and sometimes image prompts, guide the system to detect caves that become the basis for more complex hallucinations. The adversarial system is stopped in less than a hundred iterations, resulting in a distinct colourful and unresolved aesthetic that induces our own vision to further identify and interpret forms, much like with early cave paintings. See the full collection on objktcom.