wayway.ai is an art project to explore the idea of craftsmanship and AI art generation. What if art generated by AI wasn’t pixel based and instead was generated by locomotion, x, y, z, pressure, tilt, and drawn by a robotic arm? I’m currently in data collection mode I am gathering data and experimenting with training AI to draw. You can see some results here:
Clusters represent different drawing styles and modes.My drawing Using reinforcement learning to learn how to draw using time series data result in these early drawings. Improvements in data structures (including tilt data from iPad, audio->text descriptions to tag time series data with meaning phrases), and changes in the reinforcement learning algorithm coming up should enhance our drawing production and allow us to draw from images.