I was intrigued by the potential of MediaPipe for pose tracking and wanted to explore its capacity to produce unexpected or unconventional results.
Instead of using a live camera feed, I decided to experiment with public domain images of sculptures from The Met Museum. To my surprise, MediaPipe tracked the poses of the sculptures in unpredictable ways.
This unpredictability inspired me to explore the resulting pose values—such as those for hips, knees, ankles, and nose—as a dynamic source of parameters for music creation.
I mapped these values to a granular synthesis device, where the slight, continuous variations in the tracked data provided an evolving source of sonic modulation.