AI and tracking
Use a camera when a performer’s movement should shape the work without a handheld controller. These examples connect movement to a result in three different ways:
- Face landmarks and radial blur: start here to make an effect follow one point on a face. The data displays help you understand coordinates before designing a mapping.
- Face tracking to native geometry: accumulate positions into a visible motion trail. This is the more technical option: the particle buffer must be matched to the detector’s output.
- Tracking zones to MIDI: turn crossing a boundary into a note instead of continuously changing a parameter. Its simulation mode lets you explore the interaction before connecting a camera.
The downloads include face-tracking models; score still needs ONNX/Pose Detector support and access to a camera for live tracking. Begin with one face and a stable view, then use the detection displays to distinguish a lost detection from a problem in the media mapping.