The Vision: A bridge between the physical body and digital expression, FireDance allows a performer to manipulate digital visuals through natural movement. The intent was to remove the controller entirely, making the human body the interface where every gesture paints the environment in real-time.

The Engineering: Using the Cinder C++ framework and a Microsoft Kinect sensor, I built a high-performance computer vision pipeline to interpret human kinematics without relying on pre-baked skeleton tracking.

Raw Depth Processing: To reduce latency and increase accuracy, the system ingested the raw depth map stream directly. I implemented algorithms to perform background subtraction and silhouette extraction in real-time, creating a high-fidelity mask of the performer that updated at the sensor's native frame rate.

Fluid Interaction: Instead of rigid bone tracking, I utilized optical flow or velocity mapping on the depth pixels. This allowed the system to calculate the speed and direction of the dancer's limbs, translating that energy into physics-based forces that drove the digital particle systems, making the visuals react to the intensity of the movement rather than just the position.