3 cranes is a generative soundscape composition based on the motion of cranes operating at a construction site. Centered around the idea of a site specific sonification, an imperfect computer vision algorithm tries to track the movements of the cranes and transduces them into spatial and vector-based data. The sonic material comes from field recordings gathered at the construction site and is processed by the algorithm's output. Together they unfold a binaural sound space morphing through different states of sonic motion.
Videofile: https://drive.google.com/drive/folders/13PCNun4k5RFFgz9TlKmI6yCMgCJuexcu?usp=sharing
The video processing is implemented via OpenCV in Python. It uses a combination of motion detection via background substraction and the Lukas-Kanade Optical-Flow-Algorithm inside OpenCV. The optical flow data is sent via OSC to a Max/MSP-Patch to control playback and spatial rendering of the sound sources.
The Max/MSP-Patch uses the ICST-Externals for ambisonic panning and the ambiX.vst for binaural headphone rendering. You can download the ICST-Externals within Max/MSPs package manager. The ambiX.vst you can download here: http://www.matthiaskronlachner.com/?p=2015