This video shows my first steps towards a real-time video-to-sound conversion. The color-tracker was created using Python and OpenCV, using a basic contour tracker on a thresholded HSV channeled image. For each blob, a histogram is calculated.
The most prominent Hue value is then taken as input for the base tone. Information about the size of the contour and it's position is used to manipulate amplitude.
The sound was made using a simple Pure-data patch.