Windows, Kinect SDK, Cinder, C++
The program works by the following: In each frame, a ‘heatmap’ of ‘inigition’ pixels is created and these are copied into a working ‘heat buffer’. The ‘heat buffer’ then has ‘time’ applied to it, whereby all these pixels have their ‘temperature’ decreased (although not by a consistent amount) as well as their vertical position updated. Finally these temperatures are mapped to a heat colour range and then drawn to the screen, and then a new frame occur and the cycle repeats with new ignition pixels being added to the heat buffer. It uses the Kinect to create a user’s outline from the Kinect’s depth map, and uses the resulting outline pixel information as a source for ignition pixels.
One problem I encountered is the Kinect’s depth camera has a very low resolution (320×240) whereas the final installation was 1280×720. As a result, I had to scale up the image of the user which makes them look rather blocky, which is a bit annoying. This program was built in five days.