CubeFlow was a 2nd prize winner of the Intel Perceptual Computing Challenge 2013. We build a gesture-controlled version of CubeFlow that uses hand movements for motion input and navigation:
First basic research on CubeFlow started with a native implementation of the concept in OpenGL and OpenCL. Using visual computing technologies, we achieved highest graphics performance and were able to process a video stream of 320×240 and display the result with millions of triangles in realtime. The video frames are stored in 240 slices, buffering approximately 10 sec of video input and yield 18.4 million points which are used for surface reconstruction.