On Developing the Autonomous Capture of Cinematic Video