For my Embedded Systems class I'm working on a project where we are doing some webcam capture and video processing/effects work. To make life simple we are using OpenCV, which rocks. Our language of quote-in-quote choice is C# and to work with OpenCV we are using EmguCV wrapper/binding/what-have-you. So, while working on code that has to capture a video frame, process the frame to do face detection and a video effect, then draw a frame to display to the user I decided to throw everything into its own frame.