A video processing system located at a remote location such as a TV broadcasting station or an Internet media server stores a plurality of video elements at that are delivered to a viewing system located at a viewer's premises. The video processing system is capable of receiving a selection request comprising two or more video elements or subsets thereof, wherein the subsets are regions of interest or portions of the corresponding video elements. The video processing system constructs a custom video by combining them based on the viewer' selection and using the region of interest that are provided, and forwards the custom video to the viewing system. Billing based on record of usage is also supported for the custom video.