Design of a Scalable System for the Sensorization of Virtual TV Sets with Live Broadcast

Abstract

The sensorization of virtual tv sets using motion capture systems allows the automation of many of the processes implied in the live broadcast of a tv show. It facilitates the tasks of both the actor (allowing him to interact with the environment) and the technicians (camera position and orientation, real objects position, etc.). Adding new sensors usually derives in an increase of the load of the render computer and therefore a decrease in the frame rate. This increase in the load is due to the captured data processing and its application to the scene, that must be computed in parallel to the render by the same computer. This article presents a centralized hardware and software architecture that allows the addition of an unlimited number of sensors without directly affecting the frame rate of the render. The system centralizes in a single PC all the data captured by the different sensors, giving a suitable format to this data and updating the appropriate variables across the network transparently to the rendering program, keeping a stable and unaffected render time.