Abstract: A system for avoiding lost look in video conferencing utilizes multiple cameras outside a monitor screen at a conferencing station for capturing images of a participant and a processor system enabled to utilize video streams from the cameras and software including a field graph to create a stream presenting the participant as though the created stream were captured from a virtual video camera positioned within the monitor screen. In some embodiments the processor takes into account data pertaining to an active window to position the virtual camera in the active window. In other embodiments moveable cameras are employed with an audio system having directional input, which the processor can use to train cameras on a current active participant.