MPEG output is made up of three types of frames: I, P, and B. Nearly all of the pixels will be similar from frame to frame in situations when the camera and backdrop are immobile and one or two...

MPEG output is made up of three types of frames: I, P, and B. Nearly all of the pixels will be similar from frame to frame in situations when the camera and backdrop are immobile and one or two characters move around slowly. Simply subtracting each frame from the preceding one and applying JPEG on the difference will suffice in this case. This approach, on the other hand, fails miserably in situations when the camera is panning or zooming. Assume that the scenes are transmitted via UDP and that the backdrop of the scenes is stationary. UDP may lose some frames as a result of poor communication. Which of the three frames (I, P, or B) is the most important?