Enhancing Quality Of Experience For Cloud Virtual Reality Gaming: An Object-Aware Video Encoding

Abstract

Cloud gaming is an ideal solution to provide the desired GPU rendering for the Virtual Reality (VR) gaming while reducing the requirements for the weight, price, and capabilities of the head-mounted display. However, transmitting the high-resolution video requires a large amount of network bandwidth. In this paper, we present a cloud VR gaming system that significantly improves the coding efficiency by using a proposed Object-Aware Video Encoding (OAVE). Different from the video service, the game engine can offer information about the objects in the game scene. By combining this information and the eye-tracking data, the encoder compressed the rendered video in an object-dependent way that sets the quantization parameter according to the object of the user’s gaze. Through the subjective experiment conducted on the prototype of the system, above 40% to 47% bandwidth saving can be achieved compared with the conventional VR gaming system without the proposed optimization.