Remote rendering of streamed direct 3D games

Abstract

With the evolution in technology, the availability of personal devices is becoming ubiquitous. This is providing the gaming industry with the opportunity of reaching potential users by producing games on these platforms. However, devices used by the average user are not targeted for gaming purposes. Therefore the Quality of Experience on these devices cannot consistently reach high levels since current technology does not deliver the required capabilities. A new concept of remote processing for interactive applications has been under research for a number of years, in order to challenge this limitation. Interactivity within this concept requires a strict low-latency system, unlike video-on-demand and real-time video streaming for any other domestic applications, but like video conferencing for example. This requirement guarantees that the state of gameplay, perceived by the user, is not subjected to a delayed output, as interactivity would not make sense if this condition is not satisfied. Two different methodologies related to this concept are currently being researched. The first one is related to real-time video streaming with the second one being related to the streaming of issued graphical commands by the executing application. The latter was addressed through this study, by focusing on exploring ways through which graphical commands, called by an executing application to the Direct3D API, could be intercepted, processed and transmitted to another machine. The results obtained from the designed test-bed indicate the vital role of compression, in such a system, as well as identifying the fundamental benefits and flaws when compared to the video streaming methodology.