May. 11-12, 2009
SEATTLE, WA, USA
Five years have now passed since the creation of the High Profile of the H.264 / 14496-10 MPEG-4 Advanced Video Coding (AVC) standard. The development of that new standard and the Fidelity Range Extensions in which its High Profile was specified surprised some in the community who thought a major advance in video compression was not achievable. Now that new benchmark is well established. The standard has rapidly become a major market phenomenon. But time has continued to move forward – and so has the video coding community. Further new developments have continued to emerge.
After finishing the initial two versions of the standard, two more recent major extension projects were completed by the Joint Video Team of ITU-T’s VCEG and ISO/IEC’s MPEG. The first was Scalable Video Coding (SVC). In that project, important new techniques were created that extended H.264 / MPEG-4 AVC to create bitstreams that are highly scalable while retaining both excellent compression performance and consistency with the core design of the coding technology that forms its basis. SVC has recently been emerging in large-scale deployments as a highly-robust design for communication over heterogeneous and unreliable packet networks, and as a key enabler of low-delay real-time video communication. The most recent extension is Multiview Video Coding (MVC), which includes 3-D video and more. As it turns out, the core technology needed no changes below the header level to support these emerging applications. The power of multiple reference picture buffering was readily harnessed to enable the coding of multiple scene views. Inter-picture prediction is applied across camera views – through space as well as through time.
Now, a new momentum is growing in the standardization world to again tackle the fundamental problem of maximum compression capability. VCEG and MPEG are now working on new projects to harness and stimulate the community to surpass our previous best. VCEG has identified a number of Key Technology Areas (KTAs) in which improvements appear feasible, has confirmed the results in cross-verified experiments, and has defined requirements for Next Generation Video Coding. MPEG has launched studies on the feasibility of High-Performance Video Coding (HVC), mainly targeting high quality and high to ultra-high definition applications.
This talk will discuss these latest completed standardization projects and then delve more deeply into the newly-launched work. Some thoughts will be offered on the potential for further advancement and the amount of time likely to be necessary for a new state of the art to mature into products and standards.
Gary J. Sullivan has been a longstanding chairman or co-chairman of the ITU-T Video Coding Experts Group (VCEG), the video subgroup of the ISO/IEC Moving Picture Experts Group (MPEG), and the ITU-T/ISO/IEC Joint Video Team (JVT), which is a joint committee of the VCEG and MPEG organizations. He is best known for leading the development of the ITU-T H.264 | ISO/IEC 14496-10 MPEG-4 Advanced Video Coding (AVC) standard from its inception through several editions and extension efforts, including the Fidelity Range Extensions (FRExt), Scalable Video Coding (SVC) and Multiview Video Coding (MVC). He holds the position of Video/Image Technology Architect in the Core Media Processing group of the Windows division of Microsoft Corporation. At Microsoft he has been the originator and lead designer of the DirectX Video Acceleration (DXVA) video decoding feature of the Microsoft Windows operating system.
He and the team efforts that he has led have been recognized by an ATAS PrimeTime Emmy Engineering Award, a pair of NATAS Technology & Engineering Emmy Awards, the IEEE Consumer Electronics Engineering Excellence Award, the INCITS Technical Excellence Award, the IMTC Leadership Award, the Microsoft Standards Technical Achievement Award, and the Microsoft Standards Business Achievement Award. He is a Fellow of the IEEE and SPIE.