Advanced Video Coding: Difference between revisions

Revision as of 23:37, 29 November 2004

Revision as of 23:58, 2 December 2004

Line 6:

More recently, the JVT has been working on a corrigendum (a list of errata corrections) to the original standard. The drafting work on the corrigendum should be completed in May of 2004.

The intents of H.264/AVC project were to create a standard that would lead to fast implementations, using lowand bitrates that are half of H.263 rates for the same perceived video quality. That is: implementations that would demand little from the decoder hardware and from the network bandwidth.

H.264/AVC contains several new features that allow it to compress video much more effectively than older codecs. CABAC (Context-Adaptive Binary Arithmetic Coding) is a clever technique that can be used in H.264 to losslessly compress syntax elements in the video stream. H.264 also implements an in-loop deblocking filter which helps prevent the ringing and blocking artifacts common to other [[DCT]]-based image compression techniques. In previous video standards, [[motion compensation]] is handled by allowing blocks in a frame to refer only to the frame before it. H.264/AVC allows frames to be predicted from other frames that are arbitrarily far in the past. This usually allows modest improvements in bitrate and quality in most scenes. But (for example) in certain types of scenes with rapid repetitive flashing, it allows a massive reduction in bitrate. These ideas, along with many other new ideas, help H.264 to perform significantly better than MPEG-4 ASP can. H.264 can usually perform radically better than [[MPEG-2]] at a fraction of the bitrate.