The H.264 Advanced Video Compression Standard, Second Edition

Chapter 9. H.264 performance

Introduction

The driving force behind the widespread adoption of H.264/AVC is its potential for significantly better compression performance than older formats such as MPEG-2 and MPEG-4 Visual. However, getting the best possible performance from an H.264 codec is not a straightforward task. First, despite the fact that H.264/AVC is an industry standard, there is a wide variation between practical implementations of H.264 codecs, with a corresponding variation in coding performance. Second, the large number of coding options supported by H.264 introduces a problematic trade-off between compression and computation. Achieving the best possible compression performance from an H.264 codec can result in a computational cost that is prohibitive for practical applications.

Probably the best way to understand the performance trade-offs and the capabilities of H.264/AVC is to experiment. Fortunately, a number of public-domain implementations are available that make this possible. The Joint Video Team, responsible for developing and maintaining the standard, publish a reference software implementation of H.264/AVC, the Joint Model (JM) codec. The JM is intended to be a complete and accurate implementation of all the features of H.264. It is not suitable for a practical, real-time coding application but is a useful reference tool to test the potential of H.264 and to check inter-operability between codecs and bitstreams. Using the JM and/or other H.264 implementations, ...