Introduction to H.264

The H.264 standard was developed in 2003 by the International Telecommunications Union Video Coding Experts Group (ITU VCEG) and ISO’s Motion Picture Expert Group (MPEG), as an improved version of the existing MPEG-4 standard. H.264, also known as MPEG-4 AVC (Advanced Video Codec) or as MPEG-4 Part 10, is considered the best video codec available today, due to its unprecedented compression power and versatility.

H.264 offers excellent video quality at lower bitrates than previous standards, sometimes achieving similar quality in half the bitrate required by an MPEG-2-encoded video. The variety of profiles and extensions developed for the standard allow it to be used in a very broad range of applications, from very low-birate mobile communications, to HD IPTV, video-on-demand, broadcasting, and Blu-ray discs.

The H.264 standard also promises to be a favorite of businesses due to its inexpensive licensing fees.

Major Features

Improved inter-picture (temporal) prediction.

More flexible use of reference pictures, allowing up to 16 reference frames.
Variable motion compensation block size, including 16x16, 16x8, 8x8, 8x4, and 4x4 pixels, possibly combining various blocks in a single macroblock.
Quarter-pixel precision for motion compensation.
Weighted prediction, allowing scaling during motion compensation and significantly beneficial for fading transitions in or between scenes.

Improved intra-picture (spatial) prediction, allowing prediction from the edges of neighboring blocks.
Flexible interlaced-scan features, such as macroblock-adaptive frame-field (MBAFF) coding, allowing the encoder to decide whether to code each macroblock as progressive or interlaced.
In-loop deblocking filter, to smooth the blocking artifacts common to DCT-based compression techniques.
Improved entropy coding, including:

Context-adaptive binary arithmetic coding (CABAC), a powerful algorithm for lossless compression of syntax elements, based on occurrence probabilities.
Context-adaptive variable-length coding (CAVLC), a lower-complexity alternative to CABAC.

A Network Abstraction Layer (NAL) definition allows H.264 to generate self-contained packets and avoid header duplication.

Profiles

The H.264 codec offers unprecedented versatility by defining a multitude of Profiles, each designed for a different use case.

Baseline Profile: intended for applications with severe computing resource constraints, such as in mobile communications and video conferencing.
Main Profile: originally intended as the mainstream consumer profile for storage and transmission applications, this profile became eclipsed by the High Profile as the use of HD video has grown.
Extended Profile: intended as the streaming video profile, with added provisions for noise- and error-resilience.
High Profile: the primary profile for storage and broadcast applications, especially for HD content (this is the profile adopted into Blu-ray Discs).
High 10 Profile: adds support for up to 10 bits per raw picture sample.
High 4:2:2 Profile: targeting professional applications that use interlaced video, this profile adds support for the higher 4:2:2 chroma subsampling format, as well as up to 10 bits per sample.
High 4:4:4 Predictive Profile: supports the higher-still 4:4:4 chroma sampling, up to 14 bits per sample, and the coding of each color plane separately.

The H.264 continues to be extended, with the latest important amendment being the Scalable Video Coding (SVC) extension. SVC allows a single coded stream to be used by receiver devices with different framerate or spatial resolution capabilities. Thus, either by dropping packets from the bitstream, or reading a subset of the coded stream, different devices can decode the same video to match their specific requirements.