Sustainability of Digital Formats: Planning for Library of Congress Collections

Compression encoding for video identical to ITU-T (International Telecommunications Union-Telecommunication Standardization Sector) recommendation ITU-T H.264. MPEG-4_V based on H.263 is a sibling that appears to be less widely adopted. In order to distinguish the two, technical writers sometimes call this encoding H.264 video and the other MPEG-4 video. An overview of AVC encoding is provided in Notes below.

This page is used in this website to represent encodings with no profile indicated. The established profiles for MPEG-4_AVC are listed (and linked) as subtypes below. They are defined in terms of bitstream characteristics and relate to the conformance of players. The Baseline, Main, and Extended profiles were specified in Annex A of the 2003 edition of the standard. Four High profiles, part of a group called FRExt (Fidelity Range Extensions) were approved in 2005. All profiles can be wrapped in MP4_FF_2. There is another coding variant: MPEG-4_AVC_EXT, a specialized subset of AVC that appears not to be a "profile," although it requires the use of its own wrapper: MP4_FF_AVCE. Comments welcome.

For all profiles, MPEG-4_AVC (MPEG-4 Advanced Video Coding, part 10) is more efficient than the MPEG-4_V (MPEG-4 Visual Coding, part 2), i.e., MPEG-4_AVC provides better quality at the same data rate, or equal quality at a lower data rate.

Clarity is excellent for the profiles covered by the Fidelity Range Extensions (FRExt), which extend to lossless or near lossless levels, and support 4:4:4 chroma and bit depths as high as 12. See MPEG-4_AVC_HP, MPEG-4_AVC_H10P, MPEG-4_AVC_H422P, and MPEG-4_AVC_H444P.

Clarity in other profiles ranges from moderate to very good, given that chroma subsampling is limited to 4:2:0, and that these profiles will generally be implemented with significant amounts of lossy compression.See MPEG-4_AVC_BP, MPEG-4_AVC_MP, and MPEG-4_AVC_EP.

The outcome in all cases will depend on the extent of compression and the encoder used. All profiles support both progressive and interlaced video, at resolutions from sub-QCIF to HDTV.

[See Notes below for more detail on resolution terms.]

See MP4_FF_2

See MP4_FF_2

See MP4_FF_2

From the specification: "The coded representation specified in the syntax is designed to enable a high compression capability for a desired image quality. The algorithm is not lossless, as the exact source sample values are typically not preserved through the encoding and decoding processes. A number of techniques may be used to achieve highly efficient compression. Encoding algorithms (not specified in this Recommendation | International Standard) may select between inter and intra coding for block-shaped regions of each picture. Inter coding uses motion vectors for block-based inter prediction to exploit temporal statistical dependencies between different pictures. Intra coding uses various spatial prediction modes to exploit spatial statistical dependencies in the source signal for a single picture. Motion vectors and intra prediction modes may be specified for a variety of block sizes in the picture. The prediction residual is then further compressed using a transform to remove spatial correlation inside the transform block before it is quantised, producing an irreversible process that typically discards less important visual information while forming a close approximation to the source samples. Finally, the motion vectors or intra prediction modes are combined with the quantised transform coefficient information and encoded using either variable length codes or arithmetic coding." (p. ix) The specification also includes a number of additional elements.

The FRExt (Fidelity Range Extensions) as drafted in 2005 featured higher quality profiles, including one that is sometimes described as "lossless," which may mean that the second sentence in the preceding is no longer accurate. The four FRExt profiles are described on the following pages:

• MPEG-4_AVC_HP
• MPEG-4_AVC_H10P
• MPEG-4_AVC_H422P
• MPEG-4_AVC_H444P (described as lossless).

Terms used for spatial resolution: QCIF stands for Quarter Common Intermediate Format (176 pixels by 144 lines), CIF is Common Intermediate Format (352 pixels by 288 lines), while HDTV is High Definition Television (various, including 1920 pixels by 1080 lines).

See also MP4_FF_2.