MPEG-2 Video

MPEG-2 Video is a video compression standard jointly developed by the ISO/IEC Moving Image Expert Group and ITU-T, officially designated as ISO/IEC 13818-2 and ITU-T H.262. The first version was completed in 1995 with the goal of covering different resolutions, bit rates, and scanning methods—such as digital television, storage media, and communications—using a single decoding syntax.

“MPEG-2” is a set of standards and does not refer solely to video coding.

Part 1 (ISO/IEC 13818-1, ITU-T H.222.0) specifies program streams, transport streams, and system multiplexing; Part 2 specifies video; and Parts 3 and 7 cover various audio codecs. Since MPEG-2 Video is not a file container, it does not automatically include MPEG audio; files with extensions such as .mpg, .vob, or .ts must be examined separately for their video, audio, and system layers.

Encoded images are organized hierarchically into sequences, groups of images, images, slices, macroblocks, and blocks. In a typical 4:2:0 video, a macroblock covers 16×16 luminance samples and corresponding chrominance samples, and is further divided into 8×8 blocks for the discrete cosine transform. The transform converts spatial pixels into coefficients of different frequencies, and quantization further reduces the precision of these coefficients; stronger quantization typically reduces the bit rate but is also more likely to cause blockiness, ringing, and loss of detail. The quantization matrix can adjust the level of detail retained for different frequencies depending on whether the block is intra-frame or inter-frame. MPEG-2 Video uses three main types of frames: I, P, and B. I-frames do not rely on motion compensation from other frames and serve as the basis for random access and error recovery; P-frames are predicted from a previous I-frame or P-frame; B-frames can utilize both forward and backward references. Picture groups specify a typical arrangement within a bitstream, but the standard does not require a fixed length. Since the backward references cited by B-frames must be decoded first, the decoding order in the bitstream may differ from the display order. Closed and open picture groups also differ in their handling of cross-boundary references.

Motion compensation is typically performed on 16×16 macroblocks or corresponding sub-blocks, with motion vectors indicating the prediction locations in the reference frame. How the encoder searches for motion vectors, detects scene changes, or allocates bits is not mandated by the standard; as long as the output bitstream can be decoded according to the specifications, the implementation is acceptable. MPEG-2 video generally uses half-pixel motion compensation and does not feature the finer partitioning schemes found in later standards such as AVC and HEVC; therefore, complex motion and edges may require more residual data at lower bit rates.

The standard is designed to accommodate both progressive and interlaced video. Images can be encoded as complete frames or as individual fields; macroblocks within frame images can optionally use frame prediction, field prediction, or frame- or field-based discrete cosine transforms. The bitstream uses flags such as `progressive_sequence`, `progressive_frame`, `top_field_first`, and `repeat_first_field` to describe the scanning and display timing. Programs sourced from film can use progressive images combined with the repeat_field flag to create a 3:2 pull-down, or they can be converted to true interlaced frames during the production phase; relying solely on the output frame rate reported by the container may not accurately reflect the original motion structure.

Profiles define the encoding tools and chrominance capabilities, while levels define resolution, sampling rate, bitrate, and buffering resources. The Simple Profile does not use B-frames; the Main Profile is the most common foundation for broadcast and consumer distribution; subsequent profiles include configurations for scalable encoding, high chrominance, or professional production. Main Level is commonly used for standard-definition digital television and DVDs; High-1440 and High Level extend to high-definition parameters; and the 4:2:2 Profile serves program production and transmission requiring higher chrominance precision. The combination of profile and level must be evaluated as a whole; one cannot determine the image size simply by seeing “Main.”

The video buffer verifier model specifies how a hypothetical decoder receives and removes data in the order of the bitstream, and is used to constrain peak data volume and buffer usage. Encoders may generate constant-bitrate or variable-bitrate programs, but must ensure that the buffer of a compliant decoder neither overflows nor underflows. Specific methods of bitrate control are not covered by the specification; therefore, two MPEG-2 encodings with the same average bitrate may have different instantaneous allocations and picture quality.

DVD-Video primarily uses MPEG-2 Video: full frames for 525/60 systems are typically 720×480, and for 625/50 systems, typically 720×576; these are encapsulated in VOBs along with audio tracks, subtitles, and navigation packets. Digital television more commonly places the video base stream within an MPEG-2 Transport Stream to accommodate continuous broadcasting and multiplexing of multiple programs. Program streams, transport streams, and video base streams each have distinct boundaries; therefore, neither the TS nor the VOB itself can be referred to as a video codec. MPEG-2 Video is also used in satellite, cable, and terrestrial digital television; early high-definition broadcasting; professional program exchange; SVCD; and some Blu-ray and HD DVD formats. Blu-ray permits MPEG-2 as a video codec, but the larger disc capacity and high-definition designation do not alter its underlying compression method; the actual bitrate is typically higher than that of DVD. Since then, H.264/AVC and HEVC have been adopted in many new systems, offering higher compression efficiency through more complex prediction and partitioning; however, MPEG-2 continues to be decoded and preserved due to compatibility with existing programming, broadcast infrastructure, and equipment.

The MPEG-2 Video label merely indicates that the bitstream complies with H.262 / ISO/IEC 13818-2; it does not, on its own, determine resolution, scanning method, aspect ratio, or quality. The standard allows for various parameter combinations, and the final result also depends on the source material, format conversion, noise reduction, quantization matrix, picture group structure, encoder implementation, and available bitrate.