MPEG-4 Part 2: DivX and Xvid

MPEG-4 Part 2 is the visual coding standard associated with Xvid and classic MPEG-4-based DivX video.

Label on an older video What it identifies
MPEG-4 Part 2 / MPEG-4 Visual The video coding standard
Xvid An implementation of MPEG-4 Visual, commonly Advanced Simple Profile
Classic MPEG-4-based DivX A branded implementation of this coding family
AVI A possible container carrying the coded track
H.264 / AVC / MPEG-4 Part 10 A different video standard

These distinctions make a legacy AVI collection easier to interpret. “MPEG-4” names a standards family; Part 2 identifies this visual format. The DivX brand also covers other formats, including AVC in DivX Plus.

Picture coding in the common rectangular form

Common Simple/Advanced Simple video represents rectangular pictures with block-based prediction and corrections. Independently coded pictures establish scene content, while inter-predicted pictures reuse reconstructed references.

The residual correction is transformed into block coefficients and quantized before coding. The decoder reverses the coded operations and adds the correction to the selected predictor. A newer encoder library does not change the decoder's job into a different standard: it chooses another permitted representation of the same format.

The profile determines which tool set is available. Advanced Simple Profile (ASP) adds useful choices beyond the simpler form, including B-picture coding and enhanced motion tools.

Two motion options with concrete meanings

Quarter-pixel motion compensation allows finer reference positioning. An integer displacement selects an existing sample location; a quarter-position displacement requires the codec's interpolation to form a predictor between grid points. This can better follow movement that falls between whole positions.

Global motion compensation (GMC) can describe a shared transformation across a larger portion of a scene. A camera pan or related movement can make many block displacements follow one common relationship, instead of describing every region solely as an unrelated local shift.

Those modes change the encoded instructions. The Xvid project documents ASP support for them, but a particular file can use a narrower selection. A decoder must implement the modes actually present.

XVID and DX50 identify the coded track

FourCC labels such as XVID and DX50 help identify a container's track and its intended format association. They do not contain the prediction and transform operations themselves.

A classic MPEG-4 Visual track moved into another supported wrapper still contains MPEG-4 Visual instructions. Turning it into AVC requires source decoding and a new encoding process; changing “Part 2” to “Part 10” in a label does not perform those operations.

For an Emby library migration, inspect the actual codec, profile and coding features before deciding whether the work is repackaging or video conversion. The name table establishes the family, and the tool descriptions establish what its decoder needs to do.

See Also

References