macroblocks

A macroblock is a fundamental processing unit used in many traditional video compression standards, notably the MPEG family. It normally represents a 16 by 16 block of luminance samples from a video frame, together with the corresponding chrominance samples according to the format’s chroma subsampling (for example, 4:2:0). The macroblock is the unit that is predicted, transformed, quantized, and entropy coded during encoding.

In practical terms, a macroblock carries both spatial and color information for a small region of the frame. The luma (brightness) component is stored as a 16x16 block, while the chroma components (color) are downsampled according to the subsampling scheme and encoded alongside the luma data. The exact arrangement of chroma data depends on the standard and color format used.

Prediction and partitioning are core roles of macroblocks in compression. A macroblock can be coded using intra prediction, which references neighboring pixels within the same frame, or inter prediction, which references pixels from one or more previously stored frames. To model motion more accurately, a macroblock may be partitioned into smaller subblocks, each with its own motion vector. Common partition schemes include layouts such as 16x16, 16x8, 8x16, and 8x8, enabling flexible motion compensation.

After prediction, the residuals—the differences between the actual pixels and the predicted values—are transformed and quantized. The resulting coefficients are then entropy coded and transmitted. While the macroblock concept remains central in many legacy codecs, newer standards have evolved the unit of processing; for example, some modern schemes replace macroblocks with more flexible coding structures, though the term is still used in many contexts to describe the basic 16x16 luma block and its associated chroma data.