Macroblock

An image compression technique based on DCT (discrete cosine transform) where blocks of pixels from 4x4 to 16x16 are used on subdivide and encode portions (or slices) of one or more video frames. Macroblocks are usually composed of two or more blocks of pixels. Block size is codec-dependent in multiples of 4 blocks: the MPEG-2 video codec (and several other legacy codecs) have fixed 8vó8 pixel blocks; a macroblock would be a combination of several blocks of YCbCR, often with four Y blocks to one each Cb and Cr block (for YCbCr 4:2:0 chroma sub-sampling). The H.263 and H.264 (ISO/IEC 14496-10, AVC) codecs use a fixed 16vó16 pixel macroblock.