Preferably, the method further includes the step of obtaining mixed prediction mode flag information (NOT_MIX_CODE), wherein the mixed prediction mode flag information indicates whether the macroblock includes the intra prediction coded sub-macroblock and the inter prediction coded sub-macroblock.

Preferably, the method further includes the step of obtaining DC transform size flag information indicating a size of a DC (discrete cosine) transform, determining the DC transform size of the coded sub-macroblock based on the DC transform size flag information, and determining a prediction size of the intra prediction coded sub-macroblock based on the determined DC transform size.

Preferably, the method further includes the step of performing an inverse transform with reference to the determined DC transform size.

Preferably, the method further includes the step of obtaining coded block pattern information, wherein the coded block pattern information indicates whether the sub-macroblocks include coded coefficients.