Differential motion vectors

    The macroblock vector is obtained by adding predictors to the vector differences indicated by Motion Vector Data (MVD). In case of one vector per macroblock, the candidate predictors for the differential coding are taken from three surrounding macroblocks as indicated in Figure 9. The predictors are calculated separately for the horizontal and vertical components.
    In the special cases at the borders of the current GOB or picture, the following decision rules are applied in increasing order:
    1) When the corresponding macroblock was coded in INTRA mode
        (if not in PB-frames mode) or was not coded (COD = 1), the candidate
        predictor is set to zero.
    2) The candidate predictor MV1 is set to zero if the corresponding macroblock
        is outside the picture (at the left side).
    3) Then, the candidate predictors MV2 and MV3 are set to MV1 if the
        corresponding macroblocks are outside the picture (at the top) or outside
        the GOB (at the top) if the GOB header of the current GOB is non-empty.
    4) Then, the candidate predictor MV3 is set to zero if the corresponding
        macroblock is outside the picture (at the right side).
    For each component, the predictor is the median value of the three candidate predictors for this component.

Advantage is taken of the fact that the range of motion vector component values is constrained. Each VLC word for MVD represents a pair of difference values. Only one of the pair will yield a macroblock vector component falling within the permitted range [-16, 15.5]. A positive value of the horizontal or vertical component of the motion vector signifies that the prediction is formed from pixels in the previous picture which are spatially to the right or below the pixels being predicted.

The motion vector is used for all pixels in all four luminance blocks in the macroblock. Motion vectors for both chrominance blocks are derived by dividing the component values of the macroblock vector by two, due to the lower chrominance format. The component values of the resulting quarter pixel resolution vectors are modified towards the nearest half pixel position as indicated in Table 15 of the Recommendation H.263.