What are the various prediction modes in MPEG-2?
A. 24.1 Frame: Predictions are formed from a 16 x 16 pixel area in a previously reconstructed frame. Identical to MPEG-1. There can be only one source in forward or backward predicted macroblocks, and two sources in bi-directional macroblocks. The prediction frame itself may have been coded as either a frame or two fields, however once a frame is reconstructed, it is simply a frame as far as future predictions are concerned. 24.2 Field predictions in frame-coded pictures: Separate predictions are formed for the top (8 lines from field 1)and bottom (8 lines from field 2) portions of the macroblock. A total of two motion vectors in forward or backward predictions, four in bi-directional. 24.3 Field predictions in field-coded pictures: Predictions are formed from the two most recently decoded fields. Prediction sizes are 16×16, however the 16 lines have a corresponding projection onto a 16×32 pixel area of a frame. One motion vector for forward or backward predictions, and two for bi-dire
