This manual is for ENH-MC-EZBC, however, the manuals for the other two would be almost the same. MC-EZBC video codec consists of three parts: encoder, extractor and decoder. The exe files for the three parts are 3dsbcen.exe, pull.exe, 3dsbcde.exe. Encoder commands: 3dsbcen.exe *.par The *.par file is the parameter setting file, which speicifies the following parameters -inname the input file name -bitname the file name for compressed bits -statname the file name for some statistic information -start first frame number, generanlly it's 0 -last last frame number, depend on the length of the sequence -size the resolution of image, such as CIF 352 288 176 144 -framerate frame rate for the sequence, CIF will have 30 and SD will have 60 -denoise whether we need pre-filter or not. Generally it's NO -tPyrLev the number of temporal levels, CIF will have 4 with a GOP size 16 SD will have 5 with a GOP size 32 -MBSize the maximum size of blocks, generally it's 64 in HVSBM -MBlevel the level of HVSBM, generally it's 5. Then we have block size with 64x64 to 4x4 -MVaccuracy the sub-pixel accuracy, 1 means half pixel accuracy, 2 means quarter pixel accuracy -lambda used for R-D control, different temporal levels can have different lambda, such as 12 24 45 65 -searchrange the search range for motion estimation in the first temporal level, it will be doubled as temporal level increases -maxsearchrange the maximum search range for motion estimation -bi-mv always on (1). it's for alternative reconstruction with Haar filter for 5/3 filter -AGP_level how many sub-symbols we want to have for motion vector, typically 2 or 1. -layer_mv whether we want to code motion vector in two layers or not. 0/1 Extractor commands: pull.exe *.bit -r xxx -t x -s -a x x x x -y x x x x *.bit the file name for compressed bits -r xxx the target bit rate -t xxx the target frame rate. -t 0 means original frame rate, -t 1 means half frame rate ... -s xxx the target resolution. -s 0 means original resolution, -s 1 means half resolution ... -a x x x x how many sub-symbol we want to reserve for motion vector in the target bitstream if we don't input anything about -a, it will use the default method. Namely at full resolution, all sub-symbols will be used. At half resolution, the LSB of motion vector will be discarded. ... -y x x x x whether we want to discard the enhancement layer of motion vector in each temporal level. The default method will use all the layers. Decoder commands: 3dsbcde.exe *.bit output.yuv ref.yuv output.sta *.bit the target compressed bits extracted using extractor output.yuv the decoded yuv file ref.yuv the reference yuv file output.sta the PSNR information