X5

MPEG glossary 357
MPEG glossary
Motion estimation
Motion estimation is a further element for reducing data used in MPEG encoding.
Motion estimation also occurs in the B and P frames. The image difference that still
exist after prediction (view page 361) are examined. Complex algorithms are used to
search for an o
riginal occurrence of the macro block in the reference frame of each
macro block of the P or B frame (these are units of 2x2 blocks specially combined for
this purpose), which have been moved either by movement or by camera pan. They
can then be left out in the P and B frame. Only the information by how far and to
where the macro block has been moved is saved instead. This vector is called the
motion detector.
In the General encoder settings (view page 352), you can specify the quality of the
final MPEG vid
eo. This factor also influences the time required for encoding. The
longer it takes, the better the quality.
Bit rate
MPEG is a format used for storage and transferring. With older formats (e.g. AVI) you
could predict that 20 seconds of movie would result in 20 MB of data. The file size is
this a direct measurement of quality.
This is different for MPEG: The amount of data available can be used differently for
different display modes. 20 MB can be 4 seconds of DVD Video or 5 minutes Internet
streaming in thumbnail format. The quality of an MPEG video is measured by the
width of the created data stream, the bit rate. This is the amount of the transmitted
data per time unit; it is stated in kBit/s or bit per second.
Bits, not bytes are used, since the data word width has to address the transmission
restrictions.
The file size can be calculated from the average bit rate, if its length is known:
F = (BRV + BRA) * t
F=File size BRV=
Video bit rate
BRA=
Audio bit rate
t=Length in s