Specifications
10
A Digital Video Primer
Most codecs compress video using intraframe compression. With intraframe, or spatial, com-
pression, each frame of video is compressed separately. Many video compression schemes start
by discarding color detail in the picture. As long as this type of compression is not too severe, it
is generally acceptable.
A number of codecs also use interframe, or temporal, compression. is type of compression
takes advantage of the fact that any given frame of video is oen very similar to the frames before
and aer it. Instead of storing all complete frames, interframe compression saves just the image
data that is dierent, by generating three types of frames:
• I frames (which serve as the keyframes) contain a full representation of a frame of video and
use intraframe compression. I frames preserve more information than P or B frames and are,
therefore, the largest, in terms of the amount of data needed to describe them.
• P frames are predicted frames, computed from previous frames, and each may require less than
a tenth of the data needed for an I frame.
• B frames, or bidirectional frames, are interpolated from previous frames and those that follow.
B frames can be even smaller than P frames.
A typical sequence might look something like this:
I B B P B B P B B P B B P P B
How each frame is compressed depends on the type of content. If the content is fairly static (for
example, a talking head shot against a plain, still background with not much changing from
frame to frame), then few I frames will be needed, and the video can be compressed into a rela-
tively small amount of data. But if the content is action-oriented (for example, a soccer game, in
which either the action or the background moves or changes rapidly or dramatically from frame
to frame), then more I frames are required, a greater amount of data is needed to maintain good
quality, and the video cannot be compressed as much.
B and P frames contain only those portions of the adjacent I frames that change, therefore reducing the
amount of data required for a video. If more than half of a frame changes, an I frame is automatically
generated.
DV25 compression
DV25 is the compression format used for the standard DV format employed by most consumer
and prosumer camcorders. DV25 is compressed at a xed rate of 5:1 and delivers video data at 25
megabits per second (Mbps). Audio and control information is also included in the data stream,
so the total data rate in bytes is about 3.6 million bytes (megabytes or MB) per second. is
means that one hour of DV25-compressed footage will require about 13 billion bytes (gigabytes
or GB) of storage. It is impressive to realize that each 60-minute mini-DV cassette is actually
13GB of oine storage. DV25 compression uses 4:1:1 color sampling. e audio is uncompressed,
and there are two stereo audio pairs. e audio can be digitized at either 12 bits with a sampling
rate of 32kHz or 16 bits with a sampling rate of 48kHz. You should generally use the highest
quality setting (16 bit, 48kHz).