|
Digital
Compression
Lossy
and Lossless Compression
All
commonly used digital video formats use some
level of video compression. They are often
divided into lossless and lossy, although
there is no clear line between the two
categories.
With
lossless compression
there is no difference—or some people
would say "no discernible
difference" between the original and
the compressed data. Thus, no loss in
quality.
The
problem, however, is that lossless
techniques (with minimal compression) are
technically quite demanding. Thus, they
require expensive equipment.
Most
video and audio compression techniques
eliminate data to some degree to make
recording and transmission technically
easier to achieve. It then becomes a matter
of how much the data is compressed.
When to the trained eye quality starts to be
sacrificed, the term lossy
compression is used.
To
lower the costs as equipment moves from
professional to consumer, more and more
compression is used. As data is discarded in
the compression process, the resulting
signal looses quality—especially during
the copying process (which is essential for
sophisticated editing).
Compression Ratios
If you
start out with 1,000 bits of data and
compress it to 50 bits, you have a 2:1
compression ratio. If you can reduce the
original data to 25 bits, you now have a 4:1
compression ratio. Often, compression ratios
aren't in nice even numbers, but in
decimals, as for example, 2.5:1.
You can
easily use a 2:1 compression with video
without noticing any loss in quality. It's
compressed in such a way that even the
trained eye won't notice the loss of data.
In fact,
you can even compress video to 10:1 without
noticing a significant difference—and, in
the process, of course, you can record the
data in 1/10 the space.
When you
move to 20:1, you still have an excellent
picture; however, the trained eye will
notice a slight loss in quality.
As you move
to 50:1 and beyond, you can see a noticeable
difference between the original picture and
the compressed version.
MPEG2
Compression
Top-of-the-line
digital camcorders use a "no
compromise" digital 4:2:2
format. Although somewhat technical,
we often see video compression described in
these terms, and so it's probably important
to know what they mean.
The
first number refers to the luminance or
black and white part of the video, and is an
expression of the component sampling ratio
(not the compression ratio). The second two
numbers refer to the sampling ratio of the
color components of the video signal.
This
means that 4:4:4 is a pure, uncompromised
signal; 4:2:2 represents minimal and
unnoticed compression; and 4:1:1, which is
associated with DV camcorders, involves
significant signal compression.
MPEG2
compression is a sophisticated
technique that eliminates redundant
information. This is data between successive
frames that does not change, as well as data
"that we won't miss" within each
picture.
Rapidly-changing
subject matter such a hockey game is
particularly taxing for a compression scheme
such as MPEG2. In this case the discarded
data may be necessary to reproduce all of
the detail in the action. It is in this type
of subject matter that you are most apt to
see artifacts, visible problems
caused by the compression scheme not keeping
up with the speed of action.
Most
consumer camcorders use a 4:1:1
format (such as DV, DVCAM, DVCPRO,
etc.), which depend on a greater level of
quality-sacrificing compression. But even at
this level of compression, most people will
not notice the loss in quality.
JPEG Compression
MPEG2
discards video information that's repeated
between successive video frames. That makes
sense; why repeat the same information over
and over if it does not change? But, that
creates a problem for video editing.
Since
MPEG2 compressed video relies on data in
adjacent frames, if the video is stopped on
a particular frame—as would be the case in
editing—critical data will be missing.
And, there's another problem. Because MPEG2
data must be stored for brief moments in
order for the system to evaluate a sequence
of frames, there's a slight delay in
processing the video. This can befuddle
video editing equipment.
It's
possible to process MPEG2 video so it will
work in a limited way in editing systems;
however, such things as fades, crawls, and
wipe effects are not possible.
Fortunately,
there's an compression approach that does
not rely on a steady stream of pictures.
It's called JPEG compression.
Even
though higher compression ratios are
possible with MJPEG2, because of its editing
limitations, editors much prefer JPEG
compression where each video frame is
complete and able to stand on its own. Among
other things, this makes frame accurate
editing possible.
Whereas
MPEG2 can create compression ratios of 30:1
without appreciably degrading video quality,
JPEG compression is normally about
15:1—although it can vary considerably,
depending on the content of the picture.
JPEG
compression uses a highly sophisticated
mathematical approach to eliminating (to
varying degrees) nonessential picture
elements. It's also "scalable,"
meaning that you can easily chose to a range
of compression ratios.
|
