ATI Video Quality

180 Adams Avenue, Hauppauge, NY 11788 | Phone: (631) 300-4000 | Fax: (631) 434-7000 | www.aventuratechnologies.
com
1
Copyright Aventura Technologies Inc. 2008 All rights Reserved
Volume 1, issue 1 November 2008
Determining Video Quality
by Jack Cabasso
Probably one of the
most confusing issues
I have encountered in
my discussions with
security professionals
is the topic of video
quality comparisons.
A lot may be attributed
to an oxymoron or two
created by the English
written language. After
all, we speak of high
resolution cameras,
or high resolution
images. So one would
be apt to assume the
word resolution is
synonymous with qual-
ity. We all know what
happens when one
assumes though. Un-
fortunately, resolution
in video terms does not
equate to quality. This
begs two questions;
what is resolution, and
if not quality, what is?
Video Glossary
The frst order of business to put all the pieces of the puzzle together is a quick video defnitions
lesson.
Bit: a binary digit taking a value of 0 or 1 and the basic unit of measure
for digital information storage.

Bitrate: the number of bits processed per unit of time (not distance),
quantifed as bits per second (bps) or in larger increments kilo bits per
second (kbps), mega bits per second (Mbps), giga bits per second (Gbps)
or tera bits (Tbps).
Codec: a program which encodes (compresses) a digital stream for
transmission, storage and encryption and/or decodes a digital data
stream for displaying and editing Compression/Decompression.
Common Intermediate Format (CIF): Defnes the resolution in terms of
width and height of the video frame and the frame rate. CIF is stated as
352 x 240 pixels in the NTSC (National Television System Committee - used in North America)
format. Other common resolutions are: 2CIF (704 x 240), 4CIF (704 x 480). This should not
be confused with Common Image Format, a standard frame size for digital video.
Compression Artifact: the result of a signifcant data compression
scheme applied to an image, audio, or video which discards some critical
data that may be too complex to store in the available data-rate, or may
have been incorrectly determined by an algorithm to be of minimal value.
Image: a two-dimensional artifact.
Pixel: (picture element) is the smallest portion of information in an image. Pixels are merely
samples of original images.

Resolution: distinct number of pixels in each dimension that MAY be displayed. Resolution
can be specifed as the number of pixel-columns (width) by the number of pixel-rows
(height).
Sample: a set of values at a point in time and/or space.
TVL (TV Lines): Method of defning resolution in analog video.
Bit
Bitrate
Codec
Common Intermediate
Format (CIF)
Compression Artifact
Image
Pixel
Resolution
Sample
TVL (TV Lines)
Video Glossary
Knowledge You Need
Copyright Aventura Technologies Inc. 2008 All rights Reserved Rev. 112508
180 Adams Avenue, Hauppauge, NY 11788 | Phone: (631) 300-4000 | Fax: (631) 434-7000 | www.aventuratechnologies.com
2
Bitrate Equals Quality
The resolution is the
number of pixels that
MAY be displayed,
not that ARE displayed.
Therefore, you should
think of resolution as a
piece of graph paper. All
you have to start with is
a bunch of empty boxes;
which would equate to
pixels. This is the true
defnition of resolution.
CIF resolution would be
analogous to a piece of
graph paper with 352
boxes wide by 240 box-
es high (although a pixel
is not perfectly square).
A 4CIF image would be
a piece of graph paper
with 704 boxes wide
by 480 boxes high.
Since there are no bits
of information in these
boxes yet the resolution
just tells you the number
of pixels that MAY be
displayed.

Resolution Does Not Defne Quality
What actually is a key determining factor but not the only consideration in quality
is the bitrate. Bits are the information to be inserted into the pixels and the bitrate
is the rate of time at which those bits of information are placed into the pixels. The
more information or bits applied to the pixel, the denser the image and the higher the
quality should be if all other factors are equal. Think in terms of a printer, the more
dots-per-inch, the sharper the image. So when you ask someone what is the video
quality requirement and they say 4CIF, ostensibly they have not advised you of any
visual aspect of the quality.
Resolution is the
number of pixels
that MAY be
displayed
Not that are
displayed
CIF 4CIF
More bits of
data applied
means higher
image density
and resulting
quality
4CIF (Low Bitrate) CIF (High Bitrate)
3
If compression
is too aggressive
it becomes
noticeable in the
form of blocky
images.
So if all things are equal
(which is rarely the case) a
higher bitrate image with
a higher resolution may
produce a higher quality
image than a lower
bitrate image with a lower
resolution. Once we have
determined bitrate the
next primary factor to
consider is the Codec
being applied and the
quality of the resulting
data.
Video in its raw uncom-
pressed format is typi-
cally too large to accom-
modate for purposes of
storage, transmission
and broadcasting. In
order to create effcien-
cies trade-offs are made
between minimizing the
bitrate by compression
and maximizing the
recorded quality. Pro-
grams called codecs
are used to accomplish
the task.
In digital video to reduce
the bitrate a lossy com-
pression codec is utilized.
A lossy compression is a
method which discards
some of the information
or data which is not criti-
cal to the scene, thus
resulting in a lower bitrate.
When lossy data com-
pression is used on audio
or visual data, differences
from the original signal are
introduced since data
has been discarded. If
the compression is too
aggressive it becomes
noticeable in the form
of compression arti-
facts, which appear
in the form of blocky
mosaic images. Wheth-
er these affect the per-
ceived quality, and, if
so how much; depend
on the compression
How Codecs Affect Video Quality
Bitrate depends upon several factors:
The original material may be sampled at different frequencies
The samples may use different numbers of bits
The data may be encoded by different schemes
The information may be digitally compressed by different algorithms
The information may be digitally compressed to different degrees
So what is the difference in potential quality
of CIF vs. 4CIF? Resolution quantifes only how
close pixels or lines can be to each other and still
be visibly resolved. A 4CIF image can poten-
tially produce a better quality image then a CIF
image because the more pixels used to represent
an image; the closer the result can resemble the
original.
It does not necessarily have to be the case though
as 4CIF does not mention the amount or quality of
the data to be displayed in the pixels. Accordingly,
a CIF image flled with more bits and higher quality
bits than a 4CIF image can and will produce better
visual and audible results. Other considerations vis-
-vis bitrate need to be accounted for to determine
the potential video quality.
CIF vs. 4CIF
4CIF Low Bitrate
CIF High Bitrate
4CIF (High Bitrate) CIF (Low Bitrate)
4
scheme, encoder power,
characteristics of the
input data, the viewers
perception and familiar-
ity with artifacts.

Some codecs are more
effcient than others at
compressing data. Earli-
er encoding technologies
date back more than half
a century (Huffman al-
gorithm), which surpris-
ingly enough although
not very effcient are still
used today.
Newer state-of-the-art
codecs, such as H.264,
take advantage of the
more modern arithmetic
algorithms. The result
is higher quality images
at lower bit rates com-
pared to the legacy co-
decs (MPEG1, MPEG2,
MPEG4 Part 2, H.261,
H.262 and H.263)

H.264AVC is the frst
of what are known as
complex codecs, which
can yield up to 50% bet-
ter compression than
MPEG2 and up to 30%
better than MPEG4. Bet-
ter compression means
lower bandwidth require-
ments and faster down-
loads times. Also, the
resulting H.264 video
quality is up to 40% su-
perior then MPEG2 and
MPEG4.
Complex codecs have
various other attributes
that make them more
intelligent and capable
of dealing with unique
situations relative to
security. For example,
pan-tilt-zoom (PTZ)
cameras are common-
place in the surveillance
world. When swinging
the PTZ cameras older
codecs tend to have dif-
fculties in analyzing and
processing the scene
content either intelli-
gently or quick enough
and as such compres-
sion artifacts are preva-
lent and at times may be
severe. H.264AVC has
several unique charac-
teristics not available
in predecessor codecs,
which enable it to stabi-
lize fast moving images.
Modern codecs
compress video
more effciently,
resulting in higher
quality images with
accompanying
smaller fle sizes
File Size
Download Time
Bandwidth
I,P, B-frames
Interlace
Coding
Block size
1/4 pixel
GMC
Loop Filter (aka deblocking fiter)
Slice-based motion prediction
Multiple reference frames
MB AFF (improved interlaced management)
RDO (Rate Distrortion Optimisation)
WP (Weighted Prediction
Switching pictures(for fast chang channel)
MPEG -2 MPEG-4 H.264
Standards
F
e
a
t
u
r
e
s
Huffman Huffman Huffman or Arithmetic
fxed 16x16 fxed 16x16 variable down to 4x4
5
In order to compensate
for extreme motion,
whether it is from the
movement of a cam-
era or just plain old
fast moving objects,
bitrates are elevated
to extreme levels to try
to introduce as much
information into the
image as possible.
Unfortunately, massive
bitrates means massive
fle sizes and subse-
quent storage and pro-
cessing issues. Regard-
less, even at extremely
high bitrates the older
codecs for the most
part are not capable of
stabilizing fast changing
scenes. This is some-
thing you can easily
test for yourself on your
DVR or NVR. So forget
about what the manu-
facturers specifcations
say, try it for yourself;
seeing truly is believ-
ing.
Accordingly, if we say
bitrate equals quality
and 2 video fles are
using the same bitrate,
one might be better than
the other as a result of
a more effcient codec.
An H.264 video clip
recorded at 2Mbps
might look substantially
better than an MPEG2
video clip recorded
at 2Mbps due to its
effciencies. In fact,
since we say that
H.264 compresses
video more effciently,
a 1Mbps H.264 image,
may look better than
an MPEG2 or MPEG4
image recorded at as
much as 2Mbps.
Compensating for Extreme Motion
A 1Mbps H.264 image, may look better than an MPEG2 or MPEG4
image recorded at as much as 2Mbps.
4CIF
1Mbps
4CIF
2Mbps
6
While this may appear to be the end of the story there are other caveats.
We now know the following:
Resolution is nothing more than the number of pixels to be flled with data
Bitrate is the key determining factor in what actually determines the video quality
A higher resolution image (4CIF) does not guarantee a better image than one with a lower resolution
(CIF)
Some codecs can provide higher quality images and smaller fles sizes than others
It appears then that the formula for determining the ultimate video quality is:
High Resolution + High Bitrate + Best Codec = Best Quality
7
When evaluating video
quality you should view
recorded video data
with signifcant mo-
tion; which will expose
any weakness in the
compression scheme.
Codecs may perform
well if there is minimal
motion, as not much
is being asked of the
codec and processors.

A further recommen-
dation would be if its a
DVR (As opposed to an
IP camera) to record the
maximum number of si-
multaneous channels,
which may show any bot-
tlenecking or resource
sharing problem. Also,
look to see what hap-
pens when you view the
recorded video at full
screen. Many times
demonstrations are
performed and you are
shown a small little vid-
eo window. For surveil-
lance purposes if you
are trying to identify ob-
jects or subjects for evi-
dentially purposes they
are of limited value.
Average Storage Rate
Exposing Poor Video Quality

The bitrates below are a general average for acceptable quality video on a
4CIF image but actual could be signifcantly higher:
MJPEG 5 10Mbps
MPEG2 3 5Mbps
MPEG4 2 4Mbps
H.264 1 2Mbps
Again, a word of caution, the above bitrates DO NOT guarantee quality. In terms of storage,
a good rule of thumb is at 30 frames per second (real-time) you can equate Mbps to Gigabits
per hour. If you were using the above chart an H.264 stream at 30 frames per second (4CIF)
would roughly require 1 2 GB per hour for very high quality video, while MJPEG might require
as much as 10GB per hour. Again, this is very general and actual results will vary drastically. It
is safe to theorize MJPEG at low bitrates will not produce a quality image. For more accurate
information you can use a storage calculator at the Aventura Technologies website located at:
www.aventuratechnologies.com
Many people ask at
what bitrate am I go-
ing to get the best
image on a specifc
codec. The sub-
ject is quite debat-
able and you will get
arguments from both
sides of the aisle, as
manufacturers would
have you believe they
are capable of provid-
ing the highest quality
images at some in-
credibly low bitrates.
A lot has to do with
the amount of activ-
ity, quality of the data
and a host of other
variables. This is why
you need to be leery
of estimates of band-
width and storage
calculation. Storage
calculators are kind
of like the MPG stick-
ers on a new car; no
one ever seems to
get that mileage, even
going downhill in neu-
tral with a tailwind.

8
Further to note, with-
in the newer codecs
like H.264 there are
multiple profles. A
profle for a codec is a
set of features identifed
to meet certain sets of
specifcations of intend-
ed applications. There
are over a dozen pro-
fles within the H.264 co-
dec family, so obviously
some are better choic-
es than others for video
surveillance. While
the H.264AVC ini-
tial set of profles was
adopted back in 2003,
the latest update came
with the approval of the
H.264SVC (Scalable
Video Coding) standard
in November 2007, which
provides some unique
effciency for CCTV ap-
plications.
Codecs typically have
been a decade or more
in the making by stan-
dards committees be-
fore users begin to work
on their individual fa-
vor. Codecs are incred-
ibly sophisticated and
as such are created by
consortiums (Motion
Picture Experts Group
and International Tele-
communications Union)
of the most prominent
companies in the world,
from the Microsofts to
Intels. So when some-
one says they have the
ultimate proprietary co-
dec, they developed,
think twice. Most likely
it is just a favor of
something already in
existence.
To clarify one potential
misconception H.264
is also referred to as
MPEG4 Part 10 by the
Motion Picture Experts
Group, which is part
of the new advanced
codec standards. This
should not be confused
with MPEG 4 Part 2,
which most people as-
sociate with MPEG4,
which is the older gen-
eration codec using the
Huffman Algorithm.
Codec Profles
We have not dis-
cussed the qual-
ity of the coding. Just
because someone em-
barks on using a higher
resolution, higher bi-
trate and H.264 will
guarantee the perfect
image. Image coding is a
complex and highly spe-
cialized science. It is no
different than cooking in
some regards. We can
all start with the best of
intentions and the best
ingredients used in the
fnest restaurants. All
this means is you have
the potential to make
the fnest dish possible.
But then there are Mi-
chelin chefs and some
people who could burn
water. Video coding is
not add water and stir.
There is no magic pill.
It takes time, years of
research and develop-
ment, talent and a lot of
trial and error. So when
someone announces
they now have H.264 it
could be years before
they have it right, if ever.
Accordingly, an im-
mature H.264 solution
may look worse than a
mature older codec.
Quality of the Coding
9
The fnal aspect of
determining the video
quality is the hardware
by which it all comes
together. Some codecs
are clearly more eff-
cient than others and so
can be said for the hard-
ware components. It is
the hardware that actu-
ally performs the work of
processing the informa-
tion. You can begin with
best of breed all along
the route to the ultimate
in video experience but
you are only as good as
your weakest link. Com-
pression requires horse-
power; horsepower is
processing. The more
advanced the compres-
sion scheme the more
processing power re-
quired. Similarly, the
higher the quality of the
images, the larger the
fle size, the more data
involved and subse-
quently more process-
ing power needed. A
lack of processing pow-
er in a variety of areas
will cause bottlenecking
and degradation in qual-
ity.
Any shortfall in any area
of the above equation
may result in an inferior
result.
One challenge you are
going to encounter in
your quest for determin-
ing video quality is man-
ufacturers rarely pro-
vide you with adequate
information to formulate
even the most basic an-
swers. Even if a manu-
facturer is forthcoming
with specifcations on a
datasheet be careful to
note under what condi-
tions you can expect that
performance, remember
the MPG sticker. Its not
what they do tell you but
what may be omitted.
Ultimately, unless you
have scopes and a labo-
ratory, you have to rely on
visual observation.
Video compression
technology is a continu-
ously evolving science.
Advancements tend to
be incremental as op-
posed to exponential.
Whether its the ordinary
scheme or the hardware
that processes the in-
formation, it is an evo-
lutionary process. It has
taken two decades from
the lunch pail cellphone
to a touch screen Black-
berry.
Good luck in your quest
and always remember
caveat emptor, let the
buyer beware.
There are two methods by which to compress the audio and video data hardware or software.
Hardware compression tends to be the best solution if the hardware is sized correctly as it can:
Have an exponentially faster data rate throughput
Can offoad the compression task freeing up valuable CPU bandwidth
Reduce power consumption

Software compression instructs the PCs processor for encoding the video, whereas hardware
compression uses dedicated components and processing.
So when someone asks you how do you determine video quality (other than the naked eye)? The
answer is: Bitrate + resolution + codec + data quality + hardware = video quality
Hardware Factor in Processing Video
Conclusion

ATI Video Quality

Uploaded by

Copyright:

Available Formats

ATI Video Quality

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ATI Video Quality

Uploaded by

Copyright:

Available Formats

180 Adams Avenue, Hauppauge, NY 11788 | Phone: (631) 300-4000 | Fax: (631) 434-7000 | www.aventuratechnologies.

You might also like