Obviously my experience of video capture hardware is quite limited
because I don't have unlimited money and I have only dealt with
a couple of capture cards. Also keep in mind that hardware is improving
all the time so when you search there may be more options than when
this article was written. You will also be limited by how much you
are willing to spend on your hardware and what you need it to do
with it. In light of these facts I cannot recommend any particular
card or TV tuner to use. I suggest you search around and read some
reviews. Toms Hardware Guide is where I usually go to check out
the latest info on hardware so check him out at: http://www.tomshardware.com/.
The expensive dedicated video capture cards designed for professional
video editing are obviously the best quality but are usually out
of most peoples price ranges. If you are very serious about it,
there are some good quality and yet nicely priced cards sold for
both home and proffesional use by Pinnacle
Systems such as the DC10plus or perhaps even the DC50.
The less expensive Graphics Cards like the Nvidia GeForce, Matrox
and ATI All In Wonder, and more recently the Radeon are good cheap
alternatives. Just remember, despite claims of professional video
quality, these all-singing-all-dancing graphics cards almost always
produce worse results than a dedicated capture card. Of these graphics
cards I think the current opinion is that the Matrox is the best
quality, then the ATI (almost as good), then the GeForce and Voodoo
ranges respectively. Many of these cards give the option of either
Mpeg-2 or MJPEG hardware capture. Mpeg-2 gives best compression
and MJPEG best quality especially if it is hardware based like the
Matrox cards. If it doesn't offer any hardware capture option
then its probably no better than a normal graphic card with a cheap
TV Tuner stuck to it!
Basic TV Tuners cards are the cheapest option if you don't want
a big upgrade and come with many low priced graphics cards. Because
they are so cheap people assume they are not much good. Truth is,
if your computer is 'man enough' then you will often be able to
get equal or superior capture to those all-in graphics cards! In
fact, I'd go as far as to say that if you have good enough hardware
then you may be able to improve the quality of many graphic card
captures by turning off hardware compression and using it like a
TV Tuner instead!
What do I mean by 'man enough'?
The most perfect graphics format for video capture would be 'uncompressed
AVI' with 'uncompressed PCM audio'. More specifically that is one
uncompressed bitmap image for every frame along with a Wave file
played in time to it. Many TV Tuners and Capture Cards will offer
such an option or an equivalent lossless compression. But, especially
at high resolutions most computer hardware will not be able to handle
such an influx of data and you will get bad quality!
In light of this, most graphics cards use video compression as
a standard part of capture to avoid problems. The ATI graphics
cards may say something like 'AVI 1.0' but don't let the name AVI
confuse you, it is actually a highly compressed format! Your
hard disk can only save so much information per second. If you are
capturing more information than your hard disk can take, you will
loose frames and get bad, jerky video. To illustrate this effect,
just try copying a 500MB file from one folder on your hard drive
to another (don't cut and paste, just copy and paste). How
long does it say it will take to copy? 3 minutes? 5 minutes? longer?
So what happens if you are capturing 30 seconds of video at 640
x 480 in uncompressed AVI? Only 30 seconds of this format will take
over 500MB of hard disk space! Your hard disk will just not be able
to keep up! And this also highlights the other huge downside of
uncompressed AVI - it takes a huge amount of space to store! For
all these reason almost all video capture cards will offer to compress
the video file directly to MJPEG, Mpeg-1 or Mpeg-2. But, if set
correctly, any of these formats can give us VHS quality video.
A TV Tuner must use your computers Hard Disk and CPU power to compress
or store the AVI in real time, so you will need to know that your
machine is able to handle it. You certainly need a fast hard drive
if you are capturing at 640 x 480. About 7200 RPM should be enough
or 5400 RPM for 352 x 288 capture. I think Seagate do the best Hard
Disks. If money is not a problem get SCSI HDD's because you can
buy ones that reach speeds of 10000 RPM which will give you unlimited
power in the storage department! On the other hand some of the newer
IDE's are able to handle 25MB/s or more without trouble and are
much cheaper. You will always need lots HD space, you can never
have enough with video capture and 20 GB or more is probably a conservative
estamate. I also suggest at least a 300Mhz CPU and an absolute minimum
of 64MB RAM - the more the better of course.
Please bare in mind that all this is a rough estimate. Although
most PC's exceed these requirements today, there are so many factors
including slow graphics cards or motherboards that it is difficult
to give exact requirements. If you are unsure I suggest you just
invest in an ATI All in Wonder card or something like that rather
then just waste money in the hope you can upgrade your system and
capture better quality without one - you have been warned!
Video Capture Formats
MJPEG: will usually give the highest quality video especially
if its hardware based capture. Instead of using bitmaps for each
frame it uses compressed JPEG's images. As you know the quality
of JPEG is almost as good as Bitmaps but very highly compressed.
If you are capturing with a TV Tuner this is just about the only
choice. The Picvideo MJPEG codec is the best quality I know of for
real time capture.
Mpeg-1: Mpeg-1 capture is not fast enough to do in real
time unless it is hardware based capture. I know there are utilities
that will capture in Mpeg-1 in "real time" but the quality
is crud even on a machine with a very high CPU. Lets face it, why
does it take many hours to compress to Mpeg-1 or Mpeg-2 from an
AVI with Panasonic Encoder or TMPGEnc etc., if you could do it in
real time with the same quality?! Most of these real time captures
use the ligos codec to compress, but the Ligos Mpeg Encoder cannot
capture real time, so I think that says it all!
Mpeg-2: is the next highest quality to MJPEG But there are
many problems with this format. Again it must be done by hardware
compression because software compression is far too slow to produce
any good quality. It is also a format that is very hard to edit.
VirtualDub will not open an Mpeg-2 file unless we use Flask Mpeg
or AVISynth to frameserve it. This makes editing very slow indeed.
Note: As you probably know by now, Mpeg-1, 2 & 4 compress like
JPEG, but to increase effectiveness they also compare two or more
frames in a sequence and only save the differences. This means they
contain frames with only parts of the full picture in them. These
half frames are called B frames and P frames. The full JPEG-like
frames are called I-Frames. Many Mpeg capture cards such as the
ATI All In Wonder will offer to capture as I-Frame only - in other
words, as 'full frames' only. Capturing this way will increase the
quality a lot but it will take up almost as much space as an MJPEG.
This is because, with all the technical difference aside, I-Frame
is basically the same compression as MJPEG. Other advantages are
it takes less CPU power to compress and I-Frame only Mpeg-1 is fully
editable in VirtualDub! You will also have the advantage of Mpeg-2
audio compression so you can capture for a little longer than MJPEG.
If you are going to capture this way I suggest you turn off all
motion compensation settings too.
Mpeg-4: is the best compressed format and you can use VirtualDub
to capture directly to Divx or MS Mpeg-4. But again this will not
produce good quality results because, like the other Mpeg formats,
it takes too long to compress. It would only be useful if you intended
on very low bitrates where quality was not a high factor.
Framerates & Resolution
If you are going to edit and/or compress a video using VirtualDub
or any video editing package then the better the original video
the better the final result will be. Try to use the highest settings
you can without going overboard. For example, there is no need to
struggle to capture a video at 640 x 480 pixels if you are going
to resize it to 352 x 240 because not only will this take a lot
of time to resize, but it will take twice the hard drive space!
The difference in quality between the two will be very small so
why not just keep to 352 x 240 instead.
Concerning resolution, you should always keep in mind that VHS
video is roughly 320 x 240 pixels. Even so it may be best to keep
to 352 x 288 PAL or 352 x 240 NTSC if you wish to follow VCD standards.
This fact sounds really weird because a TV's active picture has
480 lines and a DVD resolution is 720 x 480 pixels. Some, in a desperate
attempt to capture full quality video, will make DVD resolutions
of 720 x 480. Let me tell you now that these captures are not
even close to DVD quality no matter how good the capture! Neither
have I just picked the size 320 x 240 out of thin air either. The
reason for this is based on a value of clarity that mathematicians
call the Kell factor. The Kell refers to the amount of detail displayed,
in this case, on an analogue TV. To simplify a rather complex subject
down to the bare bones we could say that a TV picture is very fuzzy
and a TV "pixel" works out to less than half the clarity
of a computer pixel! This means a computer resolution of less than
half that of a TV is needed to store the same information. What
is worse is the fact that, if the quality of your VCR or recording
is not very good, or if you are using a camcorder, you are probably
looking at little more than 280 x 210 pixels or something like that!
Note: I haven't done the exact mathematics on this yet, but
a DVD resolution must work out to over double the quality of VHS
video! Which brings up the subject 'why do we need high definition
DVD's if they are already so close to HDTV quality, or have I just
confused myself on this one?'
What does this mean? That capturing at 352 x 240 will give full
VHS resolution? Possibly, but its not quite that simple! It may
depend on how your capture card works. If it disregards lines then
actual information is missing when it resizes. What would be needed
is to construct a full image from all 480 TV lines and resize
them down to the equivalent PC pixels. For example, if you want
full detail you can capture at 352 x 480 and then resize precise
Bicubic with VirtualDub to 352 x 240 (or 288). This should preserve
most of the fine details and I personally think this is a good alternative.
As you can see, assuming we need 480 pixels high for total quality,
the pixels across still remain the same, about 352 pixels. In light
of this problem Mpeg-2 was developed with yet another compromise
which is supported by the ATI capture card. It can capture at 352
x 480 and then when played back stretch the video back to 720 x
480 to keep the correct aspect ratio. This takes much less space
since it is resized only when it is played back and stops
all the arguments.
I have explained the problem of interlace in great detail Here.
So I don't need to spend too much time on it. Basically there is
little you can do to solve it other than using VirtualDubs deinterlace
filters. How bad the problem is, or if you get it at all, will depend
on how your graphics card captures. If you only capture one field
(i.e. every other line making 240 or 288 pixels) then usually you
won't have the problem. Sometimes you will be given the option to
capture both fields (i.e. 480 or 576 pixels) and blend both fields
together, check the settings of your capture device. There is a
rumour going around that capturing to Mjpeg or Mpeg-2 will solve
interlace problems. This is not true, at least with every set up
I've tried so far.
Since PCM Wav audio is not too large and yet still the best quality
I suggest that you use that. It puts a large strain on your computer
to compress to Mp3 or WMA audio and there may be more problems with
audio synch if you don't. I suggest you also use 44100Hz 16 bit
stereo which is CD quality audio and best for music. This is also
the most compatible format. You can, of course, use any other setting
such as 22050Hz if you prefer and this will to save some space at
the expense of a little quality. But then again if you are going
to compress the audio anyway I wouldn't worry too much.
There is never any need to capture higher than 30 frames per second.
I'd also recommend that you stick to 25fps for PAL TV capture and
29.970fps for NTSC TV capture. This is because most capture cards
are designed for this kind of capture and you may get bad results
or synch problems if you don't.
Both VirtualDub and most capture cards will tell you if your system
cannot capture all the frames you need per second for smooth playback.
I think that if your computer cannot take the strain and starts
dropping frames it is better to capture at a lower resolution! I
would always sacrifice picture size (resolution) in favor
of a high framerate unless the picture was very small. If you have
the framerate dropping problem with an NTSC TV which is 29.97fps
then you could try capturing 25fps instead, but, as I said, this
may cause problems. Also if you intend on making a VCD at 25fps
then you should also capture at 352 x 288 instead of 352 x 240.
If your computer starts dropping frames you will almost always
get audio synch problems and annoying jerky playback problems. Its
easy for the novice to think that if they get dropped frames at
25fps then they should set it at 30fps to compensate. Then, even
if some frames were dropped, we would still get smooth playback.
This is a big mistake! If your machine is dropping frames at 25fps
then asking it to capture more per second puts even more strain
on it and you will get much worse quality. The only answer is to
lower the resolution or lower the framerate or upgrade your system
Now you know the basics and have the equipment, I will explain
in the next article how to set up and capture using VirtualDub and
your capture device. Additionally, it is always a good idea to make
some test clips before you capture and see how they play before
you do a long capture. You will also need to watch the 4GB limit,
Windows 98 cannot save single files larger than 4GB each so that
restricts what we can do even more. If you have Windows 2000 this
shouldn't be a big problem. Nevertheless all these issues will be
dealt with in following articles.
Duplication of links or content is strictly
prohibited. (C) NICKY PAGE 2000