The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: Grahame Kelly
Subject: Re: Four (4) TV images on one computer monitor
Date: Sun, 17 Nov 2002 01:01:49 GMT
NNTP-Posting-Date: Sun, 17 Nov 2002 12:01:49 EST
>> Do you know if there are any video capture boards that will allow us to
>> capture four (4) TV shows at the same time?
> I don't think so.
> There shouldn't be a problem sticking 4 in, though the drivers may not
> cope well. I know the linux BT848 drivers (the chip used in most low-end
> capture cards) will cope with at least two cards, I don't think there will
> be a problem with more than two.
> The resolution may be limited by the PCI bus bandwidth, you probably can't
> do better than around 1024*768 total resolution, as the PCI bus can't move
> much more data than that.
> But, 4 windows of 500*300 or so should be OK.
I run 4 x PCI TV (BT484) type cards under Linux concurrently without any
problem (read = As I require them to run to capture and detect the noctural
animals I am studying). One point to note is all the video capture cards
MUST BE the same type and model to be supported by the driver. You cannot
mix and match. Driver won't work correctly.
That said, you will have problems with a 4 channel card (I have one) in that
to get 4 separate channel frames into one viewable picture size ( 4 x
channels with each channel picture 256x192 pixels = 1024x768 - std. SVGA )
the s/w driver has to switch between channels (say at 0.5 sec intervals) to
capture a frame BUT be aware that each channel Mux requires a settling/sync
time probably in the order of 5->10 frames [which equates on 50Hz System to
another 0.5 second] so you will only get each channel view
updated every 1 second or so, sequentially not concurrently).
There are commercially available PCI cards that have quad capture circuits
that allow concurrentcy > 10fps viewing - BUT the viewing problem now is
limited by the PCI bus speed, the type/style of AGP card (it its off the
PCI bridge on the motherboard because you are viewing while capturing) and
other factors. I have found that saving captured frames to disk, then using
a Linux tool like VLoopback and CamStream allows the best results by giving
priority to the capture to disk process and then the viewing process.
See: http://www.exploits.org/v4l/ for more on the subject.
In my tests a 1GHz SMP (2 x PIII CPUs) performed poorly on a Gigabyte
motherboard because the PCI bus and PCI bridge where unable to maintain
throughput conditions controlling 4 video capture cards.Normally I had
experencied on an older PentiumPro SMP system the best PCI throughput even
though the PPro CPU was only 200MHz, beating hands-down the PIII SMP
system. Goes to show motherboard design is major decider on performance -
and it all depends on the applications to be run. On another system, a
uniprocessor PIII/1Ghz Gigabyte motherboard, throughput was sustained.
On another P4/1.5Ghz Apton motherboard throughput was less than PIII/1Ghz
Gigabyte momo. So, its all up in the air unless you have the resources and
time to test and select.
My results - Every motherboard/PCI combination will have good and bad points
on sustained throughput. Best to go for (in order of priority):
A. Fully PCI 2.2 compliant motherboard, 2.1 compliance is less effective.
B: 66MHz or greater PCI bus clocking,
C: PCI 2.2 compliant Quad Video Capture Card with onboard buffering memory.
D: A AGPx4 spec'd or better graphics card for viewing.
E: Fast-SCSI III and/or Fast IDE2/66 drives on seperate controllers.
F: Best speed memory SIMMS/DIMMS you can get to assure low memory latency.
G: S/W side, Drivers that allow double or triple frame buffering.
Applications that make use of priority requests.
My 2c guys.
Hope it helps.
Grahame (at) Wildpossum (dot) com
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup