[AG-TECH] Osprey 440 Cards, Matrox QID Video Card, Recommended Soundcard

Richard Naylor richard.naylor at citylink.co.nz
Tue Jun 21 22:36:50 CDT 2005


At 01:58 PM 6/20/2005 +0800, Peter DeSantis wrote:
>Didnt know there was a 4 channel osprey, gonna check that one out myself

I have an Osprey 440 - its one of the first 20 pre production units and is 
loaded with beta drivers. Previously we have used 4x Osprey 100 and lately 
Spectra-8 cards for video capture in AG nodes running under windows. The 
Spectra 8 is a nice card, but does have an annoying habit of asking which 
port goes where every time you go into a venue.

The Osprey 440 is a PCI card with 8 audio and 4 video inputs, all with 
independent capture chips. The card has 4x BNC connectors and 8x RCA 
connectors (via a fan out connector). On the board there are binder posts 
for a further 3 video inputs per capture chip and I think a similar number 
of audio inputs. Certainly S-video and balanced audio is available on these 
posts.

The card installed really well and after a reboot, the inputs show up as

video1A
video1B
video1C
video1D

I suspect the newer drivers will allow greater control and selection of 
inputs. The drivers certainly had many features that are very useful like 
logo insertion, proc amp controls, etc. There are separate controls for 
both the captured video and the displayed video.

Simulstream was available for each input. I guess this would allow 
simultaneous encoding with 6 different parameter sets, including choice of 
input. This could be awesome at a large event where several or many 
different stream products are being delivered from several video sources, 
mixes or feeds. Currently I use racks of PCs all with Simulstream. Mulitple 
CPUs gets around the big shortage - CPU capacity as well as providing lots 
of reliability, but simulstream is great for providing a really easy way of 
doing a multi-bit rate webcast.

My first test was to run it under MS encoder. I ran 4 instances of encoder 
and it performed well. Of course the capture was very quickly constrained 
by CPU capacity. I was running under a 2.8GHz Celeron in an Iwill chasis, 
with the idea of making a small but very capable capture machine. I did 
notice that under MS encoder, I got errors from ks.sys when accessing the 
sound mixer for each stereo input pair. However, the mixer did still work, 
so I suspect it is a beta driver error.

As it was late, I didn't run a long encoding session to see how stable it 
was. That is underway right now.

My next test was to run it as an Access Grid node. Sadly this wasn't so 
great. The first input loaded as a "video Producer" like a dream. The 
second one produced errors "C++ error" and complained that you can't access 
the same device twice. Third and fourth VideoProducer loads did the same 
thing. However, at least two of them then worked !! Of course the next time 
I tried it - only one input worked and only one has worked since then.

RAT didn't see any of the sound inputs, so the motherboard card was used.

I did try and get the VfW drivers loaded and going, but started to descend 
into windows driver hell. I backed out of that mission.

So in summary - its a great card with great promise. For encoding with 
plenty of CPU it will be terrific. As an Access Grid node, right now it 
needs a little work done on the WDM drivers. I suspect the final release 
drivers will be great........I'll just have to wait.

Rich





More information about the ag-tech mailing list