Differences of capture boards?

fb99999 · Post by **fb99999** » Mon Nov 08, 2004 2:40 pm

There is a wide variety of differences in prices of capture cards, but what is not clear to me is what I am gaining by purchasing the more expensive cards. For example, there is the KWORLD 8 Port card found here at new egg for $134:

http://www.newegg.com/app/ViewProductDe ... 122&depa=0

There is also the Linux Media Labs 4 port card, for $410:

http://www.linuxmedialabs.com/product_d ... prodid=310
(the site is down now, but this is the link I bookmarked)

The KWORLD card apparently captures in 320x240, while the Linux Media Labs card captures in 640x480. I saw in the FAQ that "Sometimes a large image is just two interlaced smaller frames so has no real benefit anyway", so in this case, is the larger image really just a waste of space, or is there real benefit? Also, will the more expensive card give a 'clearer' picture?

In addition, it looks like the LML card can capture at a higher frame rate, probably because it includes 4 BT chips. Is that correct? If so, that would explain some of the extra cost.

Thanks in advance,
fb

lazyleopard · Post by **lazyleopard** » Mon Nov 08, 2004 6:51 pm

Frame-rate is going to be the main difference. The KWORLD card appears to have a single video digitising chip, so don't expect high frame rates if you use it, even if you only use 4 of its inputs. Also, expect to need the same type of camera on all the inputs, as most settings (resolution, colour, brightness, contrast, etc.) happen at digitising chip level, not at individual camera level...

fb99999 · Post by **fb99999** » Mon Nov 08, 2004 11:19 pm

Ok, that makes more sense. Basically, the added cost comes from the additional digitising chips. Captured image quality is likely to be the same. So, for the extra cost one gets the following benefits:

(1) 4x the frame rate
(2) Ability to have different settings per input/chip

I'm still unclear, however, on resolution. The more expensive cards advertise a higher resolution (640x480). Is this 'Real'? i.e. will I see more detail in my captured images, or will it just quadruple my image size? There is an FAQ that says "Sometimes a large image is just two interlaced smaller frames...". So, with a card like the Linux Media Labs card, or the Spectra8 from iTuner, does one receive more detail in the larger sized images?

Thanks in advance

lazyleopard · Post by **lazyleopard** » Tue Nov 09, 2004 9:14 am

fb99999 wrote:I'm still unclear, however, on resolution. The more expensive cards advertise a higher resolution (640x480). Is this 'Real'? i.e. will I see more detail in my captured images, or will it just quadruple my image size?

An interlaced image is two frames 1/60th (or 1/50th) of a second apart, each half-frame having half of the vertical resolution but the full horizontal resolution. For static objects you will get better resolution. For fast moving objects this can present some interesting effects... Search the forum for "interlaced" as I'm sure there are some threads with examples somewhere...

fb99999 · Post by **fb99999** » Wed Nov 10, 2004 3:28 am

Ok, I think I am understanding this better. When an image is interlaced, basically 2 passes are made on camera, one pass for odd lines, one for even, and these are spliced together to form one image. I was able to demonstrate this by capturing snapshots under WinTV (don't have my Linux box running yet). Yes, motion definitely makes things interesting!

So, for example, if I decided to capture at 640x480, and the board was capturing Interlaced, I will get more detail, but if there is fast motion, the image will be blurred.

But what determines when a card captures in interlaced, versus non-interlaced? I saw some references in the formus that 'larger' sizes get captured Interlaced, while smaller images get captured from one camera pass, but I didn't see an explanation of what controls this. Can someone explain this?

Thanks in advance for your continued help!

lazyleopard · Post by **lazyleopard** » Wed Nov 10, 2004 2:46 pm

Interlacing is a standard part of analogue video signals (PAL/SECAM/NTSC as used for TV, CCTV, etc.), so if you're using a standard video input and a video capture card, and you go for a larger frame, then you'll get an interlaced image. I think the cut-in points are at vertical resolutions greather than 240 for NTSC and 288 for PAL.

Digital network cameras do things differently...

fb99999 · Post by **fb99999** » Thu Nov 11, 2004 2:05 am

Thank you very much, that clears things up 100%. Your help is truly appreciated!

-fb