50 What should I consider in buying a video capture card?
Description
This article is from the comp.sys.ibm.pc.hardware.video Frequently Asked Questions, by Michael Scott with numerous contributions by others. (v1.0).
50 What should I consider in buying a video capture card?
There are several factors that will determine which video card is the best for your purposes. It will depend on the number and type of video inputs, AD (Analog to Digital) conversion and system noise, frame rate, video overlays and whether video capture is to be integrated with other software.
Grayscale and colour video capture cards are available. Grayscale cards are usually 8 bit, but some are available for 12 bit conversion. This means that the video intensity is sampled temporally, measured as a voltage, then divided into 2^8 (2^12) or 256 (4096) discrete levels. 8-bit provides enough gray levels for most applications and approaches the noise threshold in most video systems. Noise can be reduced in this or any colour system by frame averaging.
Colour capture cards are available in 16, 24, 32 and more bit models. They convert the individual red, green and blue video streams into digital values separately, each stream being treated similarly to grayscale digitization. 16-bit cards discretize RGB into 5, 5 and 6 bits, and so can record 65535 different colours. 24-bit cards provide 8 bits for each pixel for a total of up to 16.7 million colours. 24- bit cards are also called Truecolour because most humans can distinguish 5-6 million colours. At 16.7 million, 24-bit colour can display more different colours than anyone can perceive. Cards that provide 32 bits or more of colour depth are usually Truecolour cards with overlay capabilities. The overlay planes (8 bits in the case of 32 bit) can be used to contain text or graphics overlays, or can store depth information (z-buffer). In addition, extra video memory can be used to double buffer the incoming digitized signal, up to doubling the frame capture rate.
Video capture cards can often digitize different image sizes , though the most common is 640x480. 640x480 is the maximum image size that is meaningful for NTSC video signals. Keep in mind that while the horiz- ontal resolution of a television signal is quite high, the vertical resolution is limited to the number of scan lines displayed. A VCR provides ~250 lines, while S-video or laser disc provide over 300. Many video cameras provide more - closer to the 525 that the NTSC standard can handle. This means that the capture card has to integrate vertically (or subsample) to get 480 pixels vertically. This introduces a smoothing effect in the vertical direction and results in a less sharp picture. Capture cards are available which will digitize larger images, but they require special-purpose video equipment to be used to any advantage. Ensure that the resolutions you use maintain the screen aspect ratio.
Various types of input signals can be digitized including NTSC, PAL, S-video and RGB. Some cards can handle all types, but most of the less expensive ones can only understand NTSC. Boards that can capture separate RGB signals can often be used to connect up to 3 grayscale video inputs.
Many video cards come with simple frame capture programs, but if you are planning to integrate video capture with other operations on the computer, like collecting data from an AD card, adding text data as an overlay or changing video-in channels on-the-fly, you will have to do some programming. In this case you will need good programming libraries in a language you are familiar with for the video card. Some companies include libraries with their cards, but most charge extra. Most often libraries, when available, are for C or BASIC, and sometimes Pascal.
Continue to:
My Books
