awpaeth@watcgl.waterloo.edu (Alan W. Paeth) (04/12/88)
I've noticed that many commercial video frame stores no longer operate
by digitizing the base-band video signal at four times the color subcarrier,
but instead use the "CCIR601" component video standard. Can anyone explain
what this is? From the simple sketches I've seen, it looks like there are
three data channels, which are almost certainly *not* RGB, as one channel
is labellel "luminance", the others "chrominance" (this looks suspiciously
like RGB recoded as YIQ data, but no doubt there is other trickery at work).
/Alan Paeth
Computer Graphics Lab
University of Waterlooal@cs.strath.ac.uk (Alan Lorimer) (04/14/88)
In article <3996@watcgl.waterloo.edu>, awpaeth@watcgl.waterloo.edu (Alan W. Paeth) writes: >I've noticed that many commercial video frame stores no longer operate >by digitizing the base-band video signal at four times the color subcarrier, >but instead use the "CCIR601" component video standard. Can anyone explain >what this is? From the simple sketches I've seen, it looks like there are >three data channels, which are almost certainly *not* RGB, as one channel >is labellel "luminance", the others "chrominance" (this looks suspiciously >like RGB recoded as YIQ data, but no doubt there is other trickery at work). I have no details of the precise standard you request, but I can tell you what I *remember* about component video. The Three channels you mention in component video are Y, U and V. The Y channel is the Luminance channel and contains enough information to display a presentable monochrome picture. U and V are colour difference signals, and carry enough information to allow R, G and B levels to be calculated. A few numbers: Y = 0.3R + 0.59G + 0.11B U = Y - B V = Y - R These sums are usually done in the analogue world. U and V only carry colour information hence are labelled "chrominance" One of the problems with NTSC and PAL coded signals is that in decoding the signal the picture often shows patterning in areas of high frequency colour, this is because coded colour bandwidth is limited (to about 1.3MHz in PAL I think), and also because the high colour signal will actually interfere with the luminance signal (which it does anyway, but it becomes more noticeable). Note the green/magenta junction on colour bars. If it is possible to keep the signals in component form or RGB the coding process is avoided together with this degredation of the signal. Ok, so why YUV and not RGB? If you digitise RGB, you require to treat all three channels identically ie same resolution/bandwidth/grey scale. With YUV, the important signal is Y - spend all your money there and get a good quality stored signal. U and V only carry colour information, don't have such a high bandwith and don't require the same number of bits for an acceptable signal. I vaguely remember the details of one frame store I encountered which had 12MHz sampling at 6 or 8 bits for Y, but used 1/2 the number of bits for each of U and V. Al. -- UUCP: ...!seismo!mcvax!ukc!strath-cs!al DARPA: al%cs.strath.ac.uk@ucl-cs JANET: al@uk.ac.strath.cs