NTSC versus PAL
in [Repair]
|
From: William Sommerwerck on 1 Apr 2010 09:13 "Die, die, my darling!" As both PAL and NTSC are basically dead systems (NTSC in the US, at least), there is little point in discussing their differences. But as Mr. Alison insists on displaying his ignorance in public, I'm going to, anyhow. The first color TV system approved by the FCC was a field-sequential (or frame-sequential -- I forget which) system proposed by CBS. It was developed by Peter Goldmark, the same man given credit for the modern LP phonograph record. (I say "given credit for", because there have been questions as to whether he was the principal designer.) The CBS system is a classic example of a design botched from the get-go. At that time (not long after WWII), there was no practical way to display three color images simultaneously with a single CRT. So Goldmark went with a spinning color wheel, a system that had been tried 25 years earlier for color motion pictures, and found wanting. The problems with such a system are obvious, but I'll describe them. One problem is that it requires three times as much film (or in the case of TV, three times the bandwidth). Another is that moving objects show color fringing. Then there was the problem of the spinning color-filter disk. A 10" TV would require one at least 2' in diameter. Imagine the disk needed for a 21" set! (Not to mention the noise, and the possibility, however remote, of decapitating the cat.) These obvious (and lethal) deficiencies didn't deter Goldmark or CBS, because they were in competition with RCA/NBC. The CBS argument was... Why limit TV to B&W? Why not /start/ with a color system, and be done with it? CBS pressed the FCC (as one writer pointed out, every sale of an RCA B&W TV would be another nail in the coffin of the CBS color system), and in 1950 the CBS system was approved, despite the fact it was wholly incompatible with the 480i system already in use. * David Sarnoff ("the most-nasty name in electronics") was naturally upset. RCA had to make CBS look bad, while completing development of their own color system. Sarnoff gleefully pointed out that the CBS system was "mechanical", and subject to all the limitations accruing thereto. Though this was literally true, it overlooked the fact that one can have all-electronic field-sequential color. But -- on the other hand -- CBS had nothing other than a mechanical system to offer. RCA was working on a "dot-sequential" system. Each line of the image was divided into 300 (or so) pixels **, with red, green, and blue samples alternating. This system worked fairly well -- it produced an acceptable picture on B&W sets. But (for reasons I don't remember) color receivers had problems displaying B&W images. As color receivers would (initially) be used mostly for B&W viewing, this was not acceptable The breakthrough came when engineers at Hazeltine and GE remembered Monseuir Fourier, and recognized that sampling the colors was equivalent to a "continuous" signal at the sampling frequency. They "slipped a note under RCA's door" (so to speak), and NTSC/PAL came into existance. The color information was transmitted on a subcarrier whose sidebands were interleaved with the luminance sidebands, to minimize interaction. *** "...complete with bad commercials that repeat all night, both in compatible color and black and white." -- Stan Freberg The brilliance of NTSC/PAL is that their signals produce as good (or better) an image on B&W sets, and display excellent color on a color set -- without making any existing equipment obsolete, and without requiring additional bandwidth. So... why is NTSC "better" than PAL? For one thing, it has "better" and "more" color. Although the original NTSC proposal used red and blue color signals of equal bandwidth, it was recognized that this didn't fit with the way the eye actually sees color. It turns out that for a 480-line system displayed on a 21" tube, the eye sees full color (red/green/blue) only to about 0.5MHz. From 0.5MHz to 1.5MHz, the eye sees only those colors that can be matched with red-orange and blue-green primaries. **** The system was therefore changed to the red-orange/blue-green and yellow-purple primaries, the former of 1.5MHz bandwidth, the latter of 0.5MHz bandwidth. PAL uses equal-bandwidth (1.0 MHz) red and blue primaries. If an NTSC set fully demodulates the 1.5MHz color signal (most limit it to 0.5MHz to make the set cheaper), more of the original image's color detail will be displayed (though this will be visible mostly in graphics). Much has been made of PAL's phase alternation, especially its supposed ability to eliminate the need for a tint [sic] control. (It should be hue control.) When was the last time you adjusted the hue control on an NTSC receiver? 30 years ago? This issue is confused by two factors -- the differences between European and American distribution systems, and their studio standards. If the transmission network has constant group delay, the hue setting should be set 'n forget, and never need to be changed. The American system had good group-delay characteristics -- the European did not. So switching channels could require twisting the hue knob. But that's not all there is to it. Non-linear group delay changes the colors in a way that cannot be corrected simply by adjusting the hue control. All the colors cannot be "correct" at the same time. The advantage of PAL is that these color errors "flip" with the phase, and are complementary -- the eye "averages" them to the correct color. So what's wrong with that? Well, the averaging also reduces saturation. (Mixing an additive primary with its complement pushes it toward white.) With severe group-phase error, the image shows bands of varying saturation. (In NTSC, there are bands of varying hue.) The other point of confusion is that, for many years, US broadcasters didn't pay much attention to signal quality. Cameras weren't set up properly, and burst phase wasn't properly monitored. So when you changed channels, you sometimes had to change the hue setting. Broadcasters finally got their acts together, and color quality has, for some time, been pretty consistent from channel to channel. In short, PAL's phase alternation is an advantage with transmission systems having poor group-delay characteristics -- a problem that did not exist in the US. In every other respect, it is inferior to NTSC. All of this is true, to the best of my knowledge. Corrections and additions are welcome. * Some dishonest manufacturers sold B&W TVs with a "color converter" jack on the back. It wouldn't have worked, because these sets didn't have the required IF bandwidth (AFAIK). ** No, the term didn't exist at the time. *** Some interaction is visible with objects having fine B&W detail. The set "misinterprets" this detail as color information. **** This is why two-primary color-movie systems (such as the original Technicolor) could give acceptable -- though hardly great -- results. -- "We already know the answers -- we just haven't asked the right questions." -- Edwin Land
From: William Sommerwerck on 1 Apr 2010 09:59 >> As both PAL and NTSC are basically dead systems >> (NTSC in the US, at least)... > What about PAL and NTSC videos, DVD/Blu-ray? > When did they die? I meant as broadcast systems. I have plenty of NTSC DVDs, and analog cable signals are still NTSC. Blu-ray is its own format (1080p/24 or 1080i/60).
From: Geoffrey S. Mendelson on 1 Apr 2010 10:04 Meat Plow wrote: > What about PAL and NTSC videos, DVD/BluRay? When did they die? Technically video tapes are not NTSC or PAL. They have separate tracks for luminance and chroma. The recorders all stripped them apart before recording them and put them back together when playing them. There is no technical reason not to build a video player with a digital output, which digitzes the signals and presents them as an digital data stream, with out actual NTSC nor PAL encoding. The field/frame rate would be the same as the source material, but that's not the same thing. The same with DVD's and BluRay. The data is encoded using MPEG compression, which has separate information for luminance and chroma. It can be rebuilt as red-green-blue pixels without ever going through NTSC or PAL. As reg-green-blue cameras become more common, I expect that there will be an eventual shift to rgb encoded data, but that's a long way off. Geoff. -- Geoffrey S. Mendelson, Jerusalem, Israel gsm(a)mendelson.com N3OWJ/4X1GM New word I coined 12/13/09, "Sub-Wikipedia" adj, describing knowledge or understanding, as in he has a sub-wikipedia understanding of the situation. i.e possessing less facts or information than can be found in the Wikipedia.
From: William Sommerwerck on 1 Apr 2010 11:00 >> Blu-ray is its own format (1080p/24 or 1080i/60). > Oh? So Blu-ray will play on a 50 or 60 Hz system > and the audio will be in sync? Good question. I haven't looked to see whether a Blu-ray player can be set to deliver an SD signal. I don't think it can.
From: Arfa Daily on 1 Apr 2010 21:45
<snip> > > This issue is confused by two factors -- the differences between European > and American distribution systems, and their studio standards. > > If the transmission network has constant group delay, the hue setting > should > be set 'n forget, and never need to be changed. The American system had > good > group-delay characteristics -- the European did not. So switching channels > could require twisting the hue knob. But that's not all there is to it. > > Non-linear group delay changes the colors in a way that cannot be > corrected > simply by adjusting the hue control. All the colors cannot be "correct" at > the same time. The advantage of PAL is that these color errors "flip" with > the phase, and are complementary -- the eye "averages" them to the correct > color. I don't think that is actually true. It's been a lot of years since I studied PAL decoding at college, but as far as I recall, the averaging is done totally electronically, courtesy of the PAL delay line. This is a glass block delay line of one scan-line period, so if you run a direct and a delayed path side by side in the chrominance channel, and then sum the outputs of both, you arrive at an electronically averaged result of two sequential lines, with any phase errors balanced to zero. This has nil effect on the overall colour saturation, as this is controlled by a) the ACC circuit, and b) the user saturation control > > So what's wrong with that? Well, the averaging also reduces saturation. > (Mixing an additive primary with its complement pushes it toward white.) > With severe group-phase error, the image shows bands of varying > saturation. > (In NTSC, there are bands of varying hue.) > <snip> |