free to view it up close in any sized room).
It's kinda like horsepower. More is always better. So the real question is, how much is enough?
Your bare working minimum should be 480p (480 lines, progressive scan), which is a progressive version of the 480i image that natively comes from most DVD players. Fortunately, 480p is easy to get nowadays, and at very reasonable cost (it is built into many newer DVD players). And 480p might even be enough for you. We wouldn't have thought so, until we saw Arcam's outstanding results, achieved with their DVD player feeding simple 480p into a 42" plasma display.
The next commonplace step up in resolution is the 720p native to the new DLP projectors, and also 768p or thereabouts which is native to the new (mostly larger) plasma displays. Scaling DVD sources up to 720p or 768p is especially worthwhile if you're using larger display screen sizes and/or are sitting close to the display. Or, as Revox proved with the 852p resolution of their new 32" plasma, even on smaller displays every extra degree of resolution can produce an extra degree of visual believability. And of course, if you're also using HDTV as a source, scaling it to 720p or 768p gives a perceptible advantage over a 480p display.
Is there an upper limit on how much resolution is beneficial and can be visually appreciated? Not yet. Faroudja presented a superb demonstration of their new, forthcoming (March) $10,000 professional box, the Digital Cinema Source, which up-converts 1080i to 1080p. They played some Hollywood film masters recorded at a true 1080i, and some live HDTV feed at 1080i, then demonstrating the visible improvement when they up-converted the picture to 1080p. The 1080i picture was of course excellent to begin with, looking like good film thanks to the expensive projectors, but it was still just a picture. The 1080p conversion seemed more like looking out a window at the real world. Upon direct comparison, there were subtle traces of noise and artifacts in the 1080i original, which also slightly degraded the integrity of color saturation. With the Faroudja 1080p up-conversion, these subtle traces of noise and artifacts disappeared, and colors became more truly uniform and saturated, more like looking at the real thing, more believable.
Tomorrow's digital media will include blue laser DVD and more prevalent HDTV, both of which will be able to show off the highest resolution displays to full advantage. So any extra money you invest today in a higher resolution display will pay even bigger dividends with tomorrow's source material.
-- Signal Processing
There's something to be said for the tyranny of dictatorship. When the CD format was first developed, Philips seized the reins, and with an iron fist has dictated through the years that all CDs must obey their red book standard. It might not have been the best standard at the time, and we can certainly do much better now, but at least it was a standard, so every CD player everywhere could read and process the signal coming off every CD in the same way.
Unfortunately, in the video world, democracy rules. Video images are captured, encoded, stored, manipulated, compressed, distributed, and played back in a rich but confused and ad hoc variety of formats and standards. This already means that processing of video signals is necessary, merely to convert among formats.
There's an important further factor that also conspires to mandate that all your video signals have been heavily processed. Bandwidth. Video requires wide bandwidth, which has expensive implications for video circuitry and even more expensive implications for video storage media and transmission media. To reduce this expense, video (especially consumer video) is heavily processed to reduce bandwidth requirements. A number of manipulative tricks are played with the signal, taking advantage of the fact that the eye/brain is actually relatively insensitive to certain visual degradations, and can easily be fooled in a number of certain ways (but not in other ways). For example, the chroma part of the video signal might be recorded with a different bandwidth than the luminance part of the signal, which technically cheats you and fools you, by taking advantage of the fact that the eye/brain typically has different sensitivities to bandwidth degradations in the chroma vs. luminance part of the signal. Another example of course are the various MPEG type compression schemes now prevalent on digital video consumer distribution media, including DVD recordings and dish satellite broadcasts.
In other words, there is no such thing as purist, unprocessed video image, certainly not in consumer (as opposed to professional) video systems. This means that you will need to shop for and select some video signal processing for your video system, to cope with all these processed video signals, as well as the various video signal formats. This seems virtually the opposite of the ethos of high end audio, where the mantra is usually to keep the playback signal path as pure and unprocessed as possible. But some of the same general principles still apply. For example, paying more for higher quality video processing can result in a higher quality image that is more believable. And the signal path should still be kept as simple as possible. For example, every A-D or D-A conversion of the video signal degrades it somewhat, so it is inpotant to minimize the number of such conversions in your chain.
Consider DVD playback. The native output of a DVD player is 480i (interlaced), which is good, but not nearly as good as 480p (progressive), which we regard as the minimum resolution for achieving believability. This means that you'll want to process the video signal from DVD, converting it from 480i to 480p (this conversion process is called de-interlacing). Sounds simple enough. But the big fly in the ointment is that the video content of the DVD itself might have been sourced from various types of video media, including film (which runs at 24 fps), and broadcast or recorded video (which runs at a different 30 fps), and computer generated video (which could be anything). The algorithms for optimally processing a video signal differ, depending on what kind of source the video signal originally came from. So the video processing circuitry you buy, simply to de-interlace the 480i from the DVD into superior 480p, must be adept at performing a variety of different algorithms, and must also be adept at detecting by inference what the nature of the original video source was, so it can then decide which algorithm to use for the present sequence of frames (and it must also be able to change its mind quickly, because movies on DVDs often suddenly incorporate images from a different kind of video source).
Ironically, the original sources (film, computer video, etc.) might well have been progressive, thereafter to be torn apart into interlaced fields to comply with today's DVD standard (which is intentionally compromised to a lowly 480i format in order to be able to directly feed the vast majority of cheap consumer TV sets out there). But, once the original source's progressive frames have been rendered asunder to make interlaced fields, it's surprisingly hard to figure out exactly how to simply put them back together again. It's rather like Humpty Dumpty (the egg). Once broken, you can't put the parts together exactly right, and to attempt your best you first at the very least have to figure out which parts of the smashed egg originally come from the yolk, which from the white, and which from the shell (since of course yolk, white, and shell each require a different kind of glue to best stitch them back together)..
Good de-interlacers must also make judgments about how best to blend successive interlaced frames, to provide you with the sharpest, clearest composite picture with the fewest and least obnoxious artifacts. The processor should look for certain kinds of lines in the image, for rapid motion, and other kinds of problem areas (e.g. spurious artifacts like jaggies and moiré patterns), in order to deal with these problems in the best way possible (for example, via motion adaptive interpolation). Thus, de-interlacing is actually a complex process, which can be done well or poorly by different video processors you can buy.
After your video signal is de-interlaced as well as possible, you will also want to scale it to match your chosen video display. If your video display happens to be a plasma with 480 lines, you're already set. But if your display has more vertical lines than 480, then the now 480p video signal must be processed further, to scale up the number of lines to match the maximum or optimum capability of your chosen display (fixed pixel displays have an obvious native resolution, corresponding to the discrete number of pixel rows, while analog CRT displays similarly have an optimum target resolution, relating to their circuitry and shadow mask). The added lines are generated by a sophisticated interpolation process. There are various interpolation algorithms, each best suited to a different kind of image situation. So the scaling processor must examine the nature of each video frame, as well as its relationship to other nearby frames, in order to determine and then quickly switch to the best interpolation algorithm, in order again to produce the sharpest and clearest image with the fewest and least obnoxious artifacts.
If you wish to learn more about the complex technical details of de-interlacing and scaling, we refer you to the excellent web page put together by Don Munsil and Brian Florian, who have put a lot of work into their explanation. To go there, click here.
The good news for consumers is that the video industry has attacked these complex issues head on, and has now made digital chipsets which can be easily installed in most DVD players and video displays by their manufacturers. For de-interlacing, two chipset brands in particular do a very good job, those from Faroudja/Sage and those from Silicon Image (DVDO). Many fixed pixel displays now incorporate chips to scale a 480 input up to the native resolution of the display (say 720 lines or 768 lines), and they might well also include a de-interlacing chipset that can accept 480i from a non-progressive DVD player. So you can merely look for DVD players and video displays that feature chips from either of these brands. You don't need to worry any more about needing to buy an outboard processor, as you did in home theater's earlier days.
Indeed, these capable chipsets are becoming so prevalent that you are likely to find yourself buying needlessly overlapping duplication of de-interlacing capability, an embarrassment of riches. Today's best DVD players, which you want for their superior electronic video circuitry to output the best quality analog video signal over to your display, also feature progressive scan output using one of these superior de-interlacing chipsets. Then, most of today's best fixed pixel displays (plasma, DLP, LCD) include high quality internal scaling, to scale 480 input up to their native resolution, but they also include a high quality de-interlacing chipset as well, in order to be able to work with DVD players having only 480i output. Thus, you wind up with duplication of the de-interlace function.
This needless duplication isn't quite as senseless as it might first appear. It makes sense for high quality fixed pixel displays to provide scaling on board, since this scaling can be optimized for the target native resolution of the display, more efficiently than it could be in some general purpose outboard scaler box. This means that these displays can be effectively fed directly by DVD players, without intermediate outboard scalers. And this in turn means that these displays had better come equipped to handle inputs from all DVD players out there, including the vast majority which today are non-progressive and only provide a 480i output. So this means that these displays should also provide internal de-interlacing, to convert a 480i input to 480p, in order to be compatible with the vast majority of DVD players out there today. And, if they are a high quality display, they will want to provide this de-interlacing via a high quality chipset. Meanwhile, some moderately priced overseas DVD players feature progressive scan 480p output, but achieve it via a cheap, inferior chipset, in which case a Faroudja/Sage or Silicon Image (DVDO) chipset in the display would do a better job of de-interlacing, so you'd be better off foregoing the progressive output of such a DVD player and using its 480i output instead.
Now let's look at the other shoe on the other foot. High quality DVD players invest extra money in power supplies, sophisticated video circuitry, etc. to deliver the best possible image. Naturally they also want to invest in a superior chipset to do the de-interlacing for their progressive output. Furthermore, their output signal must be compatible with driving not only fixed pixel displays (that might have their own internal high quality de-interlacers), but also the hordes of analog (non-fixed-pixel) video displays out there (CRTs, both direct view and projection). These analog displays might have no internal scaling or primitive scaling, and likewise might have no internal de-interlacing or primitive de-interlacing (especially consumer television sets, where every penny counts in parts cost). Thus, a high quality DVD player can't count on being used with a display having a high quality de-interlacer or perhaps any de-interlacer at all. So it needs to provide one, in order to be compatible with the vast majority of displays out there, which today are analog CRT based.
In other words, high quality DVD players need to provide high quality internal de-interlacing, in order to reach out a hand to and be compatible with the vast majority of inferior quality displays out there, while high quality displays (especially fixed pixel ones) need to provide high quality de-interlacing, in order to reach out a hand to and be compatible with the vast majority of DVD players out there, which provide either no de-interlacing or inferior quality de-interlacing. Of course you, our esteemed reader, will doubtless be selecting high quality units from both categories, the DVD player and the display. So when you buy high quality units today you will probably wind up with superfluous duplication of the de-interlacing function.
In case you do have overlapping duplication in your DVD player and display, you should obviously choose the better quality de-interlacer to use (you can't use both in series). From this story, you can probably tell which de-interlacer in your system is likely to be the better quality one. And feel free to try it both ways, to see which de-interlacer gives you the better picture. Incidentally, if both de-interlacer chipsets were exactly the same, then it would probably be better to use the 480i output from the DVD player and employ the de-interlacer in your display. That's because, with today's analog signal linkage, the signal traversing the cable from DVD player to display will be slightly less degraded if it has lower bandwidth, and a 480i signal has less bandwidth than a 480p signal, so it should be the 480i signal that makes the journey.
If you have a display with a high quality de-interlacer on board, it still pays to get a premium DVD player, even if you might not actually use its progressive output. That's because, with today's analog linkage, it's important that the quality of the analog video signal put out by your DVD player be very high, which requires expensive video circuitry to secure the best possible video image. This situation will change somewhat with tomorrow's digital video interface (DVI) signal connections.
Virtually all video signal processing is accomplished in the digital domain. Thus, if your video signal is in analog form, it must be converted to digital for processing. Each A-D conversion, and each D-A conversion, degrades the signal quality. Obviously, it would make sense to keep the video signal in one domain for as much of its journey as possible, with the fewest possible conversions from D-A or A-D. Most fixed pixel displays are intrinsically digital, with each pixel being individually digitally controlled. Most video media are now intrinsically digital, including DVD and satellite dish TV. Thus it makes obvious sense to have a purely digital chain for home video playback, from source through processor to display, especially for fixed pixel displays.
The industry has just standardized on a new digital video interface (DVI) standard, for interconnecting video devices with purely digital video signals. With DVI, you could theoretically have no D-A or A-D conversions at all when feeding a fixed pixel display, and only one D-A conversion when feeding an analog (CRT) display (that D-A conversion should be accomplished inside the analog display, suitably equipped with a DVI input, so that this conversion could be efficiently optimized for the native or best resolution of that analog display).
We're already starting to see the first round of fixed pixel displays featuring DVI input, including those from Dwin, Revox, and Barco. More will surely follow soon.
When will we see the first DVD players with a DVI digital output? That will have to await some politicking with powerful software interests. They are deathly afraid of piracy if any consumer player is equipped with a direct digital output. They don't care that a digital output will provide better video quality for the consumer, the guy who actually pays all their bills. In their mind the consumer is just a sucker whom they can stick with average quality merchandise. What they do care about is that the higher quality of a digital output would make it easier for pirates to make higher quality copies. Hopefully, sensible heads will prevail and, armed with a universal encryption standard, the DVI purely digital video playback chain can become reality.
Until that time, DVD players will be restricted to analog signal output. This forces an extra D-A and A-D conversion into the chain, with its accompanying signal degradation. It also forces you to pay extra bucks for high quality DVD players that provide the best possible analog output signal, using expensive circuitry to handle the video signal in its analog form.
forward to next part of Master Guide to Home Video
back to table of contents