DVD Benchmark - Progressive DVD Player Shootout - December, 2000
Don Munsil and Stacey Spears
Now that we have established a baseline for DVD benchmarks, we can get on with testing lots of players. In the present case, we tested 13 progressive scan DVD players (two of which are actually software programs that allow you to play DVDs on your computer). We found - and you will find - the results very interesting, and surprising. No hi-fi magazine anywhere in the world has performed many of the tests that we did with these players. We have found defects in certain chips that perform the decoding process. The chip manufacturers (not the companies who make the players) will be receiving information from us as to what the defects are and possible solutions. We are sure that the high tech industry is probably not used to receiving chip instruction set critiques from the media, but this is a new millennium, and consumers have a right - maybe even a duty - to let manufacturers know what is right and what is wrong with their products. Many of the writers at Secrets are electrical or computer engineers themselves, and we think it would be shortchanging ourselves, our readers, and the hi-fi manufacturers if we had the capability to test everything down to the software code, and did not do it. So . . . here we go.
- Stacey Spears -
Rather than duplicate explanations of the tests we performed, you can go to the explanatory benchmark articles and get the information there. In particular, Part 5, recently updated, will explain progressive scan.
Table of Tests:
WF1 WHQL Film 1. Pass if it stays in film mode, and doesn't comb. WF2 WHQL Film 2. Pass if it stays in film mode, and doesn't comb. Yellow circle if it goes to video mode. WMM WHQL Mixed Mode. Pass if it returns to film mode in no more than 2 frames, and doesn't comb. Number in parens is time in frames to return to film mode. WC1 WHQL Chapter Stops 1. Pass if it stays in film mode. Number in parentheses is time in frames to return to film mode. WC2 WHQL Chapter Stops 2. Pass if it stays in film mode. Number in parentheses is time in frames to return to film mode. VZP Video Essentials Snell & Wilcox Zone Plate. Pass if the stationary areas of the screen stay full resolution for the whole sequence. BL Big Lebowski Making-of. Pass if it stays in film mode and doesn't comb. Number in parentheses is total number of combs in first 30 seconds. GQM Galaxy Quest Menu. Pass if it stays in film mode and doesn't comb. Yellow circle if it goes to video mode. GQT Galaxy Quest Trailer. Pass if it stays in film mode and doesn't comb. Yellow circle if it goes to video mode. MT More Tales of the City. Pass if it stays in film mode and doesn't comb. Yellow circle if it goes to video mode. T Titanic Chapter 7/8. Pass if it stays in film mode and doesn't comb. Yellow circle if it noticeably drops to video mode. A13 Apollo 13 Making-of. Pass if it stays in video mode and doesn't comb. TS Toy Story Chapter 4. Pass if the chroma has no streaky upsampling artifacts.
The Test Results
We used a green checkmark to show that a player passed the test, and a red X to show that it failed the test. If a player technically failed a test, but it didn't comb or show obvious distracting artifacts (for example, if it didn't comb, but it used video mode when it should have been in film mode), this is indicated by a yellow circle in the box.
In a few cases, there is a number in parentheses after the mark, which tells the number of frames (at 30 per second) it took for the player to recover from glitches and return to film mode. In the case of "The Big Lebowski" test, it shows the number of times the player combed during the 30-second intro.
Player data table:
Player WF1 WF2 WMM WC1 WC2 VZP BL GQM GQT MT T A13 TS 5109 (50) (70) (60) (5) 723GD (1) (12) (1) (5) 6200 (6) (6) 9200 (7) (7) 939 (7) (7) DV-37 (2) (2) (7) S9000 (8) (36) (10) (4) Cinematrix (1) (2) (1) (6) H1000 (1) (5) (1) (6) Roundtable Voyager (2) (4) WinDVD (10) (1) (4) PowerDVD (14) (4) (5) iScan V2 N/A
The 5109 was the very first progressive player released in the US, and it's showing its age a bit. Its de-interlacing solution relies on an apparently simple detection algorithm involving looking at the flags on the disc. It seems to pretty much just use a film mode whenever the progressive_frame flag is set to “true,” and switches to a decent quality motion-adaptive video de-interlacer otherwise. It does handle the alternating progressive frame problem ("Titanic") reasonably well, but when it drops to video mode, it takes a long time before it switches back to film mode. The double-digit recovery times, almost 2 seconds, are not a misprint. Contrary to popular belief, we could not find any evidence of a Genesis chip in the player. As far as we can tell, this player (and the 9100), use a Toshiba de-interlacing solution.
As to video quality, the 5109 is not a top performer. Not only does it have the chroma problem, but saturated colors, especially reds, flicker on the top and bottom of the colored area. This is incredibly visible all through "Toy Story". (The flicker is in the luma channel. When we removed the luma to look at the red problem, the flicker went away.) There is ringing, in the form of a very large overshoot just on one side of each black line in the sharpness pattern (and all through the movies). The player couldn't resolve all the resolution in the standard wedge; it has perhaps 520 TVL of resolution or so.
All that said, because the 5109 is quick to switch to video mode if things go awry, it doesn't comb very much in practice. It's much more likely to drop suddenly to video mode and stay there for a second or two, and since the video de-interlacer in the player is fairly good, you might not notice too much depending on the material. The drop to video was very visible on the "Titanic" chapter change, and we'd expect it to be visible on other similar chapter breaks.
One annoying problem with the 5109 is that it doesn't have any kind of mode switch at all, so when you get something with really bad artifacts, like the "Making of Apollo 13" documentary, you can't do anything about it. You have to manually switch cables on the back of the unit (unless you keep two sets of cables attached) to switch to your TV's de-interlacer.
The 5109 is sold in slightly different form by Mitsubishi as the DD-6000. We were unable to get a sample to run through the full test, but we were able to check one earlier for the chroma artifact and do a few of the deinterlacing tests, and we are confident that it performs very similarly, if not identically, to the Toshiba 5109.
We had the gold-colored version. It's also sold in black as the XV-D721BK.
This player is fascinating. It's the only player in the showdown that contains an integrated single-chip MPEG decoder and de-interlacing solution, which makes it the first player we know of that is able to just read the progressive frames off the disc and send them out with no intervening processing steps. However, there didn't seem to be any significant advantage to doing things this way, from the user's perspective -- a separate deinterlacing chip in the player, operating in the digital domain (which all of them are), can, and does, produce progressive frames of the same quality as the JVC's integrated solution. And overall, the JVC didn't handle unusual material as well as we'd like. That said, most of the time the JVC was quite watchable.
The video quality on this unit was top-notch. It had almost no ringing, and was second only to the Camelot in this regard. There was no Y/C delay, and the measured numbers were good, though not outrageously good. It has a very smooth, clean output, with no artificial edge enhancement. The JVC doesn't have even a hint of the chroma upsampling problem, making it one of only four in the showdown that didn't have the problem.
The down side is that the player is loaded with strange quirks. There are no less than five progressive modes, and none of them work perfectly. We found that all five of them were useful for certain kinds of material. This seems less than optimal. A user should not have to guess which of five different modes will be correct for a particular DVD. However, we found that if you left the player in Smart mode, it would do the right thing more often than not.
The biggest drawback of the de-interlacing is that its video mode is not motion adaptive. It uses vertical filtering. Most of the time, it just slaps two fields together and softens the picture vertically to hide the combs. This loses resolution on all material, and produces something that looks like a double image whenever there is movement in the frame. It's more watchable than constant combing, but it's not by any stretch of the imagination the best way to handle video material.
Material that has the alternating progressive_frame flag, like "Titanic" and "Austin Powers", looked soft in the Smart mode, because it kept the player in video mode, and used the vertical filtering algorithm. It mostly found the right fields to combine, so there weren't many double images, but the image was soft, which defeats the purpose of having a progressive DVD player. In Auto mode, the screen would strobe subtly as it switched from film to video mode every other frame. Only in Film mode did these movies look right.
Material that had incorrect progressive flags almost always made the player comb badly, or made it drop into video mode, and since it has no way of analyzing the cadence, it was unable to recognize any film material that was transferred using any other method but the standard one.
But by far the strangest quirk of this player is that the interlaced output is derived from the progressive output, which is exactly the opposite of the way all the other players work. This means that changing the progressive mode will change the interlaced output. And since there is no one progressive mode that will always produce correct output, there is no way to get correct interlaced output in all situations. Video-based material will still be soft out the interlaced output, for example. Probably if you want to use this player's interlaced output, you will want to put it in Film mode for everything, but even then we can't be sure it will always work.
Even with the quirks and the non motion-adaptive video mode, given that the video quality is so good, and that most movies have good flags, this JVC might well be the best solution in its price range. You may need to experiment with the modes for certain DVDs, but if you're willing to do that, you can usually get something reasonable. And when everything is correct, the picture is excellent. We wish we could recommend this player more strongly, because it does have a lot going for it. Perhaps the next revision will fix some of the problems. If the player just cleaned up some of its mode detection logic and added a motion-adaptive video mode, it would be a serious contender, and would clearly leapfrog some other players costing much, much more.
These are all essentially the same, except that the 9200 and the Onkyo are heavier, and play DVD-Audio discs. Since we didn't test DVD-Audio in this showdown, we'll cover all three together. The 9200 and 939 had slightly different results on the de-interlacing tests, so it may be that Toshiba has tweaked the Genesis chip to optimize for different performance.
These players use the Genesis chipset, and it shows. They are cadence-reading players, and they comb on lots of material. On "More Tales of the City", they combed on at least 75% of the cuts, and generally for multiple frames, making the DVD basically unwatchable unless you switch to video mode. On "The Big Lebowski" making-of montage, it also combed badly, and again we would expect it to do poorly on just about any other making-of documentary. Even in major Hollywood movies like "Independence Day" and "Toy Story", we saw combing, though perhaps only once every few minutes. Frankly, this player and its cousins were below the bar for us as far as de-interlacing performance.
However, on the good side, it's quick and easy to switch de-interlacing modes, so if you see artifacts that are annoying you, it's easy to force it into video mode. You don't even need to stop the movie. We wish, though, that it wasn't necessary quite so often.
The video quality was good, though there was some ringing that we couldn't get rid of with the controls. We would highly recommend setting Edge Enhancement to Off, as any other setting produces incredibly ugly ringing that is noticeable on any material.
The players had the chroma upsampling bug, though not as bad as some. Still, it was clearly visible throughout "Toy Story". The problem will be visible on any film that has saturated red colors, but you might not see it if you don't look for it.
On the Video Essentials Snell & Wilcox Zone Plate pattern, there was flicker in the left-hand patterns whenever the center pattern moved. This is because the video deinterlacing algorithm on the Genesis doesn't appear to be weaving stationary areas of the screen. This means that when objects are moving on screen, the resolution suffers for all areas, not just the moving areas.
These players all had a strange problem with certain movies, apparently all from Warner Brothers all released at about the same time. Portions of the screen would jitter or shimmer. Some described it as looking like a ripple, like seeing the image through water. It turns out that this effect happens only on certain films, and only when the "blacker-than-black" setting is turned on in the Setup menu. This setting is only necessary when you are calibrating the black level of the TV using Video Essentials or the THX Optimode tests, so it's fine to turn it off for normal viewing. Still, it's a strange problem. It would be better if the players could just pass blacker-than-black signals normally, without requiring you to change this setting at all.
This player has some very nice features, and a much improved remote compared to earlier models. It also is an improvement on the previous elite models, the DV-09 and DV-05. Unlike them, the DV-37 has no significant chroma delay, and the ringing is improved. That said, we were still disappointed with this player. It has the worst chroma upsampling artifact of any of the players reviewed. We couldn't imagine how someone could watch "Toy Story" on this player and not notice the jagged streaks on almost every red object in the movie. And, we noticed the problem on almost every movie we used in the testing, even when we weren't looking for the problem. The red streaks were that bad.
The de-interlacing performance was not very good either. It's a flag-reading player, and like every other flag-reading player in the showdown, the Pioneer does fine when it's fed perfect material with perfect flags. (As we mentioned, this is one reason you see so many glowing reviews of players by other magazines. They are using terrific DVDs to review them. Most players will get a good report as such. It is really necessary to test them with difficult DVDs to discover the true capabilities of players.) But when the flags are wrong, the DV-37 is not up to the task. The player clearly uses the flags on the disc to make decisions, so it combed on many of the tests with bad or unusual flags. It also dropped to video mode when it didn't need to. Interestingly, the player combed in places that no other player combed, even other players that use the flags, for example the film that plays behind the “Special Features” menu in "Galaxy Quest".
The ringing on this player isn't as bad as the DV-09, but it's still too high, well above average for this group. Turning all detail and sharpness settings down on the player made the problem slightly better, but even after every setting was at zero, the player still had worse ringing problems than any other player in the showdown.
While looking at discs with this player, we saw a strange problem with the WHQL credits roll -- it looked like it was blocky and flickering. All the players had minor flicker problems here, but the DV-37 was a whole order of magnitude worse.
The S9000ES is a very good unit, but not absolutely top notch. Our chart shows that the S9000ES didn't do perfectly on our de-interlacing tests, but the chart doesn't tell the whole story. The S9000ES uses the two-chip Genesis solution, which means it uses cadence reading, and the de-interlacing has all the problems of that chipset. However, while the player failed most of the same tests as the other players using the Genesis chipset, it became clear as we were testing that the S9000ES combs less often than the other Genesis-based players, and it combs for a shorter time. You can see this on the “comb count” on "The Big Lebowski".
We think that Sony has tweaked the settings on the Genesis film-detection chip so it is faster to leave film mode and slower to return to film mode. You can see this in the long recovery times on the Mixed Mode and Chapter Stops tests. This protects against combing, though it does mean the S9000ES is more likely to use video mode, and for a longer time. In practice, most of the time when the S9000ES drops to video mode you won't notice it. Combing is much more noticeable than switching to video mode, so the Sony engineers made the right choice here.
That said, the Sony clearly didn't come close to the performance of our Silicon Image-based players. Watching some material in Auto mode on the Sony, such as stuff shot on film and edited on video, like "More Tales of the City", is an exercise in frustration, and you'll probably find yourself switching to Video mode. Sadly, the menu to change de-interlacing modes is buried deep, and it takes a lot of button pressing to get there. And you have to stop the player to change the mode, though the Sony will remember your place and return there when you hit “Play.”
The S9000ES has the chroma upsampling bug, clearly visible in "Toy Story", though not as bad as some other players. As with the 7700, when you hit pause on the remote, the color problem clears up.
The S9000ES had some ringing in the sharpness pattern on Avia, and we could see it in the image on several of our DVDs, but again, it's not as bad as most of the other players.
We saw the same flickering and distortion of the left-hand diagonal lines on the Video Essentials Zone Plate pattern, which is consistent with what we saw on the 6200/9200/939.
Overall, the S9000ES is a good performer, but we're disappointed that it's not a truly reference-quality player, especially given the hype about the player, and the high price. On the other hand, it's better than most of the other players we looked at, and it includes SACD. So if you were going to get a SACD player anyway, the S9000ES could be considered a bargain, as you're clearly getting a better player than any of the other Genesis-based units.
The Cinematrix add-on for DVD players is a nifty idea in concept. It is an add-on board that can in theory be installed on any DVD player, and completely replaces the output stage of the player with a de-interlacer plus scaler that can scale the output to almost any resolution you could desire. And since the de-interlacing and scaling are done in the digital domain, there is no loss to extra D/A and A/D steps, as there is with an external de-interlacer. We found the execution to still have some hiccups.
First off, the player they most often attach the mod to is the Sony S7700, which has the chroma upsampling problem. With the Cinematrix installed, the chroma problem is more noticeable, which is not the Cinematrix's fault, but still, the artifact is there. Strangely, like the other Sony player we looked at (the S9000ES), the chroma problem disappears when you pause or go into slow motion mode on the player. But in play mode you can clearly see the streaks in the red areas on "Toy Story".
We would recommend that people get the mod done to a Sony S7000, which does not have the chroma bug, or a decent quality Panasonic player, as the Panasonics we looked at did not have the problem. Since the Cinematrix replaces everything in the signal chain after the MPEG decoder, all you really are using from the host player is the transport, navigation firmware, and the MPEG decoder. The entire video output stage is handled by the Cinematrix.
When we looked at the resolution patterns in 480p mode, the loss of resolution was severe. It looked as though the Cinematrix was perhaps reducing the 720x480 frame to something like 640x480. For this reason, we can't recommend this mod for 480p output. However, most people get the mod to attach to a high-resolution projector, so we switched it into 720p mode, and the resolution patterns got better. It wasn't perfect, as there was a significant amount of moiré in the pattern, and there was still a significant amount of ringing, but the image was significantly smoother than in 480p mode.
The settings on the unit are controlled by a combination of dip switches and special codes entered through the remote, which had poor and convoluted documentation. We were eventually able to get the player into 720p mode and turn off all the sharpness “enhancements,” but it took some time. Once you've gotten the unit into the mode you want, you will probably just leave it alone, but be prepared for some frustration while you get everything set up.
The outputs on the Cinematrix were well above 100 IRE, which would not be a huge problem with most CRT projectors, which have a fairly high tolerance for voltage overshoot, but for the VW10HT projector we were using for our tests, and for a few video processors and de-interlacers, voltages over 100 IRE cause clipping of the whites. We were unable to fix this problem with the contrast controls on the 10HT; we just couldn't get back most of the white detail. We tried it on a CRT, and it was able to handle the hot whites just fine. Still, it shows lack of attention to detail that the output was not using the correct voltages.
The de-interlacing on the Cinematrix was not the worst we saw, but wasn't the best. It appears to be a cadence-reading solution. What isn't shown in the table is that it combed on many different films in places we weren't looking for combing. It recovers quickly, as you see in the Mixed Mode and Chapter Stops test, but it combed in places no other player combed. If you are sensitive to de-interlacing artifacts, this may not be a good choice.
We still like the idea of the Cinematrix, and we hope that the next revision gets better. A huge improvement would be to use the Silicon Image (DVDO) chip instead of whatever one they are using. If they switched to a Silicon Image chip, fixed the 480p output, and put it on a player that doesn't have the chroma problem, it would probably give the Camelot a run for its money.
This player has been around almost as long as the Toshiba 5109, but it's still hanging in there, and it's still a contender. This player is the only one to use the Genesis gmVLX1A without the gmAFMC, as far as we know. The player implements its own film detection, which seems to boil down to looking for the standard film flag sequence. If it sees that sequence, it goes into film mode. Any deviation from the standard sequence causes it to go into video mode, except the alternating progressive_frame flag material, which is still recognized as film.
You can see from the chart that the H1000, like the other flag-reading players, went into video mode on most of our film material that was encoded in a non-standard way. This looks basically OK, but it's unfortunate, as the main thing you want from a progressive DVD player is a working film mode, and it's frustrating when a player doesn't handle good film material properly. Unfortunately, forcing the player into film mode on those DVDs does not give good results. The H1000 doesn't have any way of analyzing the fields to see which ones go together, so forcing it into film mode on film material that was encoded like video will cause constant combing. And the player does comb quite often on material that it recognizes as film, but that has cadence or flag errors. Like the S9000ES, the menu to change the de-interlacing mode is buried deep, which makes it annoying to switch to video mode for material that isn't looking right.
Like the other Genesis players, this player failed the Snell and Wilcox Zone Plate test, indicating that the video-mode deinterlacing is not motion-adaptive.
The video quality on this player was excellent. It holds up well against the best players in this group, with very little ringing and no significant chroma delay. There was no chroma upsampling problem at all.
This is another boutique player that is made from an off-the-shelf standard player, in this case the Panasonic 110. However, it has a new case, power supplies, isolation transformers, video board, audio board, and front and back panel. Most importantly, it has the newest Silicon Image (DVDO) deinterlacing chip, the Sil503. The case feels quite solid, but it's not nearly as heavy as other boutique players, which we frankly think is a good thing. It includes some of Camelot's special processing technology on the audio side, but we didn't test that in this review.
We can't say enough good things about this player. It really does everything right. It has no trace of the chroma problem, though that's because the Panasonic 110's MPEG decoder doesn't have the problem. The video quality is superb. There is as close to 0 ringing as we've seen. You get full resolution, and only 1 pixel of cropping, on the very bottom. The picture is smooth and film-like, with no hint of the edgy quality you get from players with edge enhancement. The player had the lowest noise on the progressive video output of any player we measured, probably because of the 80 MHz video DACs.
As you can see from the chart, it passed all of our tests with flying colors. No other player came close. It was almost impossible to get this player to comb. And while it occasionally dropped to video mode for a moment, it has a very good motion-adaptive de-interlacer, and it was very difficult to tell the difference on most normal material. There are no modes to set at all, because there is no need to set modes. The Silicon Image film mode detection is that good.
This is an expensive player, at around $3,495. But it does deliver, in spades. In video quality and de-interlacing, this player is absolutely the top of the heap. We're spoiled now: everything else feels like a compromise after experiencing this player.
The Theta is an unusual player. Like most high-end boutique players, it's made by taking an off-the-shelf player, in this case a Pioneer Laserdisc/DVD player (possibly a 909 or 525), and adding better audio and video stages. However, generally the boutique companies take just the guts out of the player and put them into a new case. Theta has taken the entire player, minus just the top and side panels, and built a second case around it. If you open the Theta, you can see the whole original Pioneer inside, including the back panel with RCA jacks.
We had some trouble getting the player to produce a signal that the 10HT could use. The output is configured using a combination of the original Pioneer's output menus and some switches on the back, and it was not at all obvious what to do. We eventually ended up using RGBHV, but we tried, and failed, to get Component and RGBS to work. Again, since in general you would spend some time setting it up, then leave it alone, it's only an issue when you first get the player, or if you switch monitors at some point.
We believe the Theta we tested had a DVDO version 1 (DV101) chip in it, and if so, the performance was not as good as the later DVDO (now Silicon Image) versions. It did better than the Genesis solutions, but not as good as the stand alone iScan de-interlacer or the Camelot. It also had a surprising breakdown on the Film 1 test, which is usually a “freebie” for any progressive player. The de-interlacing chip lost the cadence twice during that test and dropped to video mode for a moment. It probably wouldn't be terribly noticeable on real movie material, but we still had to fail it. The Film 2 sequence worked fine.
Video performance was fine. The player does have the chroma problem, and pretty bad, which it inherited from the Pioneer. So far, all Pioneer players we've looked at have the problem, but we haven't looked at every Pioneer player. There is also a YC delay on the progressive outputs. Actually there is a delay between the two color difference channels, 'Pb' and 'Pr'.
All in all, for an approximately $10,000 player, this didn't seem to deliver the goods.
As HTPCs are becoming more popular for folks with digital front projectors, we thought it would be worthwhile to try our de-interlacing torture test on the two most popular software DVD players.
The PC used was PIII 600 with 128MB of RAM and a GeForce 256 video card set at 32-bit color. DMA was enabled on the DVD-ROM drive. We ran the PC at 60Hz.
The biggest drawback to this player was that, like the JVC, it pretty much only uses the flags and frame structure on the disc to make de-interlacing decisions. It doesn't have a motion-adaptive video-mode de-interlacer, so if it thinks the material is video, it uses vertical filtering, and there didn't seem to be a way to make it stop. Consequently, we were completely unable to get what we would consider an acceptable result with "More Tales of the City", the "Galaxy Quest" trailer, and other video mode or oddly formatted film titles.
We also were disconcerted by an overall stuttery quality in almost all pans, credit scrolls, etc. We did have DMA mode enabled, and tried both 60 Hz and 72 Hz refresh, but couldn't get it to go away. Perhaps if we dug deeper and played with more caching and other parameters, we could have made the problem better. As it was, though, we found it distracting. A good example is in Chapter 12 of "Gladiator". You can see the software decoder stutter along as you pan over the city. (Update: We have now seen one HTPC that didn't have this stutter problem, so it is avoidable by choosing good components and setting the PC up optimally. However, we are still disappointed that a wide variety of PCs we looked at had the stuttering problem, even though all of them comfortably exceeded the minimum requirements of the software.)
And it almost goes without saying that a standalone DVD player with a remote, even a bad one, has significantly better usability than a HTPC DVD player. Trying to find the right key to pause, back up, and so forth in the dark was a huge frustration. And using the mouse was even more frustrating. If one cobbles together an infrared remote solution with a Pronto or other programmable remote and an IR receiver on the PC, this gets much better.
We were pretty much stymied when it came to calibration. There don't seem to be enough controls to do the steps we're used to from calibrating video displays. That said, the color was not too far off, and the brightness and contrast could be calibrated just fine.
For big Hollywood blockbusters where the flags are OK, the player did a great job. The basic video quality was excellent. In general, you almost never get visible ringing from a video card, and there was no trace of the chroma upsampling problem. There is no overscan or pixel cropping, so you see all of the picture, something you don't get on any standalone player. Considering that if you already have a PC, buying a DVD-Rom drive and the software can be done for around $150 or so, it's definitely a bargain way to get very high quality video.
The PC used was PIII 600 with 128MB of RAM and a GeForce 256 video card set at 32-bit color. DMA was enabled on the DVD-ROM drive. We ran the PC at 60Hz.
This player had many more features than the WinDVD solution, which made navigation more pleasant. We had all the same problems that we had with PC players in general as far as usability, calibration, and the constant use of vertical filtering as a deinterlacing algorithm. We also had trouble with the stuttery pans, though again, it might be that changing parameters on the PC could help with this.
PowerDVD seemed to do a little better at identifying video it shouldn't weave together. It handled the "Apollo 13" documentary much better than WinDVD, for example.
We saw the chroma upsampling problem when in software mode. We later tried it on another machine that had a video card with hardware assist, and the problem went away.
Like WinDVD, on good material PowerDVD looked really great. Resolution was perfect, and we got the whole image. We hear each software program has a few discs that it doesn't play right, so it's probably worthwhile to get them both, considering the incredibly low price.
Even though this is an external deinterlacer rather than a player, we thought it would be worthwhile to run our tests on it and consider how it would fare if driven by a good quality interlaced DVD player. As far as we can tell, even though this is now manufactured by Silicon Image, the brand name of the stand-alone deinterlacer is still DVDO.
As you'd expect, the de-interlacing was essentially perfect. We didn't see any really significant differences between the Sil502 chip in this unit and the Sil503 in the Camelot, but perhaps we just didn't choose the right test to show the differences. Certainly it sailed through our tests with aplomb.
The biggest drawback of using this de-interlacer instead of a progressive DVD player is that it reduces resolution slightly, and increases video noise slightly. We'd estimate that you lose around 15-20 TVL of resolution, which softens the picture a bit. In a few cases, we preferred the look of the picture through the iScan, as it hid some of the MPEG artifacts.
The iScan didn't seem to change the chroma delay. If the inputs have no chroma delay, the outputs won't either. The iScan does seem to add a little ringing, though not as bad as some of the players in the showdown. In general, you'd expect the ringing to be additive, so it behooves you to mate the iScan to a player with a good signal to start with. And while the iScan doesn't add or subtract the chroma problem, it passes it right on through. Interlaced players that we know don't have the chroma problem include the Panasonic 110, 310, 120, 320, RV30, CV50, and RV80, and the Sony S7000 and S3000. We'll try to test more players and report on whether their MPEG decoders have the problem.
The iScan presents some fairly clear tradeoffs. You get de-interlacing that is far better than almost any stand alone DVD player, but at a loss of picture quality. And, of course, you get great de-interlacing for all your sources, including VHS and cable or Digital Satellite. If de-interlacing artifacts are driving you crazy, and you can't afford to buy a Camelot, the iScan may fit the bill. But if you want the absolute best quality picture and are willing to live with some artifacts, you're probably better off with one of the progressive players.
- Don Munsil -
Go to Part 3 - Functionality Part 4 - Usability Part 5 - Progressive Scan
© Copyright 2000 Secrets of Home Theater & High Fidelity
Return to Table of Contents for this Issue.