[Emil Naepflein]

Hi,


before my vacation I managed to get a used Lumagen Vison directly from the states for a good price.


I have it setup now with my XGA DLP projector and ISCO anamorphic lens and it looks really good with PAL. Now I have taken out my old NTSC AVIA disk for calibration and discovered that the moving zone plate is still full of moving artifacts, just like with my old Crystal Image scaler.


How does the moving zone plate look in your HT?


I expected that it should look clean just like the still image with the Lumagen. I always thought this artifacts are the result of the Crystalimage scaler.


May there be some other problem in the chain of my devices?


DVD Pioneer DV717 (multi region) - S video

Lumagen Vision

3M 8750 XGA DLP projector


Any hints?


Emil

 

[oferlaor]

Emil,


That moving zome plate test on AVIA is a VERY VERY difficult one to get right. I'm not sure I ever got to see it working 100% on any device yet.

 

[Dale Adams]

Emil,


The moving zone plate patterns on AVIA will never look as good as the static one. The reason is MPEG compression artifacts.


The static zone plate has a few such artifacts, which are visible if you look closely. However, the moving patterns (there are 2 of them - more on this in a minute) have many, many more artifacts. The basic reason for this is that they're, well, moving. A static image only has to be compressed once, so the storage required for this is relatively modest. A moving image, though, must have all the individual frames encoded, and so requires much more storage and a higher average bitrate. To make matters worse, test patterns with lots of high frequency information - which is exactly what the zone plate patterns are - happen to be some of the hardest images to compress. This results in more compression artifacts and a worse looking image.


There are actually 2 moving zone plate patterns on AVIA. One of them is more highly compressed than the other, and consequently has quite a few more visible compression artifacts. Both have a lot more artifacts than the static pattern.


I believe the moving images were generated as progressive frames at 30 FPS, and so are encoded on the disc using 2:2 pulldown. Very few deinterlacers appear to be able to track this particular pattern (perhaps due to the high frequencies in the pattern or the high level of compression artifacts), and so you also get deinterlacing artifacts layered on top of the MPEG compression artifacts.


- Dale Adams

 

[oferlaor]

Dale,


Looks like the authors of AVIA have screwed up then.


Very special care has to be taken, by the authors of any calibration DVD, in order to ensure that the source material is perfect. Imagine if the Y/C delay graph on AVIA had a bug and had its own slight Y/C delay during the encoding process...


As I understand it, that's why the DVE DVDs and DTheater tapes were delayed for so long - to ensure that there are no screwups like that...


If the AVIA zone plate has too much data to be adequately encoded without significant compression artifacts, then they *should have* reduced the bandwidth on that zone plate! i.e., they should have modified the plate in such a way as to ensure that, at the source, it is as perfect as possible with as little artifacts as possible.


I'm not aware of many devices capable of catching 2:2 pulldown on 60Hz video. I know the iScan series of doublers calls this "computer mode" (as computers many times use 2:2 pulldown for 60Hz output).


This might be a feature inside Sil504 (if so, it would mean that iScan, CS-X and Lumagen series of scalers also support this).


Not sure if Faroudja based, or other types of scalers test for 2:2 pulldown when presented with 60Hz material.

 

[Emil Naepflein]

Hi,


my problem are not the compression artifacts. In my opinion this are de-interlacing artefacts when diagonal lines are not stable when the zone plate moves.


If the moving zone plate is really encoded with 2:2 pulldown then I think it is clear that problems will arise and this test is then pretty useless when most of the NTSC material is with 3:2 pulldown.


How does the zone plate look with a HTPC which can decode in the digital domain?

It should then probably do the correct pulldown.


Emil

 

[yubyub]

Quote:

Not sure if Faroudja based, or other types of scalers test for 2:2 pulldown when presented with 60Hz material.

Most Faroudja based deinterlacers fail this test, but can be adjusted to detect it properly. On many Panasonic players, the "Auto 2" mode will detect 2:2 pulldown properly.


Emil,


I can suggest two things: try to use component if your Pioneer has it (probably won't make much of a difference, since the Zone Plate is mostly luma info). If you have a friend who's willing to lend you their DVD player, try a DVD player besides a Pioneer. Pioneers generally have terrible mpeg decoders. Try for a Panasonic or other good performer.


That said, I have a Panasonic RV-31 (a decent interlaced player with good video specs) fed into a Faroudja NRS via component, and the zone plate "bouncing ball" on both VE and DVE still shows artifacts. I'm not really gonna worry about it until they come out with "Zone Plate - The Feature Film."


-Jon

 

[Dale Adams]

Quote:

Originally posted by yubyub That said, I have a Panasonic RV-31 (a decent interlaced player with good video specs) fed into a Faroudja NRS via component, and the zone plate "bouncing ball" on both VE and DVE still shows artifacts.

The S&W moving zone plate pattern on Video Essentials has 3 sections: the 1st is 3:2 pulldown, the 2nd is 2:2 pulldown, and the last is 60 field/sec interlaced video (although I might have the order of the last 2 reversed). Most decent deinterlacers will detect the 3:2 PD pattern, although many (most?) do not detect the 2:2 PD pattern. Also note that 3:2 pulldown sequence on the moving zone plate ball has a sequence discontinuity every time it stops moving. I.e., your deinterlacer should drop out of lock on the 3:2 pattern whenever the ball stops moving.


- Dale Adams

 

[yubyub]

I'm not sure about when the NRS drops lock, as it really has no visible way of telling (besides via artifacts). Back when I was using the IScan Pro, I believe that unit exhibited the behavior you describe. I honestly can't remember if it locked onto the 2:2 pattern. I would be surprised if it did...


-Jon

 

[Dale Adams]

Quote:

Originally posted by oferlaor Looks like the authors of AVIA have screwed up then. Very special care has to be taken, by the authors of any calibration DVD, in order to ensure that the source material is perfect. Imagine if the Y/C delay graph on AVIA had a bug and had its own slight Y/C delay during the encoding process... As I understand it, that's why the DVE DVDs and DTheater tapes were delayed for so long - to ensure that there are no screwups like that... If the AVIA zone plate has too much data to be adequately encoded without significant compression artifacts, then they *should have* reduced the bandwidth on that zone plate! i.e., they should have modified the plate in such a way as to ensure that, at the source, it is as perfect as possible with as little artifacts as possible.

Ofer,


They may not have screwed up. I'm pretty sure that at least part of the intention of the 2 moving patterns was to show the effects of compression artifacts at different bit rates. Granted, this makes these patterns less useful for certain other purposes, such as testing deinterlacer performance. I don't have my AVIA disc handy to check, but I think that the test pattern notes on the disc for these patterns say something like this. Of course only the AVIA authors could tell us for sure . . .


- Dale Adams

 

[oferlaor]

Dale,


I thought the purpose was to replace an expensive S&W test signal generator. I guess it's not up to that level yet...

 

[Dale Adams]

Quote:

Originally posted by oferlaor Dale, I thought the purpose was to replace an expensive S&W test signal generator. I guess it's not up to that level yet...

That's certainly the purpose of the moving zone plate on Video Essentials, as the that test pattern was taken directly from an S&W video signal generator.


With respect to the AVIA test patterns, it turns out that I was only half right (or half wrong, if you will) regarding the 2 moving zone plate test patterns. The high bitrate pattern was indeed intended to test deinterlacer and Y/C separation hardware. The low bitrate pattern, however, was largely intended to show the effects of over-compression. At least that's what I gather from reading the description of each of the 2 test patterns on the AVIA disc.


The high bitrate pattern was encoded at close to 10 Mbit/sec, the limit of the DVD format. It appears as if they have a bit too much of a moving area containing very high frequencies even for this bitrate, and consequently get more compression artifacts than you might like.


- Dale Adams

 

[mark haflich]

What about the new digital VE? I purchased copies of the 1080i and 720p tapes for $70 each and 2 copies of the DVD at $20 each. I gave all but one copy of the DVD to Mike Parker. My system being up is still 3 weeks away or so. The eight addirional chairs arrived today. MP stilll has to return my SDI DVD player. The electrician still has a couple of week ends of work. The end is in sight.

 

[Emil Naepflein]

Hi,


thank you for all your replies. I am no much less worried about the performance of the Lumagen. As most of my viewing is DVB PAL the NTSC performance is not so important after all. My general impression with the Lumagen is very positive. The image looks very 3D and the sharpness control works very effective with soft low compression content.


Thanks for your comments,


Emil