AVS Forum banner
Status
Not open for further replies.
1 - 13 of 13 Posts

· Registered
Joined
·
233 Posts
Discussion Starter · #1 ·
Greetings fellow pioneers!


As the owner of an HTPC equipped with the sweet SDI-SILK card, who just upgraded from an SDI-equipped DVD player that had the chroma bug (Electrograph DVD2000) to one that did NOT have the chroma bug (modified Denon 3300), I have to say:


This is not a subtle improvement on PQ at all! At least, not on a large screen (4.5' by 8') with a high resolution projector (G15 DILA).


I waited until I could confirm with my own eyes what I had always suspected: sharpness, detail, outline clarity and color shading are greatly impoved!


I applaud all the efforts of Stacey Spears and others to raise this issue and keep it in the light until the equipment makers get it fixed.


I realize that this bug is perhaps not an issue on the small direct view sets that the majority of DVD players drive, which is why the manufacturers are letting it slide. (Denon has apparently regressed! their latest players have the chroma bug!)


However, for anyone magnifying and scaling the image for larger screens, I find it incredible that the chroma bug is considered subtle!


People with good vision, though perhaps not experienced in judging all the aspects of image quality, certainly are sensitive to the perceived detail in an image.


To my eyes (spoiled as I may be! :)) , the difference in the detail and the sharpness of outlines is obvious and once experienced, of course, there is no going back.


By now, we who frequent this forum (even as lurkers 8=>) are good at detecting de-interlacing errors and the net effect of the chroma bug is constant de-interlacing errors in the color space. And since the image is composed of pixels of color, modulated by the luma information, incorrectly combining rows of color information, due to the eyes' variable sensitivity to color hues, indirectly affects the luma or intensity of the pixel, further muddying the image reconstruction.


Much is made of the superior color space of HDTV signals, which in my opinion, goes a long way toward the perceived superiority of the image even on upconverted content. Getting the color right is almost as big a factor in image quality as the higher pixel resolution.


Shadow detail, as the Impressionist painters discovered, is ruled by subtle color choices, and here is where the non-chroma bug player shines: I can now see farther into the shadow areas of an image without straining my eyes looking in vain for detail in a muddy area. In fact, the film grain is now more apparent, evidence that the image reconstruction is getting closer to the source.


FOR SDI users:


If you have gone this far, DO NOT settle for an SDI feed from a player whose MPEG decoder has the chroma bug!


I am amazed at the difference and my daughter, who is not yet a video phile (in spite of living with a video and audio phool!), walked into the HT last night and asked me what I did to the system, as she throught that I has viewing HDTV!


I look at the money being dropped on standalone players that have the chroma bug and at the investment some have made in high end scalers and I feel both smug and flabbergasted.


My base HTPC with SDI_SILK card and running DSCALER (yeah team!) and SDI-modified player came in under $4000 and for DVD playback, I have seen nothing better at any price for film-based content.


No, I have not seen the Terranex nor the Snell and Wilcox interpolator but, fine as those pieces of video gear are, can they truly restore the messed up color even when fed an SDI stream from a chroma bug-infected MPEG chip?


In any case, those solutions are out of the reach of the middle class and my goal is to find ways of delivering state-of-the-art home theater experience that the middle class can afford (by simply foregoing that shiny new SUV!).


I am now confidant that I am squeezing as much information out of the DVD format as possible.


My heartfelt thanks to:


Dan Schmelzer - SD-SILK card

Craig Heims of Cellar Cinemas - HTPC builder extraordinaire

Stacey Spears - Secrets of Home Theater and High Fidelity

( http://www.hometheaterhifi.com )

William Phelps - Projector calibrator extraordinaire

The DSCALER software team (you know who you are!)


and everyone here who contribute to this great forum and paves the way for a future of higher quality and quantity of information and long term enjoyment for all!


Thanks for letting me rant!
 

· Registered
Joined
·
9,260 Posts
Would you mind sharing your Dscaler settings. I am alawys curious what other fellow Dscaler-s find looks best.


I was actually playing with a tough one last night - "Mr. Hollands Opus". I can't get rid of the line pairing. I suspect this was mastered from a composite video master.


RP56 w/SDI

Dscaler P4/1.5 128mb

SDI Silk

GeForce2

AA 9A60 RGB to YpBpR

Pioneer 710HD RPTV
 

· Registered
Joined
·
233 Posts
Discussion Starter · #3 ·
Hi Glimmie


I used the auto calibration feature of DSCALER to set its own settings.


[For the lurkers, the autocalibrate menu only appears when you left double-click on dscaler to make it window itself versus full screen.

From the Hardware menu, first choose "test pattern select"

I used Video Essentials (VE) range of gray first

I insert VE into the player and find the title and chapter and pause the player

Then I choose "Start card calibration" ->>Automatic

When complete, accept the settings.

Then repeat, choosing the VE color bars.]


With a Radeon 8500 OEM video card, the settings that the auto calibrate set for itself are:

Brightness (black level):15

Contrast:129

Hue -31

Color:126

ColorU: 126

ColorV:126

Overscan:12


Then, using VE again, I set the Radeon overlay:

Advanced menu->>Video Overlay->>Settings adjustments

I left the brightness at 75

Contrast was at 100 but I sweeten it to 103

Hue:0

Gamma: was at 0 or 1 but I sweeten it to 22

Sharpness defaulted to 5 but I set it to 2

I presently have the checkbox labelled: "Use DScaler settings for overlay" checked, which may ignore the tweaks to the overlay I just alluded to!!


Someday, I would like to understand this brilliant program better.

[Hint: DSCALER gurus please enlighten!]


I use the Linear Correction filter and the Temporal Comb filter all the time, unless the source is so crappy that the filters cause artifacts in the de-interlacing.


Also, for approximately one-fourth to one-third of the DVD's I have, I use the gamma filter, which boosts the low IRE range at the expense of the high IRE range and contrast, to make dark transfers watchable.


I almost feel guilty using the gamma filter, knowing I am losing contrast and possibly black level. However, since I never touch the G15 projector's settings since calibration, I have to adjust the signal using DScaler's filter, which is the preferred point in the signal path to make minor adjustments.


I just watched Mr. Holland's opus and I see what you mean; many scenes have line twitter (obvious de-interlacing artifacts) and jaggies along edges that DSCaler can't eliminate, while other, lower contrast/even toned scenes look fine.


Sometimes the line twitter looks like a poor video tape with tape stretch causing off horizontal diagonal smearing due to mistracking. I agree that the master for the DVD transfer is suspect and maybe it was a transfer from a composite NTSC tape with problems!


Regards,
 

· Registered
Joined
·
9,884 Posts
Was Mr. Hollands Opus originally from video? I'm not sure how you'd get those artifacts otherwise. But if it's video, are you using Greedy (High Motion) to deinterlace?


And on the chroma bug, I remember Stacy Spears had something posted once that explained why it happens but I didn't have the problem and never bothered to understatnd it. I'll go look for it again but I wonder if it is something that DScaler could easily adjust for. Is the chroma just left shifted or something, or is it something much harder to fix?


- Tom
 

· Registered
Joined
·
9,884 Posts
My hat is off to Don Munsil and Stacy Spears for their very excellent analysis of the chroma bug.


I just went back and found their report. I'd previously seen it but not read more than the first page or so because I was looking for something else. But anyone interested in this should see DVD Benchmark - Special Report - The Chroma Upsampling Error in DVD Players - April, 2001


Amoung other things it explains why some SDI Silk users have claimed they can get a more vibrant color than with some of the software players. It turns out that many software and hardware players do not bother to properly interpolate the chroma values vertically when upsampling from 4:2:0 to 4:2:2 chroma formats. That not only amplifies the chroma bug but makes blocky color data even when the bug is not present.


It also turns out that in the most egregious cases DScaler would be able to have an optional filter that could both correct the chroma bug and probably upsample the color better, without even using too much CPU time.


Stacy & Don had already pointed this out if I'd taken the time to read it before:
Quote:
It would also be possible to fix this problem with a de-interlacing chip. All that would be required is to resample the whole chroma channel vertically, smoothing out the errors and jaggedness. Swapping chroma lines 2 & 3, lines 6 & 7, and so on, then applying a smoothing filter, should produce excellent results.
(but maybe interpolate properly for progressive, don't smooth)


There would be some issues to be worked out, like adjusting for overscan timing etc. but if enough folks here are pestered by this bug it would probably be possible to do something about it in many cases. And there are indeed a number of different cases. Probably only the user could choose which to use. But DScaler could make the options available.


Curiously, I think DScaler could probably get the best color out of the worst players because those have not even bothered to interpolate chroma so the original info is still all there. ;)


I'm currently working on something else but if any other DScaler programmers wanted to tackle this then I'd be glad to help. Just email or PM me.


Or has everyone already gone out and bought a better DVD player? ;)


- Tom
 

· Registered
Joined
·
2,379 Posts
robobob: Thanks for sharing your experiences!


Tom: There are a substantial number of DVD players that have the chroma bug (half of them?).


I think it would super cool to be able to correct for it, although I don't know how many different kinds of errors there are. You could probably correct for one or two of the most common decoder chips and that would cover most players.


Added to that, the chroma bug may crop up in other contexts (satellite, etc.)...
 

· Registered
Joined
·
9,884 Posts
Dan -


I got the impression from the article that the most common and obnoxious form of the chroma bug occurred when:


1) The material was coded with the progressive flag on (or should have been).


2) The decoder up samples the color as if it were stored interlaced.


3) The offending decoder up samples simply by chroma line doubling, adjusting for vertical offsets but doing no interpolation.


If I understood the article right and assuming it is correct then maybe it would be worthwhile just doing it for the above case first. Even when number 3) above is violated the results would still maybe be an improvement. And a user could easily just try it and turn it off when it didn't work well.


I know there is code in DVD2AVI, for instance, that does check the progressive flags before converting from 4:2:0 to 4:2:2. I guess it would be possible to rip a couple of the problem DVD's and set break points there to see what the data looks like if anyone were to work on this.


- Tom
 

· Registered
Joined
·
9,884 Posts
Quote:
added to that, the chroma bug may crop up in other contexts (satellite, etc.)...
Anyone know of other instances of this problem besides DVD players?


- Tom
 

· Registered
Joined
·
1,912 Posts
I see it in HDTV on both the HiPix and ADTV. It's pretty rare and I only notice it in highly saturated reds with a large contrast around the red. I have a clip from the tonight show (Cirque du Soleil) which has a scene (the support struts in the upper stage are washed with a saturated red light) that shows the problem pretty clearly.


Dave
 

· Registered
Joined
·
311 Posts
Quote:
Originally posted by robobob
No, I have not seen the Terranex nor the Snell and Wilcox interpolator but, fine as those pieces of video gear are, can they truly restore the messed up color even when fed an SDI stream from a chroma bug-infected MPEG chip?
Robobob,


The Terranex setup I saw recently did not help the problem. A Meridian 800 DVD -> Miranda Component to SDI converter -> Terranex -> Vidikron Vision One with HD-SDI input was ghastly with the chroma bug.


You are quite right in that it severely affects the picture quality. Sadly, I use my Meridian for music only and a lowly Panasonic RP56 for DVD video duties.


Regards, Peter
 

· Registered
Joined
·
4,085 Posts
Hi,


Just FYI, the recent Ravisent video decoder (ATI DVD Player, Cineplayer) when running in DXVA HW accel MPEG2 decoding mode on a Radeon card gives a PERFECT chroma up-sampling picture. NO chroma bug. The Red looks even MUCH detail, solid and smooth than my standalone Panasonic RA71 interlace player line doubled by a iScan Plus.


regards,


Li On
 

· Registered
Joined
·
9,884 Posts
The article I linked above suggests that PowerDVD 3.0 has the chroma bug. Can anyone verify this would make a decent test case? Since I don't have a DVD player with that problem I'd likely have to use a 2nd computer for a test player.


I guess maybe I will write the filter if no one else can be talked into it, at least for the simple worst case I posted above.


- Tom
 
1 - 13 of 13 Posts
Status
Not open for further replies.
Top