or Connect
AVS › AVS Forum › Video Components › DVD Players (Standard Def) › FORGET the Denon 3910 -- still has MACROBLOCKING!
New Posts  All Forums:Forum Nav:

FORGET the Denon 3910 -- still has MACROBLOCKING! - Page 6  

post #151 of 311
Also tried 1080i and 480p and still no issues.
post #152 of 311
I could post some actual shots from the movie but they are much harder to notice on a still image than when it is moving. The green screen is the most obvious shot I have. Why would it make any difference if the shot was from the film itself, rather than the trailer at the beginning? If it was an inherent flaw in the image, both dvd players should produce it. Believe me, it's there in the film as well but I sure wish it wasn't. The rest of the player is great and the overall picture quality is better than the Sony, but not $1200 better. I hope this is a problem with my setup rather than with the player itself. I'm going to connect the component inputs and see if the problem is still there.
post #153 of 311
Jazzcat,

It is just the angle of the shot. I didn't have my tripod set up so I just shot it by hand so the angle changed. It is an RPTV so that accounts for the way it looks.

Vetboy.
post #154 of 311
Again I think this is a compatibility issue with the display, just like the Panny. It is sourced in the DVD player, but obviously what display you are using will make a big deal.

This is a lot of the reason I give Denon slack, there is no way they could test every display, and it is very likely their displays aren't showing any problem at all.
post #155 of 311
Thanks Vetboy. I guess I need to just get the 3910 and try it for myself. Isn't Monsters Inc. a poor excuse for a DVD? I haven't seen it myself but from other posts it is definitely not reference.

What model is your RPTV? I have the Sammy DLP. It seems that a lot of these issues are display dependent.
post #156 of 311
Quote:
Originally posted by Kris Deering
Again I think this is a compatibility issue with the display, just like the Panny. It is sourced in the DVD player, but obviously what display you are using will make a big deal.

This is a lot of the reason I give Denon slack, there is no way they could test every display, and it is very likely their displays aren't showing any problem at all.
Kris, but what is the 'reaction' that causes this? Is it a poor internal scaler for the display device? For instance my RPTV exhibits it severely, is that becuase of the processing involved to scale an incoming image to its notave resolution? Why only with some palyers?

I dont doubt your suggestion is possible but id like to understand whats going on under the hood better.

John
post #157 of 311
Jazzcat,

My display is the mitsubishi 48413, built last May. If this is just a display dependant problem, Denon is still going to have a lot of problems because I think mitsubishi is one of the biggest sellers in the U.S. Also, I hooked up the component outs and it is still very noticeable in progressive but disappears when switched to interlaced. What does all this mean? Can somebody with more familiarity with the workings of DVD players narrow down the problem? Thanks

Vetboy
post #158 of 311
Thread Starter 
quote:
------------------------------------------------------------------------
Originally posted by ZiggyB
For me, the 'best' scene which displays what I call macroblocking, is on the Superbit of Crouching Tiger, Hidden Dragon - Chapter 7.
------------------------------------------------------------------------


ZiggyB,
This is a great picture of the exact macroblocking effect I am seeing on the 3910. I added the yellow rectangle for people to focus on; it is virtually filled with the random pattern of rectangles of macroblocking (if you have an LCD monitor, viewing the image off-axis makes it easier to see).
Thanks.
LL
post #159 of 311
Thread Starter 
Quote:
Originally posted by pnichols
"quote:
--------------------------------------------------------------------------------
Originally posted by rider
While trying to get shots of it, I found:

1. when I pause, all the macroblock patterns freeze intact and are very visible.
2. if I single-step from the remote (frame-by-frame), the larger patterns (composed of many rectangles) change exactly every 18 frames, while various single rectangles or very small patterns come and go frame-by-frame or every couple of frames.
--------------------------------------------------------------------------------


This is VERY interesting because the GOP size of MPEG encoded NTSC video on DVD is usually 18 frames, i.e. the seqence looks like this:

IBBPBBPBBPBBPBBPBB,IBBPBBPBBPBBPBBPBB,etc
--------- 1 GOP ---------,--------- 1 GOP ---------,etc
(GOP = group of pictures)

I.e. there is a completely new reference picture only every 18 frames and the rest is predicted either backwards (P frames) or bi-directionally (B frames).

If the "artifact" is closely associated with the GOP sequency there is a strong reason to belive that the cause is a fault/bug/error in the MPEG decoder core.

(Assuming it's not part of the original MPEG bitstream of course, if so it should show up on all MPEG decoders and not only the one used by Denon... )


Espen Braathen"


The post above that I quoted from page 4 of this thread seems to have gone unnoticed. Espen may be on to something here. I wonder if the decoder chip revision that Denon is using in the 5900 (....and 3910) isn't really the culprit instead of the Faroudja chip? I believe the decoder chip itself is from ESS, as is in the 9000 and 3800 and maybe other Denons, but most likely it's a newer version of the ESS decoder from what was in previous Denons without the MB bug.

Perhaps this subtle thing Espen mentions points away from the deinterlacer that has been catching most of the speculative blame up to now. I wouldn't put it past the decoder to be causing this type of artifact.

I, too, am surprised this hasn't gotten a great deal of attention!!
The fact that the major macroblock patterns change exactly every 18 frames (i.e. each time a new reference picture is generated by the MPEG encoding scheme) just can't be a coincidence.

Also, the single rectangles and small patterns that come and go every frame or couple of frames would be explained by the generation of the frames in-between the reference picture frames, which are predicted from past and/or future frames on a single-frame basis.
post #160 of 311
Thread Starter 
Quote:
Originally posted by Bob Pariseau (pg. 7, post #125 on this thread)
jonapod,
The blotchiness shown in this jpeg between the wall and the head is typical of "macroblocking" -- particular if it rolls or boils as the scene moves so that it attracts the eye.

Real macroblocking is a type of over-compression problem.....

<snip>

.......if you view the same scene with another DVD player, and now knowing what to look for, you may be able to see a minor remnant of the real, original macroblocking on the DVD itself. But with the 5900, that minor problem becomes a major problem.

On scenes which exhibit the bug, the effect is not at all subtle.
--Bob
Bob,
That's a superb analysis and explanation of the issue of macroblocking -- it really helped me understand it.
Thanks.:)
post #161 of 311
rider,
The connection between the artifact and the 18 frame pattern apparently inherent in MPEG encoding may simply be due to the artifact originating as noise on the DVD itself which is then being magnified by the Denon DVD player -- i.e., the artifact starts as "real" macroblocking noise due to over-compression of that portion of the image in the DVD data -- and that noise is then emphasized by the Denon player to the point where it becomes obnoxious when it might be minor enough to ignore when played on other players.

That is, it would not surprise me to discover that a DVD mastering tool, set to over-compress "background" data in a scene, would exhibit an 18 frame pattern in the macroblocking noise resulting from that over-compression.

And of course once that pattern is built-into the DVD data itself, whatever is going on in the player to emphasize the noise would naturally just repeat that 18 frame pattern.
--Bob
post #162 of 311
Quote:
Originally posted by vetboy

My display is the mitsubishi 48413, built last May. If this is just a display dependant problem, Denon is still going to have a lot of problems because I think mitsubishi is one of the biggest sellers in the U.S.

Vetboy
Whoa, slow down, because you see the problem on you particular TV does not mean in any way, shape, or form that the problem will be seen on all Mits RPTVs. I have a 55813 and a 5900 and I have seen the problem in about 2-3% of the DVDs I watched, total being about 250-300.

By the way, has your TV been calibrated professionally? Unless you know what you are doing in the service menu and have the equipment to measure and then make adjustments I really doubt your set is running at it's optimum. I would not even consider serious viewing on an RPTV untill that was done. Something to think about.

Jim:)
post #163 of 311
Can anyone post some more locations where they think they are seeing the MB? So far I've checked the Monsters Inc. preview screens and the scene from the Larence Of Arabia superbit but can not really say I can see the MB.I don't have the CTHD superbit. I fed a component progressive signal into my Mits 65711 compared to the interlaced and the only thing I noticed was some pretty bad compression artifacts in the sand(Lof A) but this appeared on both progressive and interlaced. I even hooked up a Panny RP91 and noticed the same compression noise but it looked the same on the 3910. I am noticing some blotchiness but I am seeing it in progressive, interlaced, and on another player(RP91) so I don't think it is MB. I did have a 5900 before and I did notice MB on a few disc's but can't recall the specific disc's that had the problems. This player had other significant problems other than just the MB so that may not be a fair judgement. I can say that I have yet to see the MB in the 3910 but I have only used the progressive out. If some of you can give me some other disc's to try I will test them out.
post #164 of 311
Clark, if you can't see it why look for it ;) I thought Denon fixed this problem and for the most part it could be a Projector/DVD Player combination issue. There are a lot of very happy Denon 3910 owners so far. I should have mine tomorrow and will be testing it for a week or so and come to my own conclusions.
post #165 of 311
Keenan,

I realize that because this is happening on my model, it does not mean that all mits tv's will have the problem, just as all denon dvd players don't have the problem. All I meant was that there have got to be a lot of 48413's out there given the number of RPTVs Mits produces.

No, it's not professionally calibrated because I'll be moving when I graduate next year so it simply wouldn't be cost effective. However, this does not negate the fact that there is a problem that obviously exists with the 3910 that does not with the sony 9000, at least on my system.

I have seen the problem in a number of discs, even though I haven't watched more than 10 since I got the player.

Regardless, I think that the improvement, unfortunately, is not worth the price.

Vetboy
post #166 of 311
Quote:
Originally posted by Bob Pariseau
rider,
The connection between the artifact and the 18 frame pattern apparently inherent in MPEG encoding may simply be due to the artifact originating as noise on the DVD itself which is then being magnified by the Denon DVD player -- i.e., the artifact starts as "real" macroblocking noise due to over-compression of that portion of the image in the DVD data -- and that noise is then emphasized by the Denon player to the point where it becomes obnoxious when it might be minor enough to ignore when played on other players.

That is, it would not surprise me to discover that a DVD mastering tool, set to over-compress "background" data in a scene, would exhibit an 18 frame pattern in the macroblocking noise resulting from that over-compression.

And of course once that pattern is built-into the DVD data itself, whatever is going on in the player to emphasize the noise would naturally just repeat that 18 frame pattern.
--Bob
First of all:
ERRORS ARE NOT PROPAGATED FROM THE REFERENCE PICTURE TO THE PREDICTED PICTURES DURING MPEG ENCODING!

The reason for this is pretty obvious if you know how the process works.
The encoder side does not use the original frame as the reference picture for prediction, but actually a encoded/decoded picture - thus the prediction process in the encoder sees the same compressed I reference picture as the decoder side will have available later on!

The prediction searches for the best matching macroblock (or part of the image so to speak), then subtracts this from the reference picture block (allready in compressed and decompressed state) and the resulting predition error is then compressed using quantization of the DCT components (this quantization is the _ONLY_ lossy stage in MPEG compression).

This means that quantization errors does not proagate across frames, thus any correlating pattern which is seen inside a GOP sequence is likely caused by other effects.
post #167 of 311
Vetboy- I see it. The blotchiness in the green for the 3910, and the smoothness for the 9000. Shoot...
post #168 of 311
Quote:
Originally posted by keenan

By the way, has your TV been calibrated professionally? Unless you know what you are doing in the service menu and have the equipment to measure and then make adjustments I really doubt your set is running at it's optimum. I would not even consider serious viewing on an RPTV untill that was done. Something to think about.

Jim:) [/b]
This is a rather meaningless statement as the only thing which will seriuosly affect the image in this respect is the white- and black level settings. There is no need whatsoever for a professional calibration to achieve this.
post #169 of 311
Somone with the Crystalio scaler (which also uses the Genesis FLI320x) on another forum is also having the same problem on his Panasonic Plasma... Not sure what you can conclude from this other than FLI320x + Panasonic Plasma is not a good combination.
post #170 of 311
Garman, I agree with you but the track record I have had with Denon DVD players leads me to be a little cautious. So far I'm very pleased with the 3910. I did notice what I thought was MB on the 5900, although only on a few disc's, and have yet to notice this on the 3910. I can't remember what disc's had the problems but if anyone has some suggestions on disc's to try I would like to know if the problem exists in the 3910 and to what extent. If there are problems with only certain players I want to find out now while I can still possibly exchange my unit for one that functions optimally.
post #171 of 311
If the error is still there (and from the pictures it seems as if it is), i think it really says something that they did not fix it with the 3910. They certainly knew about it, and they WANTED to fix it with the 5900 as evidenced by the failed firmware fix earlier this year.

All this said, has anyone been able to have this issue turn up on other 2310 based players? If so, my thinking is that the 2310 is really the only viable scaling solution right now and it has the bug, so Denon (et al) have no other choice but to use it and live with the consequences. Upscaling players are all the rage right now, I doubt Denon would figure it was better to offer NO solution rather than one that has the MB bug which affects only certain displays.

John
post #172 of 311
Quick question about the MB problem. I understand that it only affects certain displays. Has anyone ever had this problem (with either the 3910 or 5900) with a Sammy DLP?
post #173 of 311
Again, not putting a real fine point on this, but this gives me a sense of deja vu: In the giant Zenith 318 thread, among other places, a bunch of us quickly determined that certain technologies and sets were going to be more sensitive, less forgiving, and have potential compatibility issues.

That's sort of what this sounds like here. We had people with all sorts of issues with the Zenith, and invariably, it was projection technology of some sort or some other high end, newer technology.

Not making any mud slinging against anyone's TV's or types of technology, but maybe there's something to it. *shrugs*
post #174 of 311
Quote:
Originally posted by Q of BanditZ
Again, not putting a real fine point on this, but this gives me a sense of deja vu: In the giant Zenith 318 thread, among other places, a bunch of us quickly determined that certain technologies and sets were going to be more sensitive, less forgiving, and have potential compatibility issues.
Let me guess: these technologies are Plasma and DLP, right? :D

Plasma display and DLP are the two technologies that have very limited gray scale capabilities and rely upon dithering heavily to simulate other shades or colors. Plasma usually can only display 256 to 1024 different shades of colors natively while DLP is even simple: on or off. That's why you see clay faces on a lot of cheap plasmas and DLP TVs.
post #175 of 311
Quote:
Originally posted by Q of BanditZ
Again, not putting a real fine point on this, but this gives me a sense of deja vu: In the giant Zenith 318 thread, among other places, a bunch of us quickly determined that certain technologies and sets were going to be more sensitive, less forgiving, and have potential compatibility issues.

That's sort of what this sounds like here. We had people with all sorts of issues with the Zenith, and invariably, it was projection technology of some sort or some other high end, newer technology.

Not making any mud slinging against anyone's TV's or types of technology, but maybe there's something to it. *shrugs*
SO are you saying that the zenith is experiencing this as well?

I have a Toshiba CRT RPTV and its horrible, i wouldn't consider this issue confined to newer fixed-pxel displays.

John
post #176 of 311
Last night I got to see for myself the MB issue. I spent some time and calibrated my BenQ8700+ to my 3910 that was delivered yesterday. I put in the DVD of Terminator 2 Ultimate Edition and played it at 720p. It reared it's ugly head with the intro screen saying it was a Mario Kassar Production. There were black blocks in the middle of the screen. Tonight I will play around with it a bit more to see what I can do adjusting more settings and playing with IRE settings.

Am I going to return this DVD player because of this, hell no! This is still a great player to me. I can live with it. If they come out with a fix for it, bonus! If not no big deal to me. I guess ignorance is bliss sometimes :) I love the audio performance of this player. It brought new life to my CDs.
post #177 of 311
Quote:
Originally posted by Foxbat121
Let me guess: these technologies are Plasma and DLP, right? :D


LOL... all of them to some degree, really. It just seemed to me that we kept having a lot of people come in with a lot of problems simillar to what we're seeing here, and invariably, you'd ask them what set they owned and the majority answer was a projection technology of some sort.

Not trying to draw any conclusions, just an observation.

Quote:


Plasma display and DLP are the two technologies that have very limited gray scale capabilities and rely upon dithering heavily to simulate other shades or colors. Plasma usually can only display 256 to 1024 different shades of colors natively while DLP is even simple: on or off. That's why you see clay faces on a lot of cheap plasmas and DLP TVs.
That right there may explain a lot of what was going on in the Zenith thread.
post #178 of 311
Quote:
Originally posted by Q of BanditZ


LOL... all of them to some degree, really. It just seemed to me that we kept having a lot of people come in with a lot of problems simillar to what we're seeing here, and invariably, you'd ask them what set they owned and the majority answer was a projection technology of some sort.

Not trying to draw any conclusions, just an observation.



That right there may explain a lot of what was going on in the Zenith thread. [/b]
So that sounds like a yes to the zenith also experiencing this issue?

John
post #179 of 311
Can the 3910 output 480i via the DVI jack?

John
post #180 of 311
Quote:
Originally posted by HT Nitwit
Last night I got to see for myself the MB issue. I spent some time and calibrated my BenQ8700+ to my 3910 that was delivered yesterday. I put in the DVD of Terminator 2 Ultimate Edition and played it at 720p. It reared it's ugly head with the intro screen saying it was a Mario Kassar Production. There were black blocks in the middle of the screen. Tonight I will play around with it a bit more to see what I can do adjusting more settings and playing with IRE settings.

Am I going to return this DVD player because of this, hell no! This is still a great player to me. I can live with it. If they come out with a fix for it, bonus! If not no big deal to me. I guess ignorance is bliss sometimes :) I love the audio performance of this player. It brought new life to my CDs.

I have the BENQ8700+/3910 combo too with a 126" Carada screen..

..and yes, I can see macroblocking too.. :(

I did fairly extensive tests last night and found some interesting results:

480p/720p/1080i thru DIV - definite MB problem
480i thru component - reduced, but still there..
480p thru component - problem gone..

This was very strange to me - 480 prog thru component does NOT show for me. Maybe I'm using a crappy cable. But it's definitely there for DVI and also more importantly, 480i thru component..

Now, if I can see thru 480i, it must be an MPEG decoder error right since my PJ is doing the re-interleaving?

Unless the BENQ8700+ has the FL23xxx chip too...?

My old POS interlaced player still has MB but to a much lesser extent. The biggest difference is less 'blocking' the block of colour are smaller.

It's like the 3910 is adding some sort of noise reduction.

To put it into context, I deliberately looked for disks which have a problem e.g. Monsters Inc.

With 99.9% of my movies, it looks fantastic.

And to see MB issues, I have to stand 3 feet from the screen, turn up the contrast and scrutinise it!?!

The wife made and interesting point - why not sit down and enjoy the show? :)

From the normal seating position - 16ft it looks fine.

I would speculate that because it is digital from source to screen, a level of analog softening is not introduced.

I bought the 3910 specifically for its DVI output, and it looks like a compromise - fab PQ with the possible spectre of MB problem or standard component with a softer PQ and less depth.

I believe that DVD players are now streching the limits of the MPEG-2 system - time to move onto HD-DVD/Bluray methinks!

I'm sticking with the DVI AND the player too.

-Jonathan
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: DVD Players (Standard Def)
This thread is locked  
AVS › AVS Forum › Video Components › DVD Players (Standard Def) › FORGET the Denon 3910 -- still has MACROBLOCKING!