FORGET the Denon 3910 -- still has MACROBLOCKING! - Page 7  

I think it can. I was playing around with the settings last night and just to see what happened, I set the output to Interlaced and it was already set to a resolution of 480. And the only video I use from the 3910 is DVI.

I feel the same way. The picture quality is still outstanding, even with the possibility of the MB issue. I went through several DVDs last night and only encountered the issue once on that Terminator 2 DVD. Obviously I didn't watch the whole movie, but I only saw that issue right there in the beginning. Other than that I did not encounter it again.

No... DVI is 480p/720p/1080i only. DVI spec doesnt allow 480i

HDMI is spec'd for 480i but Denon doesnt seem to allow it per the manual.

My bad! I guess it helps to read the manual huh? :p

My cable HD box output 480i through DVI-D just fine. Yes, we know DVI spec does not include 480i. But that doesn't mean the actual device can not implement 480i (or to deny 480i through DVI).

Just a comment on the Crouching Tiger screenshot... I really had to crank the brightness up on my monitor to see it!

Dreamaster

esp1,
I really don't know enough about how the encoding process works but it seems to me that *HOWEVER* it might be happening, if the encoding process introduces noise that shifts every 18 frames, and the player emphasizes noise, then the 18 frame pattern would still be there in what you now see out of the player -- just as would be the case if the true source material had a pattern in it that shifted every 18 frames and which was being reproduced with complete faith by the encoding process. Admittedly, finding movie material that exhibited any pattern synced to 18 frames would be bizarre, so the 18 frame pattern that rider observed really does point to something in the MPEG processing.

I'm just not sure you can immediately jump to the conclusion that it must be in the DEcoding process.

Real macroblocking noise is certainly an artifact due to poor choices made during encoding. Since it is an error (lossy encoding) to begin with, odds are that just how it is produced varies according to the particular encoding tool used and how that tool is incorrectly set. I would not be surprised to learn that some such tools were written to do something such as revisiting the choices made while generating the over-compressed portions of the image every 18 frames.
--Bob

jonapod,
If you have to strain to see the problem -- getting right up to the screen for example -- then the odds are you are *NOT* seeing the macroblocking bug in your 3910 but are, instead, seeing real macroblocking compression problems coming from the DVD itself. Again the only way to be sure is to compare using another DVD player, separately calibrated, into the same input of the same display to see if the image improves SUBSTANTIALLY.

In the detailed test I did with the 5900 and the 59avi (see my long posts at the end of the "macroblocking favor" thread) I did indeed spot some scenes where very subtle, very ignorable cases of the macroblocking bug showed up in the 5900 -- cases which I would never have seen at all except that I had the 59avi to compare against. But most of the cases I spotted stood out like a sore thumb. Not subtle at all.

That is, if the 3910 has the bug at all when used with your display, you *WILL* find scenes where the problem is obnoxious now that you know what to look for. But if it *DOESN'T* have the bug, you may very well still see minor examples of macroblocking in some DVDs just due to the production of the DVD and the particular setting of white/black levels on your display.

The fact that you are seeing problems in 480i but not in 480p is surprising. You may want to recheck the white/black levels for both settings as I found that the 5900 needed some changes in white/black levels between 480i and 480p signals. Re-adjusting your 480i levels may make the problem go away there as well.

The fact that you are seeing the problem more readily using DVI may be due to white/black levels, or may be due to lack of filtering or other processing in the DVI data path that causes the original macroblocking from the DVD to come through more visibly. That is, the DVI signal may be producing a sharper image that makes the blotches in the original data stand out more. Again the only way to be sure is to compare with another player THROUGH THE DVI INPUT.

But again, if you have to strain to see the problem then you don't really have the problem. Which is good news!
--Bob

Dreammaster,
If you had to crank up black levels (brightness) abnormally to see the problem then you don't have the "macroblocking bug". You are just forcing the orignal noise from the DVD itself to become visible. The "bug" is that a player such as the Denon 5900 emphasizes the pre-existing noise such that it becomes visible under normal viewing conditions.

If you have the "macroblocking bug" such noise will be readily visible while watching scenes with levels set properly -- and will vanish if you play the same scene through another player, separately calibrated, and using the same input on your display.
--Bob

Bob,
I **think** Dreammaster means he had to crank up the brightness on his PC monitor to see the blocks in the JPG.

Thanks for your insightful reply Bob,

So trying to distill into one sentence what I see,

The (extremely subtle, dark) color boundaries are more blocky with the 3910 than with my old interlaced player; they seem to be less dithered - there is no change in color.

Maybe this isn't the MB bug at all but just a very sensitive player, PJ and set of eyes..

I'm sure the conversation will continue to rage, but I am going to try to convince myself that I really don't have a problem.

The bottom line is the PQ is MUCH improved over component.

..and I need to buy the superbit collection. MIB2 looks FANTASTIC (I just wish they would stop using edge enhancement!)


-Jonathan

Jona - You like the PQ over component better than DVI on the 3910? However, is there no upconversion via component? That's very fascinating...

And Ben definitely deserves a gold medal for his work in this thread. Thanks, Ben!

Correct.

I checked my 3910 which is connected to my Sony 70XBR950 via DVI at 1080i (thru a Gefen switch) and also by component at 480p. I used the Monsters Inc. DVD. Now there are two preview rating screens that are green at the beginning of the DVD, one preview for Beauty and the Beast and another for Lilo and Stitch (which is the pic that vetboy attached earlier in this thread).

On the MPAA green rating screen for Beauty and the Beast - no blotches observed via component or DVI.

On the MPAA green rating screen for Lilo and Stitch - I saw the blotches in vetboys pic, but not as pronounced as in his pic. I had to get about 2 feet from the screen to make them out. At 10', I couldn't make out the blotches. Both component and DVI showed theses blotches.

In summary, I've got no clue what's causing this, but it isn't pronounced enough to make me want to return the 3910. The overall PQ is still much better than the Sony DVD player it replaced and the SQ from the 3910 is superb. I'm still gonna stick with the 3910. I'm happy with it.

The MB issue is being produced by the FLI chip. On most displays it is just exaggerating artifacts that are already there. The only exception I have seen to this is the Panny plasma, which has completely random generated noise that looks just horrible.

If you saw something in 480i, it is inherent on the disc and not an issue with the player, this is not an MPEG decoder issue. Aside from that though, there are MPEG decoders that are prone to show more issues then others, but from the tests we've seen the ESS is not one of them. Panasonic's will if the NR settings are not turned off. The Mits MPEG decoder used in a lot of Pioneer's and Denon players also doesn't have this issue.

I didn't see the issue with the Zenith when I had it. Stacey Spears also checked the Meridian and didn't see it, but he may have a display that barely shows it, I know I do. I have seen it on the Samsung 931 and Denon 5900 and it appears people are seeing it on the Crystalio, so that pretty much solidifies that it is a Faroudja issue. Funny, the new Panasonic S97 is going to be FLI based, guess that could go bad couldn't it!

ahh.. no. sorry - my post reads wrong.

I mean that DVI PQ is much better than component. My goof!

So how does it get fixed? does anyone know if this is a firmware problem that could be resloved with an update?

Without the MB issue, the 3910 is just about perfect. (assuming they fix the Y/C delay at the same time :) )


Any news on a Secrets review?.. There is a lot of anticipation to see how it fares against the 2900/5900!

BTW - someone has just posted to the 1910 thread here describing 'blotchiness' at low IRE levels on the 1910...

jonapod,
In my testing of the 5900 and 59avi, I too found the 5900's image significantly better in DVI than in component. In fact, I initially thought that the improvement in DVI was enough that I could ignore the "macroblocking bug" artifacts that I was seeing -- I thought I had reduced them back to what was merely the noise in the original DVD data. That is, some of the "macroblocking bug" problems I saw in component 480p were dramatically reduced in DVI 720p and 1080i.

Of course component 480i didn't show the problems at all, but had it's own issues unrelated to macroblocking -- i.e., the DVI combo into my Fujitsu P50 plasma produced a superior picture in other ways compared to feeding component 480i into the P50 (or my older Toshiba RPTV).

However, after quite a bit more watching, I started to notice NEW examples of the "macroblocking bug" using the 5900 via DVI (verified by seeing that the image was clean using the 59avi via HDMI/DVI) -- examples that were probably there before but which I hadn't seen since my attention was focused on the problems that showed up more easily using component 480p. The net of it was that DVI was far and away the best way for me to use the 5900, but it still left me with remaining "macroblocking bug" issues that I couldn't get past.

Here's an example that I noted in my long posts in the "macroblocking favor" thread of a problem that remained after switching to DVI:

In the opening sequence of Monsters Inc., the trainee monster enters the room of the robot boy. The monster then hides under the bed. The camera pans down to the darkness under the bed and you then see the monster's eyes open in the darkness. At this point check the smooth gray frame of the bed above the monster's eyes.

The 59avi via DVI shows that portion of the bed frame as clean, smooth, gray. The 5900 via DVI shows darker gray/blackish patches moving across that portion of the image.
--Bob

Kris,
Interesting comments re the Panny plasma, if you check out these pictures :-

http://img.photobucket.com/albums/v1...1/DSCN0709.jpg (SDI via Crystalio to TH42PHD6)
http://img.photobucket.com/albums/v1...1/c47f26ae.jpg (SDI via Crystalio to TH42PHD6)

Is that the kind of thing you're talking about? Those blocks move about all over the picture... What is it about Panny plasmas that causes this?

Cheers,
Andy.

Kris, when you say "already there" are you saying the FLI chip is exaggerating the data inherent in the DVD material, or, that the FLI chip is making noise on it's own. In other words if all DVDs were encoded perfectly then this problem would not be issue?

Curious minds,

Jim :)

I wonder whether mains noise could have anything to do with this?

Andy

You can't encod DVDs PERFECTLY with the limitation of bitrates of current DVD technology. The MPEG encoding artifacts are always there. It's just a matter of how to minize it to a degree that its hard to notice.

Well, yes, of course. :p

He pretty much answered the question in the FLI thread though. Post No. 23

http://www.avsforum.com/avs-vb/showt...88#post4359888
Faroudja FLI-23xx Problems? (Denon 3910, Zenith DVB318 Issues) - AVS Forum

Jim

As a former 5900 owner and one who suffered severely from MB, from my early testing, it seems this issue is resolved. After initially connecting it and calibrating the set i have been running through some tests and have not been able to reproduce the error.

When i calibrated the set i re-set all parameters and started over. I verified SVM was disabled and the only thing i can figure now is that sharpness is a little lower than it used to be on the 5900 but thats about it.

I'll continue to look for it but after about 5 discs all is well. I did the Monsters inc under the bed test mentioned earlier and the bed frame had no macroblocking whatsoever in it.

The ones ive watched so far are :U2 live in boston(was notorious in some scenes), monsters inc, Aliens, Monster, scooby doo2. All have looked fantastic. I would almost say the picture is as least as good as the 5900.

I'll keep you all posted as I continue to play some discs.

John

Coud anything else have been changed?

I suspect some other parameter may have been altered. On the 1910 the MB can be hidden a bit if one reduces the brightness (TV or player) and gamma (player).

Sorry let me be more clear. I didnt keep track of my old 5900 settings before i calibrated my set to the 3910. So while all of the settings could be different, the only one i am certain is different from the 5900 by any significant margin is the sharpness. The rest could be different but i didnt feel it was relevant since each player having different calibration settings is nothing unusual.

It is very odd to me, i cant see it at all anymore.

Since according my discs the player is actually 'properly calibrated' now, i owuldnt consider MB 'hidden' since this is the correct setup for this player.

John

Great to hear John, I hope you enjoy the player!

I agree with Expletive. I replaced a 5900 with the 3910 and love the result!
I never had a big MB problem, but I did see it once and awhile with the 5900. I have spent 12hr. (up till now)watching listening to DVD movies,DVD-A and SACD. This player sounds great --so much so that my wife commented about the better sound!!!!! I watched Pirates of the Caribbean last night no graininess and no MB even though I was looking for it.

I think the audio side is terrific -- very define sound stage and clearer channel separation. Menu's changes are faster than the 5900. It sound very detailed over both the D-Link to my 3805 and also over the analog ext-in for SACD. I did have to make a menu change to the SACD setup--)very easy to do) to hear the full range on my system. Glad I made the change even though I loved the sound from the 5900 and the video was very good, I believe the 3910 is better. JMO

Can anyone check Brother Bear. Not that it's a great movie but one that drives me nuts with some problem, MB or not. All of these issues are with the Denon 5900.

In almost all parts, after he changes to the bear, there are shifting blocks or color on the bears and moose. Also there seem to be bands or stripes of color in these same areas. At first I thought my Sharp 12k was not working correctly. However, switching to INHD and putting in another Non-Disney DVD and all was fine.

I put this disk on my RP-82 to my 503 cmx and still can see the same problems but to a much lesser extent. So much so that if I did not already see the problem I might not pay too much attention to it.

I tend to see this problem with all Disney titles and with a 3.5 yr/old and 1.5 yr/old, I watch these dvds alot :)

If the 3910 does not seem to show these problems then that would be good news to me at least. My kids may not care about PQ but I have to watch them too :)

Thanks,
John

The banding in Brother Bear is a mastering issue and readily apparent regardless of the player.

This is something that always worries me. People don't realize that a lot of what they are seeing is probably inherent in the DVD and they are blaming their player for it. The artifacts with the 5900 are display dependant. AGAIN, if you have a display that aggrevates the issue, GET A DIFFERENT PLAYER.