I'm currently experimenting with a Denon DVD-5900. Of course I wanted to look for "macroblocking" to see if it is something that bothers me.
Based on what I've read here so far, I expected to see the problem in Progressive mode and not in Interlaced mode, and indeed that's what I see. What surprised me is that there's another way to make the problem go away -- DVI 1080i.
Since the conversion to 1080i is supposed to include the de-interlacing that's presumed to be the source of the bug, I'm not at all sure what's going on here!
---------------------------------------------------------
My DVD-5900 was a recent purchase, with a serial # beginning with "4" and with the "-7" level ESS firmware. Denon tech support confirms to me that this means the hardware is completely up to date (including whatever hardware change they made for IEEE 1394) and that there are only two firmware upgrades not applied -- an additional fix to IEEE 1394 and a fix for certain Beethoven DVD-A disks.
The DVD-5900 is connected to a Toshiba 57HDX82 RPTV via both DVI and Component. I have separately calibrated, using Avia, the DVI 1080i signal, the Component 480P signal, and the Component 480i signal. The Progressive processing (film mode setting) was set to "Auto 1", black level to "0" IRE, and DVI Black Level to "Normal".
I found two instance of what I believe is being referred to as "macroblocking" that stand out like a sore thumb and are just about as much fun to have to put up with. Using the Component 480p signal:
1) In Monsters Inc., in the opening sequence with the robot boy in his bed and the room lights off, there is a cut to a view of the edge of his dresser on the right half of the screen and the frame of the closet door on the left half. Look at the frame of the closet door in the left half of the image. Gray, blocky, moving blotchiness is obvious. Simply horrible.
2) In Dark City, in the opening sequence, the doctor opens his pocket watch to check the approach of midnight. First you see the open pocket watch in his hand then there is a close-up of the just the face of the watch itself. During the closeup, check the left hand half of the watch face. You will see violent, brighter, moving blotchiness -- almost sparklies. Also horrible.
Seeing these makes it obvious to me why some folks have reacted so strongly to this bug. This is not minor stuff.
OK so now look at the same sequences using the Component 480i signal (i.e., turn off Progressive in the DVD-5900 setup menu). As expected, this cleans up both of these sequences.
Knowing what to look for you can just barely see the remnants of the problem in the Monsters Inc. closet door. As has been reported, the root cause of the problem is data on the DVD itself. It's just that the bug magnifies the artifact enough to make it impossible to ignore. But in 480i, the Monsters Inc. image is clean enough that I'd never have noticed it if I wasn't staring at that particular portion of the image having just seen the problem in 480p so I knew just what to look for. As for the Dark City sequence, when played at 480i it is so cleaned up that I can't even see the remnants of the problem.
-------------------------------------------------------------
Now for the fun part.
View the same sequences in DVI 1080i. To my eye the 1080i signal looks JUST AS CLEAN as the 480i version. Of course the overall display quality is much improved but for all intents and purposes the problem is gone. That is, if you like the way the 480i signal looked, the DVI 1080i signal looks just as good as regards the blotchiness and better as regards overall picture quality.
Again, the difference here is not subtle. The 480p version looks horrible, the 480i version looks clean, and the 1080i version looks clean and crisp and transparent.
As a double check, I've shown these sequences to a friend who agrees with me that as far as this particular blotchiness problem in these two sequences, there is no doubt that Component 480p is broken and Component 480i and DVI 1080i are correct and virtually identical.
Since it has been reported that the DVD-5900 generates its DVI 1080i signal by de-interlacing 480i to 480p, upscaling to 1080p, and then re-interlacing to 1080i, and since the working assumption is that the bug is in the de-interlacing of 480i to 480p, it is a mystery to me why the 1080i signal should not also show this problem, but there it is.
------------------------------------------------------------
If anyone has particular instances of this problem which show up in DVI 1080i but go away in Component 480i (so as to eliminate artifacts which arise from the DVD itself as opposed to this particular bug), I'd be interested to hear about it. If I can, I'll try such sequences in this setup and report back.
--Bob