The Pianist R1 NTSC

[raddygast]

Was wondering if anyone else had backed this up. I just did it with 3 pass VBR, everything set to defaults in RB, and main movie set to encoder default. Using CCE 2.70.

Anyway, the problem is that the movie ends up just skirting the 3Mbps barrier. I am wondering if RB accidentally set it to use the low bitrate matrix, and don't know if there's a way I can check. Actually going into RBOpt and loading the rebuilder.inf file in its working directory, I see this:

avg bitrate 3023
Passes: 3, Quantization Matrices: Standard.

Disable adaptive Q matrix switching is NOT checked. Filters set to none.
Quant characteristics 24, VBR Bias 25, Lowpass 8, Vertical 6, the rest 0.

Not sure what to do here. The source is only being compressed about 31%. I've ripped out DTS audio and French AC3, and used menushrink on the menus. I did as good a job as I could cleaning things up for max bitrate. Is it possible that this film is already seriously skirting the edge of quality?

It's about 2:22 hours in length. I've never seen CCE give me blockiness before, even with more serious compression, but maybe my sources are usually better to begin with.

The real test scene for me is the credits, where you see the guy's hands playing piano while the credits roll. IT looks horrible in my re-encode. The source itself looks ok... not great, but ok. But at various parts during the film I found myself seriously noticing artifacts/blockiness, and some noise around the edges. Then again checking the original, they did not look totally perfect either, but much better.

Any tips here? Maybe the standard matrix is very iffy when you are borderline, and perhaps I should have used the low bitrate matrix (from RB) for the whole feature? There are basically no extras so it's very surprising that I should have to be at such a low BR but I suppose the film is just too damn long or the transfer is horrible.

[raddygast]

Ok, I just noticed something VERY weird.

Is this normal? From the prepare phase in my rebuilder.log for the Pianist, which resulted in the crappy output:

- Reduction Level for DVD-5: 69.8%
- Overall Bitrate : 2,997Kbs
- Space for Video : 3,917,530KB
- HIGH/LOW/TYPICAL Bitrates: 3,008/2,975/2,997 Kbs
[04:53:08] Phase I, PREPARATION completed in 3 minutes.


Now, why are the high/low/typical so dead-on? Usually the High and Low are quite a bit apart. What is up with that tiny variation? Was I accidentally using CBR or something?

[jdobbs]

The high/low/typical bitrates are calculated based upon the allocation of space, which reflects the original DVD. If there isn't a lot of deviation that usually is just an indicator that the cell sizes are of a length that the average needs to be high. CBR is never used in DVD-RB any more. It once was used for stills, but even that is no longer true.

As for the 3,000 -- it does exactly as it says. If it is 3,000 or above it uses the standard matrix.

As for "crappy" output --- there is no reason why a bitrate of 3,000 plus should give bad output unless the original is bad as well. Remember that this is VBR and even at 3,000Kbs the bitrate can adjust on demand up to the maximum bitrate (usually around 8,000 - 9,000Kbs).

[raddygast]

Well, I just tried a prepare on this again. The avg. bitrate for the feature is 2997 Kbps. That's just below 3000, but when I go into RBOpt and check CCE settings, the matrix is set to Standard (Encoder Default in RB's terminology). That surprised me, as I thought it would be set to Custom (DVD-RB's low bitrate).

Anyway, I just watched the original credits, and then the CCE-encoded credits, and the latter is horrible. There is a definite difference. I am so surprised because I've never had any difficulty with DVD-RB and CCE regarding quality issues, that's why I posted this.

I wonder if anyone here has this film and could give it a shot to see what I'm talking about.

Perhaps if I force it to use the low bitrate matrix it'll look better, since it is hovering barely around 3 mbps.

[JohnG]

Originally posted by raddygast

Any tips here? Maybe the standard matrix is very iffy when you are borderline, and perhaps I should have used the low bitrate matrix (from RB) for the whole feature? There are basically no extras so it's very surprising that I should have to be at such a low BR but I suppose the film is just too damn long or the transfer is horrible.

Are you using any avs filters? I do for every backup regardless of input size. Also, try using the new QLB matrix for anything higher than 2000Kbps. Below that go with Jawor1cd or Bach1 (actually even qlb isn't bad in some tests I've done).

[jdobbs]

Originally posted by raddygast
Well, I just tried a prepare on this again. The avg. bitrate for the feature is 2997 Kbps. That's just below 3000, but when I go into RBOpt and check CCE settings, the matrix is set to Standard (Encoder Default in RB's terminology). That surprised me, as I thought it would be set to Custom (DVD-RB's low bitrate). Ok, this is another in a series of responses I've had to make to your posts. Please -- before you post you need to stop, think, and get your facts straight. If the bitrate is less than 3,000Kbs the output is using what is selected as the low bitrate matrix. You aren't looking in the right places... look at the ECL file -- it will tell you what the bitrate is and it will tell you whether an alternate matrix is being used for each segment.

If you're so intent on using the low bitrate matrix. Set it for the Main Feature... :rolleyes:

[raddygast]

Well, I am certain that I am missing some portion of the facts, due to my lack of understanding of the way CCE works with ECL files.

However, I have to take some issue with your insinuation that I just post random verbal diarrhea. I have thought things through, according to my limited understanding, and this is what surprises me:

On all my recent prepares, I've set Low Bitrate Matrix in RB to your "Low Bitrate" option, which uses the Low Bitrate.mtx file in the Matrixes directory of DVD-RB Pro.

Now, on some prepares, when I use RB Opt, I check various segments and notice that in CCE Settings, the matrix is set to "Custom" (which is undoubtedly your low bitrate matrix), *IF* the bitrate is below 3000.

However, in The Pianist, the avg. bitrate for the main movie (all one titleset) is 2997, and yet the CCE Settings in RB Opt points to "Standard." Which tells me RB is not treating it like it's < 3000. Now since it is very close, I was wondering if there was some smart threshold, or maybe it is being reported as 2997 but isn't in reality. That's the only part I wanted clarification on.

If RB Opt does not properly read the inf files (or ecl files, can't be sure which is important here) then I guess I'm at fault for misinterpreting the data.

By the way, I *know* that RB is supposed to use "what is set as the low bitrate matrix". I, of course, have that set to your "low bitrate" option. The fact that the 2997 title was showing up as "Standard" led me to believe the bitrate cutoffs weren't that rigid, or maybe RB Opt is misreporting the avg. bitrate.

[jdobbs]

1. I don't have a clue how RB-Opt decides whether something is "custom"

2. The average bitrate for the titleset has nothing to do with the matrix selected. Matrices are selected on individual segments based upon that segment's bitrate. Why would DVD-RB use a low bitrate matrix on a segment that may be encoded at 5,000Kbs just because the average for the titleset is 2,997Kbs?

3. Open up the ECL file with a text editor and look for "vbr_brate_avg=" for each [item] -- then look at the variable "qmat=" -- and you will see a custom matrix included for any segment that is less than 3,000Kbs.

[jdobbs]

@raddygast

I apologize if I sounded harsh. That wasn't the intent. All I'm asking is that you not state things as fact if you aren't sure.
Disable adaptive Q matrix switching is NOT checked
Incorrect. AQM is always disabled in version 0.81 and above.
The avg. bitrate for the feature is 2997 Kbps. That's just below 3000, but when I go into RBOpt and check CCE settings, the matrix is set to Standard (Encoder Default in RB's terminology) Incorrect -- DVD-RB sets it to what you've chosen. I'm not sure what RB-Opt is reporting.
The Pianist, the avg. bitrate for the main movie (all one titleset) is 2997, and yet the CCE Settings in RB Opt points to "Standard." Which tells me RB is not treating it like it's < 3000. Now since it is very close. Again incorrect, even though I'd explained it to you in the post just before this one.

I try to help as much as I can -- but it takes listening as well.

[raddygast]

Question: what is a segment? Is it a cell?

Originally posted by jdobbs
@raddygast

Again incorrect, even though I'd explained it to you in the post just before this one.

I try to help as much as I can -- but it takes listening as well.

Whoa there. I believe that was a typo -- I meant to say RB Opt wasn't treating it properly, or reporting it properly... not RB.

Anyway, it's clear to me now that almost all of my confusion arises from the use of an older version of RB Opt that predates the new DVD-RB Pro. I should probably shut my mouth and wait until robot1 releases a new version.

Although I am a bit confused as to why you seem so hostile to one of your user base? I realize most of your users are savvy DVD experts who've been using ifoedit and pgcedit and whatnot with CCE for years, but isn't the point of DVD-RB to bring the strength of CCE to the masses?

I think I've actually done a pretty good job learning how it works (and how RB Opt works). I'm just in this small window of confusion due to all the new features, but that'll pass. But when you make statements like "look at the ECL file!" it tells me that you really expect your users to be at a higher level, and some of us aren't.

I apologize if *I* sound a bit belligerent -- I really appreciate this product and the work you put into it. But DVD's are so inherently complicated that it can be a challenge to understand exactly what's going on from a user perspective. Your matrices are a step in the right direction for me, because it's one less thing to worry about if DVD-RB starts setting things automatically. I'm just nosy, that's all, and I wanted to check up on it in RB Opt. :)

So now all I should do is wait for RBOpt version 0.81 or something...