Problem after enabling the "Very Low Bitrate Matrix"

[wfn1]

Hi,

I'm using RB .79 with CCE trial 2.70.02.00, Avisynth 2.5.5.0, DGDecode 1.1.0 in multipass mode. This is the problem I encountered after enabling the "very low bitrate" matrix for extras on a disc with a lot of extras in separate title sets. During the encoding process I get this window which goes away after I click ok (CCE continues to the next segment)

http://img.villagephotos.com/p/2003-3/120500/cce_err.JPG

after the encoding was done I decided to rebuild just to see if it'll blow up and it did with this:

http://img.villagephotos.com/p/2003-3/120500/build_err.JPG

I have not done any preprocessing. The only thing I did was use RB Opt .17Beta to lower the bitrate on the extras to 1850kbps.

[pg55555]

It has been reported that in some cases RB or CCE failed to write the m2v file of some segments
I understand jdobbs added a check in the last RB version to check when this happens.
The workaround is to make cce to process just that m2v.

But your post title says
"Problem after enabling the "Very Low Bitrate Matrix"

This problem was only present with the "Very Low Matriz", have you tried without it?

Other issue is RBOpt: I'm not sure v0.17 is compatible with RB 0.79.

RB 0.79 has added quite a number of new settings

I have experienced some problems with RB 0.79 and RBOpt myself, which has been erratic and difficult to reproduce, reason why I have not posted them

[jdobbs]

I've added code to v0.80 that will automatically rerun the encoder if it finishes with no .M2V. I'm not sure what's happening, but it appears that sometimes CCE just quits before the encode is complete...

As always, though, I have to assume it's my stuff until proven otherwise. So I'm looking closely to see if there is any way I might be the culprit.

[wfn1]

@pg55555
I'll see if not using rbopt changes anything. Thanks for all your suggestions as always you're on top of things :)

@JDobbs

thank you for looking into this issue, I'm running the job as we speak without the matrix selected to check if I encounter the same problem.

[jdobbs]

Originally posted by pg55555
I have experienced some problems with RB 0.79 and RBOpt myself, which has been erratic and difficult to reproduce, reason why I have not posted them What kind of problems? The only thing I can think that might have an impact is the inclusion of the matrices in the .ECL file... I don't believe the REBUILDER.INF file which is used for most of what RB-OPT does hasn't changed at all.

[wfn1]

i just remembered something after you mentioned erratic behavior. i opended the the inf in rbopt and after i adjusted bitrate for the extras for some reason RBopt thought that the files were to be encoded in OPV when i clicked to save changes. i had to set each vts back to multipass vbr. now that i think about it more, that shouldve been my first hint. i'm sorry JDobbs if i jumped the gun on this, it could be rbopts issue.

[jdobbs]

Keep me posted.

[pg55555]

What kind of problems?
Doing Mad About You Disc 3 I selected not to do Half D1 for the extras (standard procedure for episodic), but to add the Resize line with RPOpt, which I used to change bitrates and matrices too.

When I open the output with Shrink (my standard procedure for generating a ISO file for burning, shrinking menus and eliminate audio and subs flags)I found the extras showed only yelow/red/blue bars (as from the twlight zone). Opening with WinDVD and PowerDVd the VOBs and m2v I have the same effect.
If I reencode the segment without the Rezize command in the avs, the file was just fine.
I tried with RB 0.78 and do not have problems
I redid the DVD with 0.79, but this time I selected the Half D1 for the extras VTSs from the Avisynth options. I use RBOpt to tweak the bitrates I do not remember if I used RB Advanced options for select the Low bitrate and Very low bitrates matrices or if I used RBOpt for that. I tested the m2v before rebuilding and they were OK, so I proceded to the rebuild. I tested the VOBs of the Extras, with PowerDVD and Wind DVD, and I have the problem with WinDVD (It gave me an error and quited after oking) but play just fine in Powee DVD. So I Shrinked the menus, and burnt with Decrypter. The DVD played just fine in my standalone.

So my thinking was that as you have included several new options in v0.79Pro, which in some cases do thinks that before were possible only with RBOpt, and Robot1 has not have time to check for compatibilities, maybe there were a conflict.

On the other side, this issue of one DVD player giving a problem and another not (in the final backup, which play Ok in the standalone)stoped me from reporting in this thread, as probably was just an issue with my disc or setup.

But when wfn1 reported a problem, mentioning RBOpt and "very low bitrate" matrix, I just pointed that I have had a problem (different) with RBOpt and v0.79Pro

ps: I have been experimenting the same behaviour related to RBOpt changing encoding to OPV

[wfn1]

Well, here are my findings. What caused my initial errors was in fact RBOpt so I think that would be a "do not use for the time being with .79" type thing. I was able to cause the same behavior with missing m2v's twice by using .17Beta

However, and this is somewhat pertinent to our discussion, when I have the Very Low Bitrate enabled in "Matrix for Extras" my resulting DVD is undersized by 2.5GB. The setup is like this: Steal 33% from extras, main feature set to encoder default, low bitrate and very low bitrate are set to "same as main feature" and extras is very low bitrate. Am I doing something wrong?

[wfn1]

Could someone please clear up this undersizing issue with matrices for me please? And actually the situation with matricies in RB because I could not find pertinent information on how they are used. The main thing I do not understand is why there's 4 separate submenus for matricies: Main feature, Low bitrate, Very low bitrate and Matrix for extras. Let me rephrase that, I do not understand the necessity of 2 additional options in between Main Feature and Extras. My thinking was that when you enable "steal space from extras" and apply a very low bitrate matrix for extras, RB takes titlesets smaller than the main feature and only applies the VLB matrix to those titlesets. Since my output is undersized and the main movie titleset became 1.6GB I think something must have gone amiss. I ran a couple of extra encodes since my box is rather fast and to make sure I wasn't being an idiot but the result is the same. Thanks for any suggestions you might have.

[jdobbs]

As DVD-RB is processing the source data it is also determining the average bitrate for each segment. The bitrate can change between segments (even within a VTS when dynamic allocation is selected). The four matrices are used based upon the resulting bitrate. By using the Matrix options, you can instruct DVD-RB to also dynamically select the Matrix you wish to use. So:

1. If the video bitrate is above 3000Kbs the matrix selected in "Main Feature" is used.
2. If the bitrate drops below 3000Kbs but above 2000Kbs the "Low Bitrate" matrix is used.
3. If the bitrate drops below 2000Kbs, the "Very Low Bitrate" matrix is used.
4. If a matrix is selected for "Extras" it will override any of the above selections.

The matrix you select does not affect your output size. The average bitrate determines that... Select the default for all four matrices (then no matrix will be added to the .ECL files) and run the job again... I'd be surprised if it doesn't still undersize -- something else is amiss.

[wfn1]

Originally posted by jdobbs

4. If a matrix is selected for "Extras" it will override any of the above selections.


Thank you kindly for clearing this up. One last thing, if I understood you correctly, the selection under "matrix for extras" applies only to extras which are determined as such by looking at the size of the source VTS or any other algorythm RB uses to differentiate extras for other functions (e.g. enable "stealing' or half-D1)? In essence the choice you make under matrix for extras does not have anything to do with the way the main movie is encoded?

[robot1]

Originally posted by wfn1
Well, here are my findings. What caused my initial errors was in fact RBOpt so I think that would be a "do not use for the time being with .79" type thing. I was able to cause the same behavior with missing m2v's twice by using .17Beta I'm out of time this week and I can't test RBOpt with the new DVD-RB.
Could you send me by email the Rebuilder.inf and Rebuilder.ecl (original by DVD-RB and modified by RB-Opt)? (email address is in the RB-Opt readme)
If I find the bug, I'll fix it in the weekend.

[jdobbs]

Originally posted by wfn1
Thank you kindly for clearing this up. One last thing, if I understood you correctly, the selection under "matrix for extras" applies only to extras which are determined as such by looking at the size of the source VTS or any other algorythm RB uses to differentiate extras for other functions (e.g. enable "stealing' or half-D1)? In essence the choice you make under matrix for extras does not have anything to do with the way the main movie is encoded? Correct.

[wfn1]

Originally posted by robot1
I'm out of time this week and I can't test RBOpt with the new DVD-RB.
Could you send me by email the Rebuilder.inf and Rebuilder.ecl (original by DVD-RB and modified by RB-Opt)? (email address is in the RB-Opt readme)
If I find the bug, I'll fix it in the weekend.

I just sent you an email with both files. Thank you for looking into this on such a short notice.

@JDobbs

Thank you!

[wfn1]

Originally posted by jdobbs
Select the default for all four matrices (then no matrix will be added to the .ECL files) and run the job again... I'd be surprised if it doesn't still undersize -- something else is amiss.

You are correct, it did undersize with all default settings. This is strange. If you have a free second could you tell me some of the reasons for undersizing in multipass VBR transcodes ?

[Sharc]

As DVD-RB is processing the source data it is also determining the average bitrate for each segment. The bitrate can change between segments (even within a VTS when dynamic allocation is selected). The four matrices are used based upon the resulting bitrate. By using the Matrix options, you can instruct DVD-RB to also dynamically select the Matrix you wish to use. So:

1. If the video bitrate is above 3000Kbs the matrix selected in "Main Feature" is used.
2. If the bitrate drops below 3000Kbs but above 2000Kbs the "Low Bitrate" matrix is used.
3. If the bitrate drops below 2000Kbs, the "Very Low Bitrate" matrix is used.
4. If a matrix is selected for "Extras" it will override any of the above selections.

Assuming that the selection for matrices in the Advanced Options looks like:

1. Main Feature = [A]
2. Low Bitrate = [B]
3. Very Low Bitrate = [C]
4. Matrix for Extras = "same as Main Feature"

What matrices are now applied for the Extras? Is it just the matrix [A] for all segments of the Extras, or does this mean that matrices [A], [B] and [C] are kicked in for the segments of the Extras in the same way as for the Main Feature (= Movie), i.e. according to the average bit rate of the individual segments?

[jdobbs]

It should follow the same set of rules you have provided for the main feature. So could be [A], [b], or [C] depending upon the bitrate.

[Sharc]

ok, thanks.

Another question:
If I tweak the bitrates by means of the beta viewer/editor (e.g. by "allocating saved space to the main feature") , will the matrix allocation follow the modified (=higher) bitrates, or will the original matrix from the "Prepare" phase be kept? Means some segments, although boosted by the tweaking above the 3000 bps threshold, will still be encoded with the "Low Bitrate Matrix"?

If yes, could that be a reason for occasional undersizing due to encoder saturation as reported in other posts of this forum?

(In fact I hardly ever experienced undersizing, perhaps because I always set CCEAQM=1 which modifiies the Low Bitrate Matrix dynamically to a sharper matrix, by halving or quartering its coefficients.)

[jdobbs]

Hmmm... that's a very good observation. Changing the bitrates in the editor viewer doesn't modify the matrix. I'll take a look at that.

[Sharc]

Thank you for looking into that.

I just noticed that it happens also the other way round: When I reduce the bitrate (eg of the Menu and Extras) by means of the beta viewer/editor to values below the 3000 kbps threshold, the "Low Bitrate Matrix" or "Very Low Bitrate Matrix" won't be kicked in.

[jdobbs]

As I said, changing bitrates with the menu editor currently does not change the matrix. The matrix selected during the PREPARE phase remains in effect. That's something I'll need to address.

[Sharc]

Does the bitrate dependent matrix selection also apply to menus (provided that menu encoding has been enabled), or is it only applied to the feature and extras? I wish that it is applied to the menus as well, but so far I had no evidence because the menus were always above the 3000 bps threshold after the Prepare.

Finally, would it be very difficult to introduce a prefix "m:" for filters that should be applied to the menus only? I normally do heavy menu compression (2000 kbps or less) and would like to apply softening filters in order to keep artefacts to a minimum. I don''t care much to loosing details in the menu artwork. So far I have done menu filtering by manually editing the Wxxxxxxxxxx.avs scripts, but an "m:" prefix would be most convenient. What do you think?

[Sharc]

As I said, changing bitrates with the menu editor currently does not change the matrix. The matrix selected during the PREPARE phase remains in effect. That's something I'll need to address.

Any progress since? It just happened once more to me that the main feature was encoded with a conservative (i.e. low bitrate) matrix after blanking some of the extras which actually lifted the bitrate above 3000 kbps. (My mistake, I should have remembered that ...... )