RB-Opt v0.23 BETA a tool to change titles bitrate, CCE parameters and AVS scripts.

[robot1]

Hi robot1,

I'd like to do a feature request. In DVD-RB, there's the feature to use a different matrix for cells of high, low and very low bitrate. It would be nice if you could do this distribution in RB-Opt and set the boundaries to whatever you like. I could ask for this in the Wish List thread but I don't want to burden jdobbs with my endless feature requests:) Besides, the option is an advanced one so it would fit in much better in RB-Opt.

Should I add a function to automatically set the matrices for every cell, according to the cell bitrate, and to the bitrate limit set by the user?
This should not be a problem.


Then there's the infamous "redistribute the bits between the cells" request, which probably will never make it to DVD-RB ;) If you're interested, the thread discussing the issue is in this forum.
Could you link me the discussion, or explain the problem? PM if you prefer.

[Boulder]

Should I add a function to automatically set the matrices for every cell, according to the cell bitrate, and to the bitrate limit set by the user?
This should not be a problem.Yes, that's exactly what I meant. Basically the same thing that DVD-RB already does automatically but with user defined bitrate limits. In DVD-RB the division is "over 3000kbps", "2000-3000kbps" and "under 2000kbps". I think DVD-RB also doesn't reassign matrices after you've adjusted the average bitrate in the segment editor. The feature would com e in real handy in this case.Could you link me the discussion, or explain the problem? PM if you prefer.Here it is: http://forum.doom9.org/showthread.php?t=122351. If you want me to explain anything, just let me know. In short, the bitrate distribution between cells is often sub-optimal in the original DVD. My suggestion calculates the proportionate bitrate demand in each cell and adjusts the average bitrates accordingly.

[~bT~]

I always use RB-Opt whether I do any changes or not as I noticed my encodes always come out between 4.35-37GB.

Waiting for the new version...

Thanks a lot!

[Sharc]

If I may add something to the wish list:

I would like to see the menus (VIDEO_TS.VOB, VTS_XX_0.VOB) included in a new version of RB-Opt.
RB-Opt would then offer the possibility to add filters (AVS script) to the menus, select a suitable matrix for higher compressability and reduce the bitrate for the menu stuff accordingly, still maintaining an acceptable quality)

And why not add a Donate option ....... ;-)

[Boulder]

And why not add a Donate option ....... ;-)I'd be happy to donate if my requests ever appeared in RB-Opt, they would both save a lot of manual work :)

[linx05]

I'd be happy to donate if my requests ever appeared in RB-Opt, they would both save a lot of manual work :)I will donate aswell. It's a ****** good program!

[archaeo]

same here.

[stereo]

Yeah, I would also be happy to donate. I love the DVD-RB/RB-Opt combo, and donation would seem perfectly appropriate.

[Sharc]

Then there's the infamous "redistribute the bits between the cells" request, which probably will never make it to DVD-RB ;) If you're interested, the thread discussing the issue is in this forum.

Hmmm..... maybe I miss the point. But why don't you just go for an OPV (or CQ) encode? I think it does much what you are aiming at as it allocates the bits where needed for maintaining a "constant quality" over all the segments, irrespective of their original average bitrate. In RB-Opt you have in addition the possibility to link several VOB-IDs for a joint estimation of the Q factor.
I feel that what you describe is the reason that for many cases an OPV encode looks even better than a multi-pass VBR which operates and optimizes on average bitrates of individual segments rather than on the whole title.
The only difference I see is that one would hit the target size more precisely using the procedure you suggest (i.e. mapping the OPV predictions to average bitrates of individual cells for multi-pass VBR), or that you could of course apply your method to an encoder which does not support one pass encoding.
Still you might have to set DISABLE_DYNAMIC_PEAK=1 to loosen the handbreak (at your risk).....

Maybe I misunderstood something?

[Boulder]

No, you got the idea perfectly. You could say "let's bring the benefits of OPV to the masses" as when the redistribution has been accurate enough (that is, the max bitrate limit has not restricted the compressibility test encode too much), the final result shows that the average quants between the segments are very close to each other. In fact, as I use this method always with my DVB re-encodes, HC's log has showed that the difference is usually less than 0.1.

OPV is a bit cumbersome and this method will have its constant quality benefit but is still accurate in filesize. The downside is that you have to spend the time creating the compressibility test encodes. With current CPUs that is not a big problem IMO.

[robot1]

Here is the new version.
RB-Opt v0.25 (http://www.savefile.com/files/636578)
I've fixed all the bugs I was aware of, and now it fully supports CCE SP2 and parameters for Menu Encoding.
Consider the beta status of this version, as I need more testing, especially for the new features.

Next version will have bitrate redistribution and dynamic matrices selection (as requested by Boulder). Anyway i can add these features only when using CCE. I think DVD-RB generates the parameters for HC and QuEnc at runtime (there isn't the equivalent of Rebuilder.ECL). Can anyone confirm this?
(if there is a request, I could code the bitrate redistribution also for HC and/or QuEnc)

[Rippraff]

Thanks a bunch robot1, I'll give it a go. :)
Edit: Seems you don't like even numbers... :D

Cu Rippraff

[SpazzHH]

Ah robot1, Bless you.

[Boulder]

Next version will have bitrate redistribution and dynamic matrices selection (as requested by Boulder). Anyway i can add these features only when using CCE. I think DVD-RB generates the parameters for HC and QuEnc at runtime (there isn't the equivalent of Rebuilder.ECL). Can anyone confirm this?
(if there is a request, I could code the bitrate redistribution also for HC and/or QuEnc)Hey, that's great news :)

I think you are correct, DVD-RB creates the ini files on the fly based on the data in rebuilder.ecl. This means you can actually use any encoder to do the redistribution pass. Just do a constant quant/constant quality encode using the DVD-RB created avs files with any encoder you like and use the results to calculate the new average bitrates, then edit the rebuilder.ecl file accordingly. DVD-RB will happily do the rest.

[robot1]

Hey, that's great news :)

I think you are correct, DVD-RB creates the ini files on the fly based on the data in rebuilder.ecl.
I think rebuilder.ecl is used for the bitrate, but not for other parameters (matrices, dc precision...).

This means you can actually use any encoder to do the redistribution pass. Just do a constant quant/constant quality encode using the DVD-RB created avs files with any encoder you like and use the results to calculate the new average bitrates, then edit the rebuilder.ecl file accordingly. DVD-RB will happily do the rest.
Yes, that's possible and will be the first release. But what if I have only free encoders (HC and QuEnc) and want to run bitrate redistribution? Probably RB-Opt have to use also these encoders.

[Boulder]

I think rebuilder.ecl is used for the bitrate, but not for other parameters (matrices, dc precision...).I'm not sure..I think I've tampered with the ecl file and the changes have been there in the HC GUI.

Yes, that's possible and will be the first release. But what if I have only free encoders (HC and QuEnc) and want to run bitrate redistribution? Probably RB-Opt have to use also these encoders.You can simply use the avs files DVD-RB created, if needed, replace ConverttoYUY2 with ConverttoYV12 in them. Then for every avs file, create an HC ini file which does the CQ encode and run the ini files. For the calculations, you need the length in frames in every avs file and the average bitrate of every created m2v file.

The ini file can be as simple as this:
*INFILE d:\temp\dvd-rip\rebuilder\d2vavs\v01000000001001.avs
*OUTFILE D:\Temp\DVD-Rip\Rebuilder\D2VAVS\V01000000001001.m2v
*MAXBITRATE 8500
*PROFILE best
*ASPECT 4:3
*CQ_MAXBITRATE 4.000
*DC_PREC 10
*MATRIX mpeg
*WAIT 0
The only requirement for the redistribution pass is that every avs file is encoded in constant quality/quant mode using the exact same parameters. The encoder you wish to use (in DVD-RB or for the redistribution pass) doesn't matter :)

[Rippraff]

I think rebuilder.ecl is used for the bitrate, but not for other parameters (matrices, dc precision...).
No, DC precision is stored in rebuilder.ecl and matrices for CCE as well. But I don't know how matrices for HC are stored.

Cu Rippraff

[Boulder]

No, DC precision is stored in rebuilder.ecl and matrices for CCE as well. But I don't know how matrices for HC are stored.

Cu RippraffIt's easy to parse the matrices from rebuilder.ecl. In fact they could probably be used without any editing at all. Though I'm not sure if the format changes along with the different CCE versions.

[robot1]

RBOpt parses (and can edit) the rebuilder.ecl.
But will DVD-RB consider this editing for the actual encoding in HC?
Example:
segment 1 has DC precision = 10.
RB-Opt can change DC precision to 8 or 9.
Using CCE or Procoder, you will have the real encoding with the new parameter.
But what if you encode with HC or QuEnc? Will DVD-RB use it's internal parameter?
Same thing for other parameters (probably only bitrate editing will be taken in account).

[Boulder]

I'll give it a whirl. I'll run the prepare phase in HC mode to make sure it's HC-compatible rebuilder.ecl we're seeing and see what happens when I edit the items.

[Rippraff]

It's easy to parse the matrices from rebuilder.ecl.
There are no matrices in .ecl to be parsed if you choose HC for example.

Cu Rippraff

[Rippraff]

But will DVD-RB consider this editing for the actual encoding in HC?
Example:
segment 1 has DC precision = 10.
RB-Opt can change DC precision to 8 or 9.
Using CCE or Procoder, you will have the real encoding with the new parameter.
But what if you encode with HC or QuEnc? Will DVD-RB use it's internal parameter?
I guess RB will use the setting which are originally stored for CCE basic (intra_dc_prec=1) and will parse this to HC.ini.

But maybe we should stop the speculation and wait for jdobbs' reply. ;)

Cu Rippraff

[Voodoochild]

Hi I'm new to RB-Opt and played now with it to see what I can understand of it. I saw I can choose matrix from the DVD-RB folder even when using OPV.
I know CCE using the default Matrix of CCE when in OPV mode, can I change that using RB-Opt ?? If yes and I can do OPV with avamat6 for example that will be awesome.

10x Elad

[Boulder]

There are no matrices in .ecl to be parsed if you choose HC for example.

Cu RippraffYes, there are matrices.

qmat=
08 20 24 28 33 34 37 43
20 20 28 30 34 37 43 47
24 28 33 34 37 43 43 48
28 28 33 34 37 43 47 50
28 33 34 37 40 44 50 60
33 34 37 40 44 50 60 73
33 34 37 43 48 58 70 87
34 37 44 48 58 70 87 104
16 22 23 24 25 27 28 29
22 23 24 25 27 28 29 30
23 24 25 27 28 29 30 32
24 25 27 28 29 30 33 34
25 27 28 29 32 33 34 35
27 28 29 30 33 34 35 38
28 29 30 33 34 35 38 39
29 30 32 34 35 38 39 42
I did the tests and everything else except for the intra DC precision changed. There might be some key values to use (2 means DC precision of 10bits, that much I know), but jdobbs sure knows if he wishes to reply here.

[robot1]

Hi I'm new to RB-Opt and played now with it to see what I can understand of it. I saw I can choose matrix from the DVD-RB folder even when using OPV.
I know CCE using the default Matrix of CCE when in OPV mode, can I change that using RB-Opt ?? If yes and I can do OPV with avamat6 for example that will be awesome.

10x Elad
Yes, you can change matrices in OPV mode.

[robot1]

I did the tests and everything else except for the intra DC precision changed. There might be some key values to use (2 means DC precision of 10bits, that much I know), but jdobbs sure knows if he wishes to reply here.
Thank you for the test. It should possible to change HC parameters editing the rebuilder.ecl. Am I right?

[Boulder]

Yes, according to the quick test. The only thing that wouldn't change is the DC precision but it's possible that there are just some specific things you need to do to edit it.

[Sir Didymus]

Yes, Robot1, you are right.

The prepare phase is both logically and from the implementation point of view separated from the following encoding and rebuild steps. The "database" linking the prepare step to the following phases is exactely the rebuilder.ecl file. This holds for all of the supported encoders, AFAIK...

Best regards,
and thanks a lot for your tool!

Edit: don't know why the change of the DC prec. in the test of Boulder is not accepted, however... :confused:

[Rippraff]

Yes, there are matrices.
Guess what the problem was? I've forgotten that I had mistakenly OPV ticked... :rolleyes:

There might be some key values to use (2 means DC precision of 10bits, that much I know), but jdobbs sure knows if he wishes to reply here.
The precision is set for CCE Basic by:
intra_dc_prec=n
Where n is 1 for 8 bit, 2 for 9 bit, and 3 for 10 bit.
Edit: Made some additional tests:
For HC it seems
n is 0 for 8 bit, 1 for 9 bit, and 2 for 10 bit (which is also default/auto)

Cu Rippraff

[Boulder]

Setting intra_dc_prec to 0 doesn't change anything here...it's still 10bits.

[Rippraff]

Well, what I wrote above are only the rebuilder.ecl entries, I can't tell anything how jdobbs parse this to HC.ini.

Cu Rippraff

[Sharc]

Awesome! Thanks robot1 for the new version!

[jdobbs]

Ok... I zipped through the numerous posts -- hope I got the gist of it.

The ECL file is used by DVD-RB to set the values for HC on the Pro Version. It is read and interpreted, and an INI file is generated at encode time. For DC_PREC the value comes from REBUILDER.INI (or the associated project file) using the "DCPrec=n" (which is set by the selection in the OPTIONS tab) -- if n = 1 it sets it to 8 bit, 2 to 9 bit, and 3 to 10 bit. Zero uses the encoder default.

Lots of values are derived from the ECL, including:

- Aspect Ratio
- OPV setting
- Bitrates (min, average, and max)
- progressive/interlaced settings
- TFF or BFF
- The selected Matrix.
- Input/Output filenames.

[Boulder]

Yep, I kind of guessed that DC precision would be in the ini file because editing wouldn't work.

What happens when you prepare a project, then edit DC prec in the DVD-RB options and then exit DVD-RB, open it again and begin encoding? Would it be possible to add the feature to use the value set in rebuilder.ecl in DVD-RB?

[jdobbs]

For CCE the value is set... because it was already added to the ECL file. For HC and QuEnc, it is read at startup or when a project is loaded, and only changes when it is done via the Options Panel. I could make it read the value from the ECL file easily enough...

[Boulder]

IMO it would be great if all the settings were available in the ecl file, at the moment the DC precision is the only one not available there :)

[jdobbs]

GOP size is also read from the INI rather than the ECL. I'm trying to decide whether moving it is a good idea. If it's stored in the ECL file, you'd have to reaccomplish the PREPARE in order to change it. But, on the other hand, that's how it is with CCE now (since it uses the ECL no matter what).

[Voodoochild]

Hi
I saw that Q value from DVDRB is Different from RB-Opt.
DVDRB uses 2% of the movie, I think as default while RB-Opt use 1% even though that can be changed easily.

My Question, is it better to set the DVD-RB Q value with RB-Opt, or to let RB-Opt to do its own calculation?

10x
Elad

[Boulder]

Does the rebuilding phase need the GOP length and structure information? Many times I've had to rerun the prepare phase because I've forgotten to switch GOP length to 15 frames or to auto depending on the encoder I've used.

[Boulder]

Hi
I saw that Q value from DVDRB is Different from RB-Opt.
DVDRB uses 2% of the movie, I think as default while RB-Opt use 1% even though that can be changed easily.

My Question, is it better to set the DVD-RB Q value with RB-Opt, or to let RB-Opt to do its own calculation?

10x
EladYou can't say, it's more like a matter of luck. Theoretically, the larger the sample size, the better the accuracy but in real life, it's not that simple. The calculations probably differ as well. tylo's CCEFront has a quite good accuracy, I don't know if either DVD-RB or RB-Opt uses its method.

[Voodoochild]

You can't say, it's more like a matter of luck. Theoretically, the larger the sample size, the better the accuracy but in real life, it's not that simple. The calculations probably differ as well. tylo's CCEFront has a quite good accuracy, I don't know if either DVD-RB or RB-Opt uses its method.

10x Boulder, I guess I'll use DVD-RB Q value, if that comes undersize by match or Oversize, I can always set the opv higher / lower and encode again.

10x
Elad

[Boulder]

I caught some bugs in v0.25beta: when you are in Tweak Cells menu and use the arrow up or down keys on your keyboard, the focus jumps to the first cell. This also at least once reset all the changes I had made.

Also when I had edited the avg bitrates, applied the changes and then tweaked the average bitrate of the whole VTS in the main window to avoid oversizing, the bitrates were reset to the original values. I don't know if this is done on purpose, DVD-RB does the same thing. Still it's not convenient at all :)

EDIT: actually when you tweak the cells, click on apply and then go back to tweaking them, the bitrates are reset. The "autosized" tick box is not reset though.

[robot1]

I caught some bugs in v0.25beta: when you are in Tweak Cells menu and use the arrow up or down keys on your keyboard, the focus jumps to the first cell. This also at least once reset all the changes I had made.
Thank you for your report.
This bug is fixed for v0.26beta.


Also when I had edited the avg bitrates, applied the changes and then tweaked the average bitrate of the whole VTS in the main window to avoid oversizing, the bitrates were reset to the original values. I don't know if this is done on purpose, DVD-RB does the same thing. Still it's not convenient at all :)

EDIT: actually when you tweak the cells, click on apply and then go back to tweaking them, the bitrates are reset. The "autosized" tick box is not reset though.
I'm not sure about your process.
You should:
1) tweak avg bitrates (in the main windows)
2) tweak single cells bitrates.
If you change the avg bitrate of the VobID (even if you join vobid's and then change the avg bitrate) you lose your edits.
Please report back your results.

[humax]

Nice - a new version available

thx robot

[Boulder]

I'm not sure about your process.
You should:
1) tweak avg bitrates (in the main windows)
2) tweak single cells bitrates.
If you change the avg bitrate of the VobID (even if you join vobid's and then change the avg bitrate) you lose your edits.
Please report back your results.
I just got it reproduced.

1) Tweaked a couple of avg bitrates of in Tweak cells.
2) Switched to Excel (Office 2003) to get more numbers to feed. I didn't actually do anything, just ALT+TABbing there and then back again.
3) Bitrates were reset.

I noticed that it doesn't happen every time but very often.

Try disabling autosized from the first cell, then enter some value in the input field. Choose the second cell and switch to Excel (I don't know if any other programs cause this, maybe it's just the Office 2003 package), then switch back to RB-Opt. The first cell is reset. If you have tweaked many cells, all of them will be reset.

When you tweak only one cell and keep it selected when switching to Excel, the cell is not reset. When it is not selected, it will be reset.

The funny thing is that you don't even need to do anything in Excel:confused:

[Sharc]

@Boulder:
I could reproduce this 'oddity' here.
Workaround is to adhere to this sequence:
1) Tweak the cells in RB-Opt
2) Push <Apply> (=> Close the "Tweak Cells" window)
3) Switch to Excel
4) Switch back to RB-Opt and push <Tweak cells>
5) go to 1) ......

Not so convenient, but it will preserve your tweaks.

Actually it's not related to EXCEL. It happens whenever the "Tweak Cells" window looses focus.

[Boulder]

Actually it's not related to EXCEL. It happens whenever the "Tweak Cells" window looses focus.When I wrote that, I had tried switching to DVD-RB and Notepad but they didn't cause the problem, I could reproduce it only with Excel:devil:

[robot1]

Actually it's not related to EXCEL. It happens whenever the "Tweak Cells" window looses focus.
Ok, bug reproduced.
:devil:

[Boulder]

If you change the avg bitrate of the VobID (even if you join vobid's and then change the avg bitrate) you lose your edits.This option is something that would be nice to expand so that when you have edited the bitrates and touch the average bitrate of the VOBID/joined VOBIDs, the adjustments would be done according to the edited values instead of the original ones. This means that the calculations would have to be adjusted so that the ratios of avg bitrates between cells remain the same.

[robot1]

Here is the new version, with some new features and bug fixing
RB-Opt v0.26 (http://www.savefile.com/files/648426)

Changelog:
Fixed the bugs reported in the posts previous posts
Added a new feature: "Bitrate redistribution". This feature now works with CCE (and also with ProCoder2 and AQE). I will support HC and QuEnc for the next release.

In short, the idea is by Boulder in the thread http://forum.doom9.org/showthread.php?t=122351
For 99.9% of the discs, the way DVD-RB allocates the bitrate is optimal. But there are some bad mastered discs, encoded in CBR. In these cases the default bitrate allocation is sub-optimal.
For these backups, an OPV method could give better quality.
With the "Bitrate redistribution" you can have an encoding similiar to OPV, but with two advantages:
1) Perfect size
2) ability to further tweak the end credits cell, and/or use different matrices.

I will appreciate some feedback.
Mich