VBR bias, Quality Precision recommendations?

[archaeo]

Being new to RB, I've used it primarily for dvd backups with the default VBR bias and Quality Precision settings set at 25 and 16, respectively. This has worked very well for most of what I've done, but there have been a few sources where tweaking these settings would probably have helped. Problem is, I am new to understanding how these values work, and I'd like to learn how to adjust these settings for achieving the best possible results on my backups, based upon the source dvd characteristics.

I've been following a few of the posts on this forum that discuss changing these settings based upon bitrate and length of the original movie source. Just so you know, I use CCE basic, so it appears that I cannot use OPV for predictive measures. However, I do have bitrate viewer which gives me a basic anaysis tool for the original.

The general 'rules of thumb' I have seen so far have been:

Original sources that have an average bitrate >3000, Quantizer(Quality Precision) should be set between 27-30.

VBR bias is recommended to be 40-50% of QP, or more loosely, should be set to 23-25 for movies <2hrs, and 15-23 for those >2hrs. 3hr movies should be set to 15-20 VBR bias...

These are pretty general, and I welcome being steered toward some sort of guideline or tool that I could use to help me make better estimations of settings. Again, without being able to use OPV, I may be limited in what type of predictive measures I could utilize.

If you have a thread or any suggestions on what you use to predict these settings, I'm wide open. Is bitrate viewer something that could be of some use for these predictive measures?

thanks

[archaeo]

no suggestions or threads to recommend?

[Fr4nz]

You could use the search option.....

Anyway take a look here: http://forum.doom9.org/showthread.php?s=&threadid=77171

[robot1]

Originally posted by archaeo
Just so you know, I use CCE basic, so it appears that I cannot use OPV for predictive measures. If I'm not wrong, CCE basic doesn't support neither VBR bias nor Quantizer.

[archaeo]

Fr4nz,

Thanks for the thread - I have done searches, and I usually try to answer my questions in that way to begin with.

However, in my searches I've turned up some threads that state: "I tend to concur with RB's findings that ..."

Where are these RB findings? Are they posted somewhere? Only having been on this board for a month or so, I may have missed something that I haven't turned up on a search yet.

In any event, I'm just looking for some input from members on how they predict optimum VBR and QP settings without the use of OPV.

[archaeo]

If I'm not wrong, CCE basic doesn't support neither VBR bias nor Quantizer.

really?? I thought that those settings were supported through RB, even with CCE Basic (2.69).

[robot1]

It's not a DVD-RB fault. It's the basic version limited.
Anyway, for normal disks the result is excellent with internal default parameters.

[archaeo]

Is this true that CCE basic will not support the RB VBR bias and QP settings adjustment?

I ask because in RB the advanced expert setup box for CCE basic (2.67+) has VBR and QP values boxes that can be adjusted - They're not greyed out or anything...

[archaeo]

robot1 is correct - the vbr_bias and quality_prec default settings are FIXED when using CCE Basic.

[RB]

The latest CCE Basic at least supports quality_prec (AKA Quantizer Characteristics, Flat Part Priority).

[archaeo]

The latest CCE Basic at least supports quality_prec

Would that be v2.69?

[RB]

Yes, 2.69.


----------

Version 2.69.01.00 Monday, January 19, 2004

Change
- Updated encoder engine to 2.69.

Enhancements
- Supported Adobe Premiere Pro. (plug-in version)
- Enabled to output bottom field first stream.
- Added automatic resolution conversion function.
- Added quantization characteristics setting.
- Added de-interlacing function.
- Added blanking function.
- Enabled to save/load template to/from file.
- Enabled to use the last modified parameter setting.
- Enabled to specify file size.
- Enabled to shut down the system when finished encoding.
- Enabled to play a sound file when finished encoding.

----------

[jdobbs]

Originally posted by RB
Yes, 2.69.


----------

Version 2.69.01.00 Monday, January 19, 2004

Change
- Updated encoder engine to 2.69.

Enhancements
- Supported Adobe Premiere Pro. (plug-in version)
- Enabled to output bottom field first stream.
- Added automatic resolution conversion function.
- Added quantization characteristics setting.
- Added de-interlacing function.
- Added blanking function.
- Enabled to save/load template to/from file.
- Enabled to use the last modified parameter setting.
- Enabled to specify file size.
- Enabled to shut down the system when finished encoding.
- Enabled to play a sound file when finished encoding.

----------
Great. Now I have to come up with a way of telling whether CCE Basic is 2.69 and above... as much as I love CCE I wish they'd remain consistent. Does anyone know if the syntax is the same as other versions?

[robot1]

Originally posted by jdobbs
Great. Now I have to come up with a way of telling whether CCE Basic is 2.69 and above... as much as I love CCE I wish they'd remain consistent. Does anyone know if the syntax is the same as other versions? Quantization characteristic (flat part priority) has 0-100 range, and is saved as "pict_qchar" in ecl.
Default is "pict_qchar=16", if I'm not wrong.
Quantization characteristic can't be used if you use filters :confused:

[RB]

I'm a little surprised nobody noticed this earlier :), CCE-Basic 2.69 is out for several months now, the last update to 2.69.01.08 was just released today.
Originally posted by jdobbs
Great. Now I have to come up with a way of telling whether CCE Basic is 2.69 and above...
Maybe they store the version in the registry, but it would be safest to query the .EXE version resource. Here's some C code I'm using in EclCCE

LARGE_INTEGER WINAPI GetCCEVer(char *szCCEPath)
{
DWORD dwVerSize, dwDummy;
VS_FIXEDFILEINFO *pvfi;
void *pData = NULL;
LARGE_INTEGER Version = {0,0};

__try
{
if (0 == (dwVerSize = GetFileVersionInfoSize(szCCEPath, &dwDummy)))
__leave;

if (NULL == (pData = malloc(dwVerSize)))
__leave;

if (!GetFileVersionInfo(szCCEPath, NULL, dwVerSize, pData))
__leave;

if (!VerQueryValue(pData, "\\", (void**)&pvfi, (UINT*)&dwVerSize))
__leave;

Version.LowPart = pvfi->dwProductVersionLS;
Version.HighPart = pvfi->dwProductVersionMS;
}

__finally
{
if (pData)
free(pData);
}

return Version;
}

[jdobbs]

Cool RB -- you're a real code warrior!

[hobyho]

I did note this in the bugs section of DVD-Rebuilder, but I guess it got lost by other posts.

Though I guess it really isn't a bug, but a feature that needed to be added/implemented to reflect the new version.

Atm, I have to manually edit the Rebuilder.ecl file by replacing pict_complexity=2 with pict_qchar=X. Works fine, but would be better if rebuilder can make the necessary adjustments.

Anyway, we appreciate your work jdobbs and thanks for a great piece of software.

[EddieTH]

Originally posted by RB
I'm a little surprised nobody noticed this earlier :), CCE-Basic 2.69 is out for several months now, the last update to 2.69.01.08 was just released today.
Actually I mentioned it a couple of times in one of the DVD-RB Suggestions threads and another time in one of the other threads in this forum. I think it just got lost in the shuffle along the way. I didn't consider it a huge deal since I was already adding it manually and I figured it would get done eventually.

[jdobbs]

Just one more rant... I wish Cinema Craft would make up their mind as to the scale they are using... In CCE SP 2.50 the scale was 0-100, then in version 2.66 it was changed to 0-64, now they've added it to CCE Basic -- and the scale is back to 0-100... so it doesn't match between the newer versions.

A little consistency would be nice. I can't even count the number of lines of code I've had to write for all the special circumstances associated with different versions on all their settings. (for example they've changed "filter_off=1" to "filter_on=0" -- what is up with that?)
:angry:

[Fr4nz]

Originally posted by jdobbs
Just one more rant... I wish Cinema Craft would make up their mind as to the scale they are using... In CCE SP 2.50 the scale was 0-100, then in version 2.66 it was changed to 0-64, now they've added it to CCE Basic -- and the scale is back to 0-100... so it doesn't match between the newer versions.

One question jdobbs: are these two ranges proportional?

I'll explain mysel better: if I use vbr_bias 20 in CCE 2.50, will I have to use 13 in CCE 2.66? And regarding quality_precision: if I use 15 with 2.50, will I have to use 9-10 with 2.66, in order have CCE working with similar parameters?

[jdobbs]

Originally posted by Fr4nz
One question jdobbs: are these two ranges proportional?

I'll explain mysel better: if I use vbr_bias 20 in CCE 2.50, will I have to use 13 in CCE 2.66? And regarding quality_precision: if I use 15 with 2.50, will I have to use 9-10 with 2.66, in order have CCE working with similar parameters? Yes they are proportional, so the value set in the CCE dialog gets multiplied by 1.56 before it is written to ECLs for 2.50 or CCE Basic (starting in DVD-RB version 0.53).

[robot1]

Originally posted by jdobbs
(for example they've changed "filter_off=1" to "filter_on=0" -- what is up with that?)
:angry: Anyway filter_off still works with 2.69.

[quantum]

Originally posted by robot1
Anyway filter_off still works with 2.69.

True. If I recall correctly, 2.69 saves ecl's using new keys, but will correctly open ecl's using the 2.67 format.

[jdobbs]

I just put both sets in. The new one will read its setting and the old will read the ones for it. I also noticed that older versions won't choke on the pict_qchar= setting.

[Jeffster]

Strangely enough I reported a problem with the new ecl format in this thread (http://forum.doom9.org/showthread.php?threadid=76681) three weeks ago, with 270 views and not a single reply. I didn't know what to make of that, I still don't know if I'm the only one who has experienced this?

Anyway I haven't used RB since, but I just ran a test now with 0.52 and still get the same error.

Comparing the ecl that RB creates and one from 2.69 and there are seventeen new settings in the 2.69 ecl... only 2 have been mentioned in this thread, so I wonder what impact they may have also, if any. :confused:

Regards
Jeff