DVD PAL - Always oversized !

[crabougnaf]

Hello everybody,

I've a problem with DVD RB, the result is always oversized. I'm in zone 2 and trying to back up a 6.09 GB DVD, result is always 5.4GB
I think avisynth and CCE (2.5) are properly installed since I had good results with previous versions of DVD RB (0.46 & 0.49).
I'm now using v0.51c

Here's a my config file :
--------------
[Options]
LoadPlugin=1
EncoderMinimized=1
BackColor=-2147483633
DECOMBFlag=1
CCE=0
HalfExtras=1
NoWarn=1
OneClick=0
Completed=10
Mode=1
CCETargetSectors=2260007
idct7Opt=1
[CCEOptions]
VBR_bias=25
Quality_prec=16
eclPasses=3
OPV=1
-------------------

I've checked paths, and tried different settings ( half D1 yes/no, deinterlace with decomb yes/no,...).
This DVD contains interlaced material (some extras) and I removed 1 soundtrack. Whatever my settings are, I always get an 5.4GB output.

Here's a part of the status window during preparation :
----------------------------
-----------------
[19:15:48] Phase I, PREPARATION started.
- "Half-D1/Half Space for Extras" mode is enabled.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_01: 2_282_758 sectors.
-- Scanning and writing .D2V file
-- Processed 116_000 frames.
-- Building .AVS and .ECL files
- VTS_03: 143_686 sectors.
-- Scanning and writing .D2V file
-- Processed 8_675 frames.
-- Building .AVS and .ECL files
- VTS_07: 113_210 sectors.
-- Scanning and writing .D2V file
-- Processed 7_425 frames.
-- Building .AVS and .ECL files
- VTS_08: 644_453 sectors.
-- Scanning and writing .D2V file
-- Processed 40_900 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 69,0%
- Overall Bitrate : 4_641Kbs
- Space for Video : 3_920_294KB
- Analyzing VTS_01 for optimal Q factor.
- TargetSize (sectors): 1668063
-- Predicted size (sectors) at Q=19: 1696299
- Q Value selected: 20
- Analyzing VTS_03 for optimal Q factor.
- TargetSize (sectors): 93008
-- Predicted size (sectors) at Q=25: 123619
-- Predicted size (sectors) at Q=33: 123619
-- Predicted size (sectors) at Q=44: 123619
-- Predicted size (sectors) at Q=58: 123619
-- Predicted size (sectors) at Q=77: 123619
-- Predicted size (sectors) at Q=102: 123619
- Q Value selected: 137
- Analyzing VTS_07 for optimal Q factor.
- TargetSize (sectors): 75315
-- Predicted size (sectors) at Q=27: 105806
-- Predicted size (sectors) at Q=38: 105806
-- Predicted size (sectors) at Q=53: 105806
-- Predicted size (sectors) at Q=74: 105806
-- Predicted size (sectors) at Q=104: 105806
- Q Value selected: 147
- Analyzing VTS_08 for optimal Q factor.
- TargetSize (sectors): 415835
-- Predicted size (sectors) at Q=27: 596258
-- Predicted size (sectors) at Q=39: 596258
-- Predicted size (sectors) at Q=56: 596258
-- Predicted size (sectors) at Q=80: 596258
- Q Value selected: 116
- Movie improvement from extra reduction = 21,2%
- HIGH/LOW/AVERAGE Cell Bitrates: 6_196/688/4_641 Kbs
[19:26:52] Phase I, PREPARATION completed in 11 minutes.
---------------------------------------------


We can see that predicted size is always the same, is it the expected behaviour ??

I'm really new to DVD backup, so it may be a really stupid problem. I'm sorry about that, but any help would be really welcome.

Thank you in advance,

Crabougnaf

P.S. : Please forgive my froggy's English ;-)

[nimbles]

this maybe a bit thick but i can't see any reference to the mpegdec2.dll (or valid variants of this like mpeg2dec3dg.dll)

this is from wmansir's setup sticky:

You will need ALL of these tools (all are free):

* DVD Decrypter (Version 3.2.1.0 or above) or another ripping program (if backing up encrypted DVDs)
* AVISYNTH installed (Version 2.5 or above recommended)
* MPEG2DEC3 Avisynth Plugin (without lost-frame-bug) DVD-RB is not compatible with MPEG2Dec 1.02 and greater and DGDecoder. If you experience problems with your .avs files, be sure and check your version. When in doubt, use the version linked here.
* DVD Rebuilder, of course.


have you pointed dvd-rb to the mpeg2dec.dll/mpeg2dec3dg.dll?

I think the one you get in the decodefix110 (http://www.avisynth.org/warpenterprises/files/decodefix110.zip) is the last valid mpeg2dec3dg.dll that can be used, unless i missed an update :confused:

[writersblock29]

@crabougnaf

Two things you'll be warned about, based off of your log file.

1) OPV is still highly experimental, and even though Jdobbs is getting closer to refining it, you're still bound to have a bit of hit-or-miss involving the target sizing. Based off of other posts I've read, he's working on it, though!

2) CCETargetSectors=2260007 isn't Rebuilder's default setting. I myself had fairly consistant luck with this setting using previous versions of Rebuilder (always using standard multipass protocol)... but the last few versions have been iffy.

You might try deleting the CCETargetSectors line from your file, and give 'er another go to see what happens. Either way, post that log file so we can see how you did. If it doesn't give you what you need, you can always try multipass.

(Sound about right, Jdobbs? ;) )

[crabougnaf]

Hi,

Thanks for the answerw.

@nimbles : I didn't paste the first part of my config file because I thought it was not relevant. Versions and paths are correct : I used them before with RB 0.46 & 0.49 successfully.

@writersblock29

1) I know OPV is under development, but I'd like to give it a try. I've seen in the forum there was problems with target size with PAL DVD before, but it's supposed to be corrected in 0.51c. I wonder if I missused RB or if there's still a bug with size prediction algorithm (maybe with interlaced material ?)

2) I added CCETargetSector because of this problem. But it was the same when I used default ini file.

Is there any file I could provide to help the understanding (avs ?).

If someone succeed in the rebuilding of a PAL DVD with some interlaced material using RB >= 0.50 I would be happy to have some feedback...

I will try multipass tonight to confirm the problem is with OPV and not with this particular DVD. But in the future, I would like to use OPV ;)

ThanX
Crabougnaf


Edit : I've seen v0.52 adds some improvements int OPV. I will try to increase the sample size for the analysis : this DVD is composed of many small VTS (lot of tiny extras), maybe this leads size prediction to be wrong.

[crabougnaf]

Hi,

I've tried RB 0.52 yersteday.

1) First test : I kept my previous settings to give OPV a chance. RB entered what seems to be an endless loop. RB never stopped to try an higher Q value during analysis of the first VTS, predicted sector size was always the same. I had to interrupt it and it crashed, so I don't have any trace of the problem.

2) Secondly, I performed a multipass backup tonight (VAF + 3 pass encoding). The resulting VIDEO_TS was 5.4GB. So I still have the same problem.

Does anybody have an idea about the origin of this problem ?
Can anybody confirm he has performed a zone 2 PAL DVD backup successfully ?

Any help or piece of advice would be welcome, I'm a bit lost...

ThanX
Crabougnaf

[wmansir]

1. I've seen this happen too, but only when I was accidently using an old version of EclCCE. The problem was CCE was outputing .m2v files instead of .mpv after the first test. So the first test output is never overwritten and is subsequently measured over and over again. It looks like the same thing was happening in your previous attempts, but it didn't get stuck in a loop then.

The latest version of EclCCE should fix this if that is causing your problem.

2. I don't know about this because multipass has been reported to be pretty accurate for several releases now. Are you using the "steal space from extras" option, or Half&Half?

[crabougnaf]

Hi,

Thank you for your answer.

1) I'm quite sure I've the last EclCCE because I had such problems before (RB v0.4x) and fixed it by installing the last EclCCE. I will check tonight (I'm at my office now).

2) I had good results with RB v0.46 & v0.49. So I know that multipass is usually accurate. I've tried with and without Half D1, I've never tried "steal space". Both test end with the same oversize problem (exactly the same size.

I will try to re-install avisynth and EclCCE and will check again settings. But I already tried this without any improvement...
Could it be a PAL specific problem ?

Crabougnaf

[Sir Didymus]

Originally posted by crabougnaf
Hi,

[... skip...]

2) Secondly, I performed a multipass backup tonight (VAF + 3 pass encoding). The resulting VIDEO_TS was 5.4GB. So I still have the same problem.

Does anybody have an idea about the origin of this problem ?
Can anybody confirm he has performed a zone 2 PAL DVD backup successfully ?

Any help or piece of advice would be welcome, I'm a bit lost...

ThanX
Crabougnaf

Did you use some preprocessing application after ripping the DVD on your HD ?

[crabougnaf]

Hi,

No, I tried : 1) Rip only iso with DVDDecrypter, mount with Daemons Tools
2) Rip in file mode to get VIDEO_TS and AUDIO_TS and work directly on file.

I have the same problem with the 2 methods...

[jdobbs]

I can tell you there is definitely something wrong. This should never happen:

- TargetSize (sectors): 93008
-- Predicted size (sectors) at Q=25: 123619
-- Predicted size (sectors) at Q=33: 123619
-- Predicted size (sectors) at Q=44: 123619
-- Predicted size (sectors) at Q=58: 123619
-- Predicted size (sectors) at Q=77: 123619
-- Predicted size (sectors) at Q=102: 123619

I'm at a loss to explain it... It appears that the .M2V file is not changing between passes at all... That makes me think it isn't actually being encoded for each. wmansir may be on the right track. Can you post the AVS and ECL settings for the first segment of VTS_03? You might also post the contents of Q_ANALYSIS.ECL if it is still available.

[Added] Also look in the D2VAVS directory and see what date and time you have on each of the .M2V files.

[crabougnaf]

Hi jdobbs,

I will check everything this evening again and, if I don't have better results, I post here the details (AVS,ECL and Q_Analysis).

Thank you,
Crabougnaf

[crabougnaf]

Hi,

I'checked everything last evening, it was a really stupid mistake actually.

Few days ago, I gave CCE 2.67 a try, so I moved my old "CCE" folder to "CCE25" to keep my previous version (CCE 2.50), and installed v2.67 in a new "CCE" folder.
As CCE 2.67 was slower on my machine (ThunderBird), I've decided to give up with 2.67 and ticked CCE 2.5 in RB again. But - stupid guy - I didn't edit EclCCE.ini file : CCE executable path was still pointing to "CCE" folder instead of "CCE25".

So, RB was sending 2.5 style command to EclCCE in the "CCE25" folder, and EclCCE was transmitting these command sequence to the "wrong" CCE situated in the "CCE" folder...

I've said it was stupid...

I'm sorry, I shouldn't have post such a simple problem.
Everything is working now (4.27GB output, great quality in OPV mode).

Thank you all for your help,
Crabougnaf

[EnTaroAdun]

Hey crabougnaf!

I've got exactly the same error, always oversized. How did u exactly fixed it? Thanks for finding out this error for me :)

@jdobbs: Great Work, I think the programm is now nearly perfekt, the percentage stuff for extras is great and the one pass method, too!!

EnTaroAdun

[wmansir]

The problem was like I originally guessed, but for a different reason. Because of the flaky EclCCE install CCE was outputting files with the wrong name, so DVD-RB wasn't reading the right output for each Q value test.

Notice that each Q value gave the same sized output:

-- Predicted size (sectors) at Q=25: 123619
-- Predicted size (sectors) at Q=33: 123619
-- Predicted size (sectors) at Q=44: 123619
-- Predicted size (sectors) at Q=58: 123619
-- Predicted size (sectors) at Q=77: 123619
-- Predicted size (sectors) at Q=102: 123619

That shouldn't happen. For crabougnaf the fix was to make EclCCE point to the right version of CCE (by deleting his eclcce.ini and running eclcce.exe to set it up right). For me, it was updating to the latest version of EclCCE.

[EnTaroAdun]

Hi,

I installed everything new and in the newest version. Now all works fine, thanks for the hint!


EnTaroAdun