OPV issue in .75

[wfn1]

when OPV is selected in 1C mode, the preparation phase goes through the regular Q calculations but when RB goes to encode it does the usual multipass on my machine (CCE 2.70.01.05)

[TheSeeker]

That is strange. Try using it in 3 click mode and let us know what happens.. I have used OPV with 2.70 and .75 already so I know it works... I just may be something about one click mode that is messing it up

[wfn1]

well, i tried it in 3C mode and the same thing happens, the opv analysis goes through but when i hit the encode button it starts building a .vaf file for my usual 5 pass multipass encode (the cce window shows 1/5 passes)

maybe it has to do something with the .ini? i copied it over from .74 but it worked just fine in OPV...

this is what i have in the .ini:

[CCEOptions]
VBR_bias=25
Quality_prec=16
eclPasses=5
OPV=1

[Sir Didymus]

It once happened to me exactely the same problem...

But it wasn't a problem... it was my fault: (unwanted, of course, and very silly, but...) I changed inadvertitely the encoding mode after the prepare step...

Please check and double check that ALL OF your settings (Mode, Setup, Options) are coherent; if you are not sure about the origin of your rebuilder.ecl file, start a new, clean project from scratch...

If necessary you may delete the rebuilder.ini file...

[wfn1]

the project was prepared with OPV checked (if it werent RB would give an error about preparation/encode mode mismatch when i'd try to hit encode). Please note in 1C mode you cant even possibly have that because there's no user intervention after you hit transcode. i'll do a screen cap later so you can see the RB log window showing OPV events and the CCE window with a multipass encode.

also, i tried starting the project from scratch and that didn't do it. i will also try to erase the .ini tonight and see if it does anything.

[The eDealer]

Same problem to me, with 3C mode.
And the Q is always 1 or 0 (CCE 2.70)?! Same DVD with OPV prediction of RB Opt Q is also 1 or 0 (Source DVD 7.561.713.664 Bytes)?

[jdobbs]

I'll look at it... but I've done numerous OPV encodes since releasing 0.75 and they've all worked.

[wfn1]

yes, the q is actually 0 in all of my encodes. i tried 3 discs so far.

[The eDealer]

Some more testing using DVD-RB 0.75 and
- Versions:
-- CCE SP Version: v2.70.1.5
-- DECODER Version: DGMPEGDec 1.2.0RC2
-- AVISYNTH Version: v2.5.5.0

Is it a dvd-source size/reduction-level problem?


Hellboy, RC2, (8.024.717.312 Bytes)

[06:15:56] Phase I, PREPARATION started.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_04: 3.096.720 sectors.
-- Scanning and writing .D2V file
-- Processed 190.575 frames.
-- Building .AVS and .ECL files
- VTS_07: 352.394 sectors.
-- Scanning and writing .D2V file
-- Processed 30.104 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 57,8%
- Overall Bitrate : 2.865Kbs
- Space for Video : 3.086.862KB
- Analyzing VTS_04 for optimal Q factor.
- TargetSize (sectors):1.369.807
- Sampling 9540 of 190575 frames.
-- Predicted size (sectors) at Q=38: 819.410
-- Predicted size (sectors) at Q=17: 985.867
-- Predicted size (sectors) at Q=11: 1.098.410
-- Predicted size (sectors) at Q=8: 1.124.408
-- Predicted size (sectors) at Q=6: 1.253.890
-- Predicted size (sectors) at Q=5: 1.306.283
- Q Value selected: 5
- Analyzing VTS_07 for optimal Q factor.
- TargetSize (sectors):196.775
- Sampling 1512 of 30104 frames.
-- Predicted size (sectors) at Q=42: 124.923
-- Predicted size (sectors) at Q=21: 153.320
-- Predicted size (sectors) at Q=15: 161.494
-- Predicted size (sectors) at Q=11: 173.888
-- Predicted size (sectors) at Q=9: 176.603
-- Predicted size (sectors) at Q=8: 178.022
-- Predicted size (sectors) at Q=7: 184.107
-- Predicted size (sectors) at Q=6: 198.101
- Q Value selected: 7
- HIGH/LOW/AVERAGE Cell Bitrates: 4.270/143/2.865 Kbs
[06:40:17] Phase I, PREPARATION completed in 25 minutes.


New York Minute, RC2 (5.201.422.336 Bytes)
-----------------
[07:13:55] Phase I, PREPARATION started.
- "Half-D1/Half Space for Extras" mode is enabled.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_01: 2.085.539 sectors.
-- Scanning and writing .D2V file
-- Processed 134.029 frames.
-- Building .AVS and .ECL files
- VTS_02: 447.702 sectors.
-- Scanning and writing .D2V file
-- Processed 38.318 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 94,5%
- Overall Bitrate : 5.022Kbs
- Space for Video : 4.226.356KB
- Analyzing VTS_01 for optimal Q factor.
- TargetSize (sectors):1.941.140
- Sampling 6708 of 134029 frames.
-- Predicted size (sectors) at Q=19: 686.514
-- Predicted size (sectors) at Q=2: 950.847
-- Predicted size (sectors) at Q=1: 1.013.638
-- Predicted size (sectors) at Q=0: 1.013.638
- Q Value selected: 0
- Analyzing VTS_02 for optimal Q factor.
- TargetSize (sectors):203.736
- Sampling 1920 of 38318 frames.
-- Predicted size (sectors) at Q=51: 91.777
-- Predicted size (sectors) at Q=13: 129.338
-- Predicted size (sectors) at Q=7: 143.975
-- Predicted size (sectors) at Q=4: 154.029
-- Predicted size (sectors) at Q=3: 155.642
-- Predicted size (sectors) at Q=2: 160.084
-- Predicted size (sectors) at Q=1: 165.854
- Q Value selected: 1
- Movie improvement from extra reduction = 11,7%
- HIGH/LOW/AVERAGE Cell Bitrates: 7.016/2.145/5.022 Kbs
[07:26:46] Phase I, PREPARATION completed in 13 minutes.

[wfn1]

*** EDIT: I was able to reproduce the error with CCE SP 2.50 and 2.66.01.07. Moreover, erasing the .ini file and re-setting everything up did not help in any shape. RB's Q value choices are consistently a 0 or a -1. I tried a total of 4 DVDs already.


well, this is what i have for my latest attempt:

Sex and the city R1, S6 Part2 Disc2

-----------------
[02:23:09] Phase I, PREPARATION started.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_01: 3,399,652 sectors.
-- Scanning and writing .D2V file
-- Processed 201,843 frames.
-- Building .AVS and .ECL files
- VTS_02: 111,463 sectors.
-- Scanning and writing .D2V file
-- Processed 9,152 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 62.8%
- Overall Bitrate : 3,521Kbs
- Space for Video : 3,782,720KB
- Analyzing VTS_01 for optimal Q factor.
- TargetSize (sectors):1,851,425
- Sampling 2028 of 201843 frames.
-- Predicted size (sectors) at Q=28: 206,762
-- Predicted size (sectors) at Q=-6: 632,414
-- Predicted size (sectors) at Q=-1: 342,599
- Q Value selected: -1
- Analyzing VTS_02 for optimal Q factor.
- TargetSize (sectors):68,305
- Sampling 96 of 9152 frames.
-- Predicted size (sectors) at Q=34: 9,653
-- Predicted size (sectors) at Q=-5: 27,381
-- Predicted size (sectors) at Q=-1: 16,595
- Q Value selected: -1
- HIGH/LOW/AVERAGE Cell Bitrates: 3,595/393/3,521 Kbs
[02:28:48] Phase I, PREPARATION completed in 5 minutes.

- Versions:
-- CCE SP Version: v2.70.1.5
-- DECODER Version: vUNKNOWN
-- AVISYNTH Version: v2.5.5.0


the decoder is 1.2.0rc4

[pg55555]

Looking at wfn1 and The eDealer posts it is clear there is something wrong with the Q estimation: No way for a 55% reducion you get a Q of 5 or 7. And a negative value? Maybe thats the reason CCE is going back to VBR

Both of them are using the DGMPEGDec 1.2.0RC2 or RC4. I would suggest try DGMPEGDec 1.1

[jdobbs]

Hmmmm.... looking at the resulting output sizes of the passes it looks like it is doing the right thing... lowering the Q until it maximizes the size.

I'll check tonight -- it looks like I should have set a minimum value for Q of 1, but the problem appears to be the output size from CCE based upon the Q provided. Have you looked at the .AVS files to see if it is being fed valid data?

[wfn1]

OK, for the record the .avs files are 100% kosher whether I use dgdecode 1.2.0rc4 or 1.1.0.0

BUT, with 1.1.0.0 the Q values are actually valid. I just did the same disc that gave me wacky Qs and here's the status window:

-----------------
[10:43:59] Phase I, PREPARATION started.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_01: 3,399,652 sectors.
-- Scanning and writing .D2V file
-- Processed 201,843 frames.
-- Building .AVS and .ECL files
- VTS_02: 111,463 sectors.
-- Scanning and writing .D2V file
-- Processed 9,152 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 62.8%
- Overall Bitrate : 3,521Kbs
- Space for Video : 3,782,720KB
- Analyzing VTS_01 for optimal Q factor.
- TargetSize (sectors):1,851,425
- Sampling 2028 of 201843 frames.
-- Predicted size (sectors) at Q=28: 979,017
-- Predicted size (sectors) at Q=10: 1,386,864
-- Predicted size (sectors) at Q=7: 1,408,265
-- Predicted size (sectors) at Q=5: 1,416,905
-- Predicted size (sectors) at Q=3: 1,423,227
-- Predicted size (sectors) at Q=2: 1,426,159
-- Predicted size (sectors) at Q=1: 1,427,902
- Q Value selected: 1
- Analyzing VTS_02 for optimal Q factor.
- TargetSize (sectors):68,305
- Sampling 96 of 9152 frames.
-- Predicted size (sectors) at Q=34: 60,187
-- Predicted size (sectors) at Q=29: 65,965
-- Predicted size (sectors) at Q=28: 67,249
-- Predicted size (sectors) at Q=27: 68,791
- Q Value selected: 27
- HIGH/LOW/AVERAGE Cell Bitrates: 3,595/393/3,521 Kbs
[10:52:25] Phase I, PREPARATION completed in 9 minutes.


i'm trying to encode it right now and see if it'll go through. Thanks pg55555

[The eDealer]

AVS File from the "New York minute" trial:
#------------------
# AVS File Created by DVD Rebuilder
# VOBID:03, CELLID:03
#------------------
LoadPlugin("C:\Programme\AviSynth 2.5\plugins\DGDecode.dll")
mpeg2source("E:\!DVD-RB WORKING\D2VAVS\V01.D2V",idct=7)
trim(15614,19725)
ConvertToYUY2()

Extras:

#------------------
# AVS File Created by DVD Rebuilder
# VOBID:04, CELLID:01
#------------------
LoadPlugin("C:\Programme\AviSynth 2.5\plugins\DGDecode.dll")
mpeg2source("E:\!DVD-RB WORKING\D2VAVS\V02.D2V",idct=7)
trim(33677,38317)
BilinearResize(352,576)
ConvertToYUY2(interlaced=true)

Everything is fine.


I have made another test with an old DGDecode.dll version.
with this version the quantizer scale (I;P;B)and the bitrates (avg. & curr.) changes during the analyse scan, with the new version it's constant!
Here the result for the same DVD with old encoder:

[18:23:09] Phase I, PREPARATION started.
- "Half-D1/Half Space for Extras" mode is enabled.
- "One Pass VBR (w/analysis)" mode is enabled.
- VTS_01: 2.085.539 sectors.
-- Scanning and writing .D2V file
-- Processed 134.029 frames.
-- Building .AVS and .ECL files
- VTS_02: 447.702 sectors.
-- Scanning and writing .D2V file
-- Processed 38.318 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 94,5%
- Overall Bitrate : 5.022Kbs
- Space for Video : 4.226.356KB
- Analyzing VTS_01 for optimal Q factor.
- TargetSize (sectors):1.941.140
- Sampling 6708 of 134029 frames.
-- Predicted size (sectors) at Q=19: 2.174.942
-- Predicted size (sectors) at Q=22: 2.061.321
-- Predicted size (sectors) at Q=24: 1.988.770
- Q Value selected: 24
- Analyzing VTS_02 for optimal Q factor.
- TargetSize (sectors):203.736
- Sampling 1920 of 38318 frames.
-- Predicted size (sectors) at Q=51: 180.030
-- Predicted size (sectors) at Q=43: 199.856
-- Predicted size (sectors) at Q=42: 202.563
- Q Value selected: 42
- Movie improvement from extra reduction = 11,7%
- HIGH/LOW/AVERAGE Cell Bitrates: 7.016/2.145/5.022 Kbs
[18:42:53] Phase I, PREPARATION completed in 19 minutes.


So the problem is the coder version!
I will test now the coder versions beginning from 1.1.0 till 1.2.0RC2.

Results will follow.

[The eDealer]

OK, all version work fine till 1.2.0rc1!
With 1.2.0rc2 the problem comes up.

[jdobbs]

How in the world can the decoder affect this? It must be related to the way it handles random access -- since the .ECLs created for prediction use the "SelectEvery" functions.

[The eDealer]

I dont know, but the cce doesn't change the BR an the number of I,P and B frames. You can test it.

[wfn1]

hmm, my latest encode with dgdecode 1.1.0.0 completed fine in opv mode but i got an error 6 overflow on the rebuild. (the target folder was over 6GB when RB errored out)

[hobyho]

I can verify that the latest dgdecode 1.2.0rc2 does not work with the current 0.75RB. Says the d2v file is obsolete.

[jdobbs]

Don't use it. ;) Upgrade for a reason, not just to upgrade. What does it do that improves its use in DVD-RB? You can install DGIndex separate from DVD-RB if you need to use it for some reason. Just install the DGDECODE.DLL version that DVD-RB needs in the RB directory -- and point to it in DVD-RB.

[pg55555]

jdobbs

I'm glad to confirm my first experience with RB 0.75 OPV has resulted in a BIG improvement over previous version.

I did Sex and The City Season 6 Disk 3 (R4 NTSC) and the output size was right on target (RB 0.75 CCE 2.66, Avisinth 2.55 and DGDECODE.DLL 1.1).

With RB 0.74, same setup, the output has been undersized for almost 15%.

Keep the good job.

[wfn1]

pg55555,

i'm doing that series too and i'm having super good luck with the most discs but for some reason i cannot do Season 6 Part 1 Disc 2 and Season 6 Part 2 Disc 2. Now that the DGDecode issue has been sorted out those are the only 2 discs I just cannot rebuild. I get error 6 on both. Is there a chance you could try to do on and tell me if it wors for you? Thanks!

[pg55555]

I have completed all the Season 6 successfully, but probably the release here in R4 (South America) is different from yours as we do not have Part 1 and 2, just Disc 1 to 5.

I used to have problems (error 6) with RB 0.73 and OPV, but it was solved in 0.74

I would suggest trying with CCE 2.66. good luck

[gizzin]

It seems like .75 does accurate encodes to the right file size. I've done Gladiator R1 with .73 and it was slightly oversized
but with .75 it came out to 4.38 keep up the good work jdobbs. Remember that this is only 1 DVD.

[The eDealer]

@jdobbs
btw. is it possible to contact dg to discuss the missmatch in opv mode beginning with the version 1.2.0rc2?
It would be nice to need only one version on the system :) .

[jdobbs]

All you need is the correct dgdecode.dll file for DVD-RB. Just keep it in the DVD-RB executable directory and point to it. DVD-RB works independently of DGINDEX. You can keep whatever version you want of DGIndex installed and they won't interfere with each other.

It's just an incredible amount of effort having to constantly change DVD-RB... I have to do it for DGDECODE, QuEnc, and CCE. There seems to be a change on almost a daily basis.

[wfn1]

i'd like to report that once i got everything sorted the OPV mode works beautifully. i mean i get an awesomely close approximation every time. did 7 or 8 discs and they all fall between 4,600,000,000 and 4,66x,xxx,xxx bytes. thanks JD!