AutoIt Error in D2SRoba

[pitchshifter]

Hi
I've been encoding SVCDs with D2S for a while, and have recently begun experimenting with converting avi eps to DVD format. So far i'm having no luck using D2SRoba - the same error message keeps coming up 95% of the time. I have trawled through all subjects on D2Sroba and have only found one similar looking AutoIt error which has since been apparently fixed.

After the end of the D2s log (below), the following message comes up:

(AutoIt Error)
Line 0 (File "D:\Program Files\DVD2SVCD\Tylo\D2SRoBa.exe"):

If ($g_QBitrate[$Q] <> 0) Then
If (^ERROR

Error: Array varible has incorrect number of subscripts or subscript dimension range exceeded.

This has occurred with D2SRoBa versions 3.50, 3.60, 3.63, 3.70, 3.72, the latest build of DVD2SVCD and the one before it. Also after clean reformat with the latest DVD2SVCD and roba 3.70 then installed. My computer also runs 12 hour batch svcd encodes etc perfectly.

I have tried every combination of bitrates, resizing filters, and other settings (completely grabbing at straws but i have no idea what's causing this error). here you can see the Q was detected as 4290 initially?? the one time it randomly worked, it encoded the same movie in Q=1 i believe.

Any suggestions for this bizarre problem would be much appreciated!

--------------------------------------------------------
D2SRoBa v3.72, by Tylo
--------------------------------------------------------
- WIN_XP - AutoIt 3.0.103.0- Run mode : Normal
--------------------------------------------------------
- Waiting for CCE window, or 'Ripping done' message box...

--------------------------------------------------------
- INI file : D:\Program Files\DVD2SVCD\Tylo\D2SRoBa.ini
- Output type : DVD- Detected a CCE encode window
- Detected the movie encode window - shutting down
--------------------------------------------------------
- 2004-09-30 23:04:10
- Copying D2SRoBa.ini to project dir C:\Nip Tuck\01\
- Avisynth setup: BicubicResize
- D2S ver. 1.2.2.3, CCE ver. 2.66.1.7
- Setting process priority 0
--------------------------------------------------------
- Movie length : 01:06:51 (96183 frames, 23.976 fps)
- D2S audio select : 224 + 0 kbps
- D2S video calc. : 1600 kbps, 1 CDs
- Select ranges : every 1200, select 12 frames
- Sample frames : 972
--------------------------------------------------------
Calculations:
- 1 CD: video_br=1600 (1600) audio_br=224+0 video_sz=802327327 fill=100% cbr=0
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Using num CDs : 1
- Target mpv br : 1600 (max 9000) kbps, size=802327327 bytes
--------------------------------------------------------
Search for Q:1978 kbps, err=23.6%, size=992084968, sample sz=10123316
--------------------------------------------------------
- Determined Q : 4290 = Round(200 + (23.6 - 1.2)/0)

- Sample enc. Q=50 : 2144 kbps, err=34%, size=1075404176, sample sz=10973512
- Sample enc. Q=67 : 2045 kbps, err=27.8%, size=1025740128, sample sz=10466736
- Sample enc. Q=143 : 1983 kbps, err=23.9%, size=994633360, sample sz=10149320
- Sample enc. Q=200 :
- CD size : 900
- Threshold Q : 36
- Sample file size : 1.0% (delete previous: Yes)
- Adjust estimated : 1.2%
- Sizing pass : Encode (1 pass) if opv pass < -2.5% from target, or > 0.0% above, or Q > 40
- Cond. filtering : No

[pitchshifter]

I'm having some success getting d2sroba to do one at a time if i fiddle around with the max bitrate beforehand...in the end i got the aforementioned file to work by putting the max bitrate down to 8000...it ended up generating a q of 51 and coming out fine..

So i thought it must be that i was just expecting too much for too small a filesize or something, but then the next one generated the same AutoIt error for 8000 max bitrate...but worked when i put it back to 9000?

I'm not sure whats going on, there's no logic to my experimentation here and i hope someone can figure out what the deal is because this is just bizarre :confused:

Thanks!

[tylo]

Determined Q : 4290 = Round(200 + (23.6 - 1.2)/0)
OK, I see one of the problems: division by 0 (actually near 0).

(I'm about to change this way of calculating "Determined Q" anyway).

The above leads to the error you've got. But your main problem here is that CCE is not able to fit your (mini) DVD into 900MB, even with Q=200, which gives untolerable quality in any settings. I find this strange though, because with a BR of 1600, you should easily be able to get a Q below 60, also for DVD output. I believe that max bitrate need seldom be more that 3X of average br (or even lower) - in this case about 5000. It may even be a setting in future D2SRoBa (factor of avg br, I mean)?

Tips 1: With such a short movie, use 2% sample size
Tips 2: Don't copy log directly from log windows - it gets messy if you click in it during processing. Open the D2SRoBa_log.txt in the output dir, and copy from there instead.

[pitchshifter]

Thanks for the tips, i'm sure this will fix the problem plus in general increase the quality of my rips by not wasting size available potentially allocating 9000kbps to certain parts of such a low bitrate movie....
they're approximately 45 minute movies i'm trying to squeeze into about 730 megs..i know the quality won't be great but i'd like to fit 6 per cd, and it still looks a hell of a lot better than SVCD, and actually reasonably close to the original avi source.

I wasn't sure why it was going to Q 200 either, after fiddling around with bitrates etc i'm finding the q stays around 40..this is using bicubic resize as i'd imagine lanczos wouldn't be appropriate for such low bitrates? although its certain soapies so its mainly just low motion of people talking where skin detail is more important than capturing any movement...

anyway, more just musings to myself, thanks so much again for your help!

[pitchshifter]

Well, i did what you said, and I still get this happening:
--------------------------------------------------------
D2SRoBa v3.72, by Tylo
--------------------------------------------------------
- WIN_XP - AutoIt 3.0.103.0
- 2004-10-13 13:22:24
--------------------------------------------------------
- INI file : D:\Program Files\DVD2SVCD\Tylo\D2SRoBa.ini
- Output type : DVD
- CD size : 730
- Threshold Q : 36
- Sample file size : 2.0% (delete previous: Yes)
- Adjust estimated : 1.2%
- Sizing pass : Encode (1 pass) if opv pass < -2.5% from target, or > 0.0% above, or Q > 40
- Cond. filtering : No
- Run mode : Normal
--------------------------------------------------------
- Waiting for CCE window, or 'Ripping done' message box...
- Detected a CCE encode window
- Detected the movie encode window - shutting down
--------------------------------------------------------
- 2004-10-13 13:26:48
- Copying D2SRoBa.ini to project dir C:\Nip Tuck\06\
- Avisynth setup: BicubicResize
- D2S ver. 1.2.2.3, CCE ver. 2.66.1.7
- Setting process priority 0
--------------------------------------------------------
- Movie length : 00:47:35 (68467 frames, 23.976 fps)
- D2S audio select : 192 + 0 kbps
- D2S video calc. : 1887 kbps, 1 CDs
- Select ranges : every 600, select 12 frames
- Sample frames : 1380
--------------------------------------------------------
Calculations:
- 1 CD: video_br=1887 (1887) audio_br=192+0 video_sz=673575810 fill=100% cbr=0
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Using num CDs : 1
- Target mpv br : 1887 (max 6000) kbps, size=673575810 bytes
--------------------------------------------------------
Search for Q:
- Sample enc. Q=42 : 2094 kbps, err=11%, size=747582612, sample sz=15256788
- Sample enc. Q=46 : 2065 kbps, err=9.4%, size=737163644, sample sz=15044156
- Sample enc. Q=71 : 1992 kbps, err=5.6%, size=711328296, sample sz=14516904
- Sample enc. Q=107 : 1979 kbps, err=4.9%, size=706717396, sample sz=14422804
- Sample enc. Q=200 : 1977 kbps, err=4.8%, size=705942216, sample sz=14406984
--------------------------------------------------------
- Determined Q : 3332 = Round(200 + (4.8 - 1.2)/0)

Then it crashes because it divides by 0, as you pointed out.

There is no way the Q=3332, other files of the same/size/quality normally end up having a q from 38-58...any more ideas? :scared:

[tylo]

First, v3.77 fixes this, but it will still give you a Q=200, which will oversize. The sizing pass will put it down to correct size, but the quality will be bad. If you really want to go for this low bitrate, at least enable conditional filtering, and use FACAR to "black out" the invisible overscan areas on the TV. Also, try to use a low-bitrate CCE matrix (set in DVD2SVCD). Still, I think you'd be better off creating SVCD's in this case.

/Add: to automatically enable FACAR with conditional filtering, install D2SRoBa after FACAR.