D2SRoBa Plugin

[r6d2]

Originally posted by tylo
I'm close to implement the "Newton-Raphson" Q-search, so I may throw some questions at you while doing that, r6d2

Please be my guest. I'm on the "Babylonians to go" class.

[Crackhead]

i just tried the new version and it worked very well to the point it stopped....as you said

i just wondered if there is a way to cancel D2S's audio extraction and conversion if you got that audio stream already on your hard disk!
I mostly do 2-3 different encodes of 1 movie to see if i can get a better quality cause i only do 1cd encodes at the moment, ok it takes only about 45-60mins but the wasted time get bigger with every encode and..you know time is money!

Greetz, Crackhead

[telemike]

Latest success story with DVD2SVCD 1.1.3B2 and RC2 of Tylo

========================================================
D2SRoBa v2.7.0 RC2 - DVD2SVCD plugin, by tylo
OS Version: WIN_XP
--------------------------------------------------------
- DVD2SVCD Shutdown
- 2003-10-25 06:17:48
--------------------------------------------------------
Settings:
- Encoding mode : OPV RoBa (Recover=No)
- CD size : 800
- Number of CD's : auto
- Auto Q. (worst) : 40
- Maximum Q. : 64
- Sample percentage : 1.0
- Adjust Q. : 1.2
- Sample GOPs : 1
- Sizing pass : on=0 (-4.0 .. 1.0)%
- Audio after video : on=0 (Fixed bitrate, 10%, min 128 kbps)
- Adjust max video br : on=1 (total 2500 kbps)
--------------------------------------------------------
- Movie length : 02:18:17 (198939 frames, 23.976 fps)
- D2S audio settings : (160 kbps + 0) kbps
- D2S video estimate : 2193 kbps, 3 CDs
- SelectRangeEvery : every 1500, select 15 frames
- Num. sample frames : 1995
--------------------------------------------------------
Calculations:
- 1 CD: Video BR=614 Audio 1 BR=160 Video SZ=636827171.
- 2 CD: Video BR=1404 Audio 1 BR=160 Video SZ=1456197636.
- 3 CD: Video BR=2193 Audio 1 BR=160 Video SZ=2274530922.
- 4 CD: Video BR=2983 Audio 1 BR=160 Video SZ=3093901386.
--------------------------------------------------------
- Using Max bitrate=2340 for auto samples (adj. for 2 CDs)
- Estimated mpv size : 1028843104. (Q=40, 991 kbps, 10317444 sample sz)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Number of CDs : 2
- Target mpv size : 1456197636. (1404 / max 2340) kbps
--------------------------------------------------------
- Binary search for Q:
- Estimated mpv size : 1561395328. (Q=20, 1505 kbps, 15657984 sample sz)
- Computed mpv size : 1295119216. (Q=30, 1248 kbps, 12987714.=(Q40+Q20)/2)
- Estimated mpv size : 1363136809. (Q=25, 1314 kbps, 13669808 sample sz)
- Estimated mpv size : 1479109371. (Q=22, 1426 kbps, 14832804 sample sz)
- Computed mpv size : 1421123090. (Q=23, 1370 kbps, 14251306.=(Q25+Q22)/2)
--------------------------------------------------------
- Q Weight : 1.0 - (1456197636. / 99.7187969924812 - 14251306.) / (14832804 - 14251306.)
- Adjust Q : (23 + (0.395123372370019) - (1.2)) = 22.19512337237
- Determined Q. : 22
--------------------------------------------------------
- Encoding Movie
- 2003-10-25 06:26:34
- Actual mpv file size:
10/25/2003 08:50 AM 1396953432 Encoded_Video_CCE_NTSC.mpv
- OPV pass result: 95.9% on target ( 1396953432 / 1456197636.)
--------------------------------------------------------
- Recover from pulldown/muxing
- 2003-10-25 08:50:55

[tylo]

Interesting results! 2:18 long movie with Q=22 on 2CDs is impressive. Which filters do you use?
Note, although you "only" got 95.9% on target, you're only one Q from the optimal Q=21 in this case (the Q-steps here is about 40MB), i.e Q=20 would have oversized.
PS. I also think the 1.1.3B2 release was particulary good. (D2SRoBa still works with it, as you just documented)

[telemike]

I only used Bilinearresize. No Filters. It was The Matrix: Reloaded btw.

I left the adjust Q at 1.2, should I adjust it to a different number?

Can't wait for the new version.

[tylo]

No, I have used a higher Q adjust, but it sometimes oversizes then. IMO, 1.2 is the optimal value, at least if you don't want oversizing.

A few results with my new DVD burner:
To get started, I tried with one of the one-click tools (DVD2One).
- Burned on a DVD+RW media with Nero (both 5.5.10.54 and 6.0.0.19) 4x speed --> Failed at end of burn (I've heard 5.5.10.45 is good).
- Burned with Stomp RecordNow Max --> Worked.
- Visual result --> Mediocre, much artifacts (as expected)
Most of my SVCDs looks better (on my 32" ws TV). Maybe DVDShrink is better, but it uses basically the same transcoding method.

For DVD backups, I'll still go with SVCD (D2SRoBa, FACAR). Later I'll try DVD output. For my DVs, I'll make DVDs.

[telemike]

Tylo-

I just did "the phantom menace" on 2 CD's at a Q of 23. I got 97.7%!

I like using DVDShrink to rip the extras to my harddrive so I can make a CD of extra material. I like the preview function in dvdshrink.

[r6d2]

Originally posted by telemike
I just did "the phantom menace" on 2 CD's at a Q of 23. I got 97.7%!

I just did LOTR:TTT, 2 CDs, Q=34, 100%!!!

@Tylo,

You may be interested in taking a look at this thread.

Some guys are thinking of implementing "RoBa" in MPEG-4

I have just started to play with MPEG-4 and I am really, really surprised of the good quality.

[syzygytec]

My first results with RC2 on 1.1.3b2 were good with my limited understanding. 2 full discs. My best results yet with d2sroba. So my log submission.

========================================================
D2SRoBa v2.7.0 RC2 - DVD2SVCD plugin, by tylo
OS Version: WIN_XP
--------------------------------------------------------
- DVD2SVCD Shutdown
- 2003-11-01 01:05:51
--------------------------------------------------------
Settings:
- Encoding mode : OPV RoBa (Recover=No)
- CD size : 800
- Number of CD's : auto
- Auto Q. (worst) : 40
- Maximum Q. : 64
- Sample percentage : 5.0
- Adjust Q. : 1.2
- Sample GOPs : 1
- Sizing pass : on=1 (-2.0 .. 1.0)%
- Audio after video : on=0 (% of Video, 10%, min 128 kbps)
- Adjust max video br : on=1 (total 2754 kbps)
--------------------------------------------------------
- Movie length : 02:06:51 (182501 frames, 23.976 fps)
- D2S audio settings : (192 kbps + 0) kbps
- D2S video estimate : 1517 kbps, 2 CDs
- SelectRangeEvery : every 300, select 15 frames
- Num. sample frames : 9135
--------------------------------------------------------
Calculations:
- 1 CD: Video BR=655 Audio 1 BR=192 Video SZ=623217776.
- 2 CD: Video BR=1517 Audio 1 BR=192 Video SZ=1443391397.
- 3 CD: Video BR=2380 Audio 1 BR=192 Video SZ=2264516496.
- 4 CD: Video BR=3242 Audio 1 BR=192 Video SZ=3084690118.
--------------------------------------------------------
- Using Max bitrate=2562 for auto samples (adj. for 2 CDs)
- Estimated mpv size : 1173180758. (Q=40, 1233 kbps, 58723000 sample sz)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Number of CDs : 2
- Target mpv size : 1443391397. (1517 / max 2562) kbps
--------------------------------------------------------
- Binary search for Q:
- Estimated mpv size : 1742389051. (Q=20, 1831 kbps, 87214448 sample sz)
- Estimated mpv size : 1397462201. (Q=30, 1468 kbps, 69949300 sample sz)
- Computed mpv size : 1569925626. (Q=25, 1649 kbps, 78581874.=(Q30+Q20)/2)
- Computed mpv size : 1483693913. (Q=27, 1559 kbps, 74265587.=(Q30+Q25)/2)
- Estimated mpv size : 1454801837. (Q=28, 1528 kbps, 72819408 sample sz)
- Computed mpv size : 1426132019. (Q=29, 1498 kbps, 71384354.=(Q30+Q28)/2)
--------------------------------------------------------
- Q Weight : 1.0 - (1443391397. / 19.9782156540777 - 71384354.) / (72819408 - 71384354.)
- Adjust Q : (29 + (0.397994842937199) - (1.2)) = 28.1979948429372
- Determined Q. : 28
--------------------------------------------------------
- Encoding Movie
- 2003-11-01 01:32:45
- Actual mpv file size:
11/01/2003 03:16 AM 1400137164 Encoded_Video_CCE_NTSC.mpv
- OPV pass result: 97.% on target ( 1400137164 / 1443391397.)
--------------------------------------------------------
- Executing VBR Sizing Pass - 1517 kbps.
- Actual mpv file size:
11/01/2003 05:00 AM 1443249492 Encoded_Video_CCE_NTSC.mpv
- Sizing pass result: 100.% on target ( 1443249492 / 1443391397.)
--------------------------------------------------------
- Recover from pulldown/muxing
- 2003-11-01 05:00:43

[r6d2]

Hi, folks,

I think you may be interested in taking a look at this thread.

@DDogg, no jokes about babylonian stuff again please

[tylo]

Almost there...
I have put my sole into making D2SRoBa 3.0 worth the while.
It's working great - only one minor item left, then final testing. Release this week.

Highlights
----------
- Completely rewritten using AutoIt3
- No more annoying splash screen - a nice log window instead
- Less bugs
- Faster
- No more use of the clipboard

New functionality
-----------------
- Uses Newton-Raphson for Q search - only 3 sample tests in total usually needed! (thanks r6d2)
- Corrected much of the misbehaviors in the 2.x versions:
e.g. it handles when target br > max avg bitrate. Also, both max avg and max bitrates are adjusted.
- I may get this in: A "silent" mode, where D2S (after restore), and CCE (after termin.) are running minimized!

Removed functionality
---------------------
- Sorry, currently it doesn't work with CCE 2.50, probably because of a bug in AutoIt3 (very strange one - it can't detect the encoding window).
- The "Credits tweaking" is gone - but was really an unrelated feature

I would like to create a fresh HTML page for the new release, so if anyone could create/suggest a pretty layout, it would be great.

[Siku]

@tylo

That sounds great! Looking forward to the final release!

Regards,
Siku

[telemike]

- Sorry, currently it doesn't work with CCE 2.50, probably because of a bug in AutoIt3 (very strange one - it can't detect the encoding window).

Bummer, I still have 2.67 so I guess I'll switch back.

Can't wait to try it.

[tylo]

- It works with CCE 2.66 and 2.67, just not 2.50.
- I'd say Friday for release, but you never know...

[r6d2]

@Tylo, just curious... Did you get my PM?

[sPeziFisH]

Whow Tylo, your tool is just great. I changed from 4-pass to D2SRoba, at least about 10 encodes were made since that decision. ...and they all are damn fine.
The last months I made the experience that encodes with bitrates down to 1400 are fine enough (in the past I refused everything under 1900).
In comparison to telemikes 'Matrix Reloaded' here's my encode (ok, it's not the same: I got PAL, set audio to 192kbps (IMO (now) I should had set it to 160 but why encoding it again) and I used bicubic precize (like it a bit more sharper, even it enforces a higher Q-value)).
Don't know why the NTSC-version is a bit longer than its Pal-crony. maybe these additional 6min results in a 30bit higher bitrate (plz don't waste time calculating also when we all know it's done quite fast).
Also I forced to make a 2nd pass (sizing pass).

Whoo, and we all are waiting for friday or the start of the next week
*doing the 'D2SRoba 3'-move*

========================================================
D2SRoBa v2.7.0 RC2 - DVD2SVCD plugin, by tylo
OS Version: WIN_XP
--------------------------------------------------------
- DVD2SVCD Shutdown
- 2003-11-02 21:59:14
--------------------------------------------------------
Settings:
- Encoding mode : OPV RoBa (Recover=No)
- CD size : 800
- Number of CD's : auto
- Auto Q. (worst) : 30
- Maximum Q. : 40
- Sample percentage : 1.0
- Adjust Q. : 1.2
- Sample GOPs : 1
- Sizing pass : on=1 (-2.0 .. 0.0)%
- Audio after video : on=0 (% of Video, 10%, min 128 kbps)
- Adjust max video br : on=1 (total 2754 kbps)
--------------------------------------------------------
- Movie length : 02:12:37 (198945 frames, 25 fps)
- D2S audio settings : (192 kbps + 0) kbps
- D2S video estimate : 1434 kbps, 2 CDs
- SelectRangeEvery : every 1200, select 12 frames
- Num. sample frames : 1992
--------------------------------------------------------
Calculations:
- 1 CD: Video BR=613 Audio 1 BR=192 Video SZ=609766425.
- 2 CD: Video BR=1434 Audio 1 BR=192 Video SZ=1426435650.
- 3 CD: Video BR=2255 Audio 1 BR=192 Video SZ=2243104875.
- 4 CD: Video BR=3076 Audio 1 BR=192 Video SZ=3059774100.
--------------------------------------------------------
- Using Max bitrate=2562 for auto samples (adj. for 3 CDs)
- Estimated mpv size : 1362317035. (Q=30, 1369 kbps, 13640632 sample sz)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Number of CDs : 2
- Target mpv size : 1426435650. (1434 / max 2562) kbps
--------------------------------------------------------
- Binary search for Q:
- Estimated mpv size : 1873295283. (Q=15, 1883 kbps, 18756964 sample sz)
- Computed mpv size : 1617806159. (Q=22, 1626 kbps, 16198798.=(Q30+Q15)/2)
- Computed mpv size : 1490061597. (Q=26, 1497 kbps, 14919715.=(Q30+Q22)/2)
- Estimated mpv size : 1412800727. (Q=28, 1420 kbps, 14146116 sample sz)
- Computed mpv size : 1451431212. (Q=27, 1459 kbps, 14532916.=(Q28+Q26)/2)
--------------------------------------------------------
- Q Weight : 1.0 - (1426435650. / 99.8719879518072 - 14146116) / (14532916. - 14146116)
- Adjust Q : (28 + (0.647042399172689) - (1.2)) = 27.4470423991727
- Determined Q. : 27
--------------------------------------------------------
- Encoding Movie
- 2003-11-02 22:03:28
- Actual mpv file size:
02.11.2003 23:27 1367113620 Encoded_Video_CCE_PAL.mpv
- OPV pass result: 95.8% on target ( 1367113620 / 1426435650.)
--------------------------------------------------------
- Executing VBR Sizing Pass - 1434 kbps.
- Actual mpv file size:
03.11.2003 00:51 1425005536 Encoded_Video_CCE_PAL.mpv
- Sizing pass result: 99.9% on target ( 1425005536 / 1426435650.)
--------------------------------------------------------
- Recover from pulldown/muxing
- 2003-11-03 00:51:23

[r6d2]

Originally posted by sPeziFisH
Don't know why the NTSC-version is a bit longer than its Pal-crony. maybe these additional 6min results in a 30bit higher bitrate.

I think this has to do with the very nature of PAL, NTSC and Films being shot at 24-fps. In fact, your PAL encode should also have higher audio pitch. Check this if interested.

[DDogg]

tylo, yes, I certainly agree you should start a new thread with the next version. This one is getting mighty long and has a lot of content that is now out of date.

New Thread Started

[syzygytec]

Yes I know v3 is out in it's own new topic, I'll leave the testing to the pro's I think, I'd be glad to test it, but for now on this slow P3 1.2ghz I'll stay with this version as I am having great results, so much so I wanted to post my latest log. T e r m i n a t o r 3.
Total time from Rip & Convert until completion: 11/12/2003 1:19:40 AM - 11/12/2003 6:22:06 AM

D2S 1.2.1b3 D2SRoBa 2.70rc2,
FACAR 1.06.6 (Lancos Resize)
CCE 2.67
Avisynth 2.5.2
no other filters,
standard 2.6x matrice
WinXP Pro SP1 Pentium3 1.2ghz 512RAM FAT32 pagefile off

========================================================
D2SRoBa v2.7.0 RC2 - DVD2SVCD plugin, by tylo
OS Version: WIN_XP
--------------------------------------------------------
- DVD2SVCD Shutdown
- 2003-11-12 02:00:33
--------------------------------------------------------
Settings:
- Encoding mode : OPV RoBa (Recover=No)
- CD size : 800
- Number of CD's : auto
- Auto Q. (worst) : 35
- Maximum Q. : 64
- Sample percentage : 2.0
- Adjust Q. : 1.2
- Sample GOPs : 1
- Sizing pass : on=1 (-2.0 .. 1.0)%
- Audio after video : on=1 (% of Video, 10%, min 128 kbps)
- Adjust max video br : on=1 (total 2754 kbps)
--------------------------------------------------------
- Movie length : 01:49:00 (156808 frames, 23.976 fps)
- D2S audio settings : (128 kbps + 0) kbps
- D2S video estimate : 1860 kbps, 2 CDs
- SelectRangeEvery : every 600, select 12 frames
- Num. sample frames : 3144
--------------------------------------------------------
Calculations:
- 1 CD: Video BR=858 Audio 1 BR=128 Video SZ=701437188.
- 2 CD: Video BR=1828 Audio 1 BR=160 Video SZ=1494437271.
- 3 CD: Video BR=2734 Audio 1 BR=256 Video SZ=2235115700.
- 4 CD: Video BR=3672 Audio 1 BR=320 Video SZ=3001954955.
--------------------------------------------------------
- Using Max bitrate=2594 for auto samples (adj. for 2 CDs)
- Estimated mpv size : 968584463. (Q=35, 1184 kbps, 19420116 sample sz)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Number of CDs : 2
- Target mpv size : 1494437271. (1828 / max 2594) kbps
--------------------------------------------------------
- Binary search for Q:
- Estimated mpv size : 1431290357. (Q=17, 1750 kbps, 28697368 sample sz)
- Estimated mpv size : 1777644117. (Q=8, 2174 kbps, 35641760 sample sz)
- Computed mpv size : 1604467237. (Q=12, 1962 kbps, 32169564.=(Q17+Q8)/2)
- Estimated mpv size : 1541051768. (Q=14, 1885 kbps, 30898084 sample sz)
- Estimated mpv size : 1503782935. (Q=15, 1839 kbps, 30150844 sample sz)
- Computed mpv size : 1467536646. (Q=16, 1795 kbps, 29424106.=(Q17+Q15)/2)
--------------------------------------------------------
- Q Weight : 1.0 - (1494437271. / 49.8753180661578 - 29424106.) / (30150844 - 29424106.)
- Adjust Q : (16 + (0.257837802062784) - (1.2)) = 15.0578378020628
- Determined Q. : 15
--------------------------------------------------------
- Encoding Movie
- 2003-11-12 02:14:42
- Actual mpv file size:
11/12/2003 03:54 AM 1442881340 Encoded_Video_CCE_NTSC.mpv
- OPV pass result: 96.6% on target ( 1442881340 / 1494437271.)
--------------------------------------------------------
- Executing VBR Sizing Pass - 1828 kbps.
- Actual mpv file size:
11/12/2003 05:34 AM 1493146520 Encoded_Video_CCE_NTSC.mpv
- Sizing pass result: 99.9% on target ( 1493146520 / 1494437271.)
--------------------------------------------------------
- Executing: C:\DVD2SVCD\BeSweet\BeSweet.exe -core( -input "C:\DVD2SVCD\Movie\Extracted_audio_1.ac3" -output "C:\DVD2SVCD\Movie\Encoded_audio_1.mp2" -logfile "C:\DVD2SVCD\Movie\Encoded_audio_1.log" ) -azid( -L -3db -C 0db -S 0db -c normal -s surround -g max ) -shibatch( --rate 44100 ) -2lame( -e -b 160 -m s )
--------------------------------------------------------
- Recover from pulldown/muxing
- 2003-11-12 06:06:11

cd sizes
1) 824,850 Playing Time: 56:44
2) 824,280 Playing Time: 52:16

[syzygytec]

oops wrong thread