ReJig 0.5 Released

[Abnormal1]

Hi,

I have a small request to make.

Is there a chance that a CLI for the DVD Author mode could be added as I can see this being very usefull to me and maybe others.

Thanks
Abnormal

[Makira]

Fixed a small bug (but visible), updated the archive:

http://www.info.polymtl.ca/~anmis/m2vrequant_211203.tgz

[mmgrover]

@Makira

Thanx for the update :-)

mike

[Nic]

http://nic.dnsalias.com/ReJig.zip

This basically has LPCM support (although may contain bugs, dont have enough test material) and Makira's new engine. This new engine is definitely an improvement on the "new" engine. But in one short test I did, I still think I preferred the "old" engine. Ill have to test more...Let me know what you think.

Ill add commandline DVD Authoring and try and fix E-Male's problems (as well as others) before moving onto 0.6

Cheers,
-Nic

[Rombaldi]

Suggestion for 'engine management'...

with people making may tweaks to the engine, might I suggest a change to the distribution/packaging of ReJig?? Altho it is quite handy to have everything in one small EXE, it could be easier if it was split into an EXE and two DLL's.

The main EXE of course being the main interface processing logic. The DLL's being the ENGINES.

ReJig-E0.DLL being the 'old' encoding engine
ReJij-E1.DLL being the 'new' engine.

that way if a 'new' engine fix comed out, it would just be a simple download of the new DLL and plug in the new code, without having to require a complete re-release of everything..

take this a step further.. decide 'how many engines are enough' (grin) and make part of the settings dialog be

Encoding Engine Manager

(X) E1 (description) (filename/path)
( ) E2 (description) (filename/path)
( ) E3 (description) (filename/path)
( ) E4 (description) (filename/path)
:
( ) Ex (description) (filename/path)

X=Default Engine

where (description) and (filename/path) could be specified in the settings, as well as the 'default' engine.

You could when setting up a compression job, specify which engine (drop down list?)if you didn't want the default to use to compress a file, and obviously if you stack up multiple compressions in a list, a different one for each file.

TAKE IT A STEP FARTHER, lets's say that someone makes a mod of the engine that does VERY VERY VERY good work on animated stuff but not so hot on live action. That could be a seperate DLL that is invoked when needed, someone else does a great engine that compresses very well on slow motion, but not so good on fast action, etc.. etc.. etc...

The obvious advantage of being able to QUICKLY and easily test multiple engine mods is obvious.

Think of it as a set of tools with different plug in adaptors for specific needs (dare I even think about being able to specify a given engine for each title in a backup [well, this is going to be lightly compressed, so use E3, but these extras need to be tightly compress, so use E5, etc...)

[kempfand]

@ Nic, Makira, All:

This is off-topic. I'm a silent reader in this section. My application of ReJig is 'simply' to compress my MPEG from DV-conversions, which are sometimes a tiny bit too big to fit ona DVD-R.

But I wanted to say that I really really admire your work on this.

It's fascinating to see the progress, and most importantly, what a 'community' can do to push things forward.

Nuff said. A Big Thank You, and keep up the truely exceptional work.

Happy holidays!
Andreas

[DVDRFreak]

@Nic

Tried the new 0.5d verion and it crashes right after multiplexing the streams. It says creating DVD files and then I get a windows exeption error.

Did run it on NARC. Using the DVD backup mode.

[Nic]

Grrrr. I think there's a small stack overflow somewhere, as I got a crash once too. But I ripped two films last night and there where no problems.
(However, I have found one of my DVDs does cause subtitle de-sync so I can use that to fix the problem. yay!)

@Rombaldi: That is a nice idea, and would be easy to implement, but Im trying to keep things as simple as possible for the user in this program. DLLs can make things messy as well as harder to debug. Eventually I hope we have just one engine, which will stop any confusion as which one to use.

-Nic

[unixfs]

While we still don't have a real 2-pass ReJig ;) is it possible to
instruct the engine not to recompress scenes that have a high
amount of motion (more than a user-specified threshold) ?

Thanks.

[m1482]

@Nic: Something is really wrong with new engine!!!

Quality is bad. Compressed image is very blocky, jumpy, and it has some trembling in the screen. The best engine of all is the original ReJig...

[katjarella]

Hello NIC,

Unfortunately I do not know good English, therefore is not good it much to understand. But I try it.

My ReJig05d test:

1.Original IFOTitle Set (Movie) attributes:
Video: MPEG-2 720x576 (PAL) (PAL 625/50) (16:9) (letterboxed)
Audio 1: English (Linear PCM) 2ch 48Kbps 16bps (ID: 0xa0)
Audio 2: English (Dolby AC-3) 6ch 48Kbps DRC (ID: 0x81)

PGC_1 (program chain): [Title(TTN): 1] [00:03:08.16 / 25 fps] (Programs: 1) (Cells: 1) (uses VOB-IDs: 1)
[Ch 01] [Pg 01] [Cell 01] [V/C Id: 1/ 1] : time: 00:03:08.16 / 25 fps [Pos: 00:03:08.16] [Frames: 4716]

2.ReJig05d > IFO Mode
all Streams, Demux Video (No Compress !) , Correct AC3 Delay=OFF

Result:
AC3_Audio_0xBD_0x81_Delay_0ms.ac3 ( Sound perfectly )
LPCM_Audio_0xBD_0xA0_Delay_0ms.wav ( Sound perfectly )
MPEG_Video_0xE0.m2v

3.ReJig05d > DVD Author
Video = MPEG_Video_0xE0.m2v
Audio1 = LPCM_Audio_0xBD_0xA0_Delay_0ms.wav
Audio2 = AC3_Audio_0xBD_0x81_Delay_0ms.ac3

Result and comparison to original the Ifo:Title Set (Movie) attributes:
Video: MPEG-2 720x576 (PAL) (PAL 625/50) (16:9) (pan-scan & letterboxed)
Audio 1: English (Linear PCM) 2ch 48Kbps 16bps (ID: 0xa0)
Audio 2: English (Dolby AC-3) 6ch 48Kbps DRC (ID: 0x81)

PGC_1 (program chain): [Title(TTN): 1] [00:03:08.16 / 25 fps] (Programs: 1) (Cells: 1) (uses VOB-IDs: 1)
[Ch 01] [Pg 01] [Cell 01] [V/C Id: 1/ 1] : time: 00:03:08.16 / 25 fps [Pos: 00:03:08.16] [Frames: 4716]
perfectly !!!!


---------------------------------------------------

BUT!

The sound of the LPCM is very bad and is changed. One hears a little of the original, then noise gets ahead and in such a way.
Then nich again demuxt and a CRC check made.

File - Original Demux CRC - New Demux CRC
AC3_Audio_0xBD_0x81_Delay_0ms.ac3 - EB5CBB8C - 5474DEFD
LPCM_Audio_0xBD_0xA0_Delay_0ms.wav - D8CD3A4A - 38B59E19
MPEG_Video_0xE0.m2v - 8F59F8DD - 8F59F8DD

From where do the differences come?
Are all sounds changed?

Greeting your Katjarella.

[ffroms]

Can someone post different link to new version (0.5d) of ReJig 'couse I can't access Nic's page.
Thanks

EDIT: Ignore this. I've got it.

[Makira]

@m1482

Your statement doesn't really help. I can show you numerous examples where the new engine gives improved visual quality, both psnr and subjectively. But... you can do the same the other way, I suppose.

Most of the samples I've tried give better quality with the new engine, _when the final size is good_. The old engine is known to produce too big files, ie not following the requested compression factor, and then comparison is invalid.

But I do have some samples where the new one performs worse. While the psnr is still better, the subjective quality due to temporal flickering is worse. I'm thinking of modifying mpeg2psnr so it prints temporal psnr (ie comparing the difference between frames instead of comparing the frames themselves).

Also, bear in mind that testing with files smaller than say 75 megs is not helpful either, since it does not reflect the intented use: the new bitrate controller tends to work better with longer files.

So what can we do? We have to find out _why_ some samples perform better with the old, some others with the new. Then translate this in machine code, and make the engine smarter.

For my particuliar sample where this is most noticeable (a scene from Baraka), while the psnr is better with the new engine, the image seems to flicker. It looks like the I and P frames are really good, while the B frames are not, which make the video flicker because of the alternance of good/bad looking frames. In the old engine, all frames look equally bad (really, they are spatially much worse), but temporaly, it looks a lot better, leading to an overall better subjective quality.

It is _not_ as simple as saying X is always better than Y.

Maybe we could support a 'slow' mode, where the engine could aim for a constant psnr, or maybe we shouldn't rely on the quantiser in the stream to find our own, or maybe there's a whole different way to solve the problem.

I don't know, but I'm sure we can find out. The quickest path to solving this quality problem is properly documenting for what kind of samples, and if possible why, one engine performs better than the other.

[E-Male]

i noticed that alternating good and bad frames on "M"
i think it is caused by the grainy source

[TCmullet]

Originally posted by Makira
For my particuliar sample where this is most noticeable (a scene from Baraka), while the psnr is better with the new engine, the image seems to flicker. It looks like the I and P frames are really good, while the B frames are not, which make the video flicker because of the alternance of good/bad looking frames. In the old engine, all frames look equally bad (really, they are spatially much worse), but temporaly, it looks a lot better, leading to an overall better subjective quality.Whatever you all do to change rejig, I hope you will apply all compression logic processes equally to I, P, and B frames. I want very much to be able to pause playback and have consistent quality no matter whether I'm paused on a P, B or even an I. And even apart from pausing, I would strongly prefer a result that does not rely on any psychovisual tricks. (Unless you wish to do something like reduce more in the black or nearly black areas--THAT I'd heartily approve of. Or maybe reduce more around the perimeter of the picture, if that's even possible.)

[TCmullet]

Originally posted by unixfs
While we still don't have a real 2-pass ReJig ;) is it possible to
instruct the engine not to recompress scenes that have a high
amount of motion (more than a user-specified threshold)?
I second the motion. But I suspect that it might not be appropriate as good mpeg encoders have already divvied up the bits to have more in the fast scenes. But I still like the idea--maybe it's worth a try.

[dragongodz]

Makira - i said back in the old thread about some ways that increased quality could be gained. things like a payback system where GOPS with a lower bitrate than the average could be reduced slightly more and the extra could be used for GOPS with a higher bitrate than average. thus higher motion scenes should get less reduction preserving better quality while static/low motion would get reduced more but should be able to handle that. also things like original frame size compressed to the point of actual amount required(with a frame type bias) before writing should work better aswell. that is read original frame size, reduce frame to the percentage required(do not write until it is) with the bias being that I frames are reduced less(say 10%) than the target reduction and B frames are reduced slightly more to make up the correct reduction amount per GOP. for extremly high compression there are other ways to preserve some quality that are not really desirable for low compression. these other methods though could be used on things like extras etc.
if interested then email me but i will be away until after new year so dont expect a reply until then.

Unixfs - if a movie was 50%(or more) action and not compressed at all for action scenes the rest of the movie would be compressed way to much and you would see it(unless only compressing something like a few percent total). so less compression for action would be better. thats basically what the part above about different compression based on original bitrate etc is aimed at. this asumes the original compression used higher bitrate for the action or highly detailed scenes of course.

[ffroms]

A little possible bug in DVDAuthoring.
When I run DVDAuthoring first time and everything completed OK. I've exited authoring (not ReJig) and started again authoring, set everything and clicked on Create and program just disappeared from screen. I've tried that few time and with same resolute.

FFS

[Nic]

@ffroms: Ahh, that doesn't surprise me. Mplex is coded quite badly in some respects...it also has huge memory leaks afters it's been done. Thanks for the update. I should be able to fix it without too much bother :)

@unixfs: I think the best option would be to support a proper full 2-pass rather than try and change the compression ratios on the fly depending on the motion...I dont think that would always work well.

@Makira: The new engine is definitely better in some scenes that ive tested...Thanks for your feedback and development :)

@katjarella: So, the LPCM was fine when done in IFOMode but when done with DVD Backup it sounded really bad and the files produced were not the same? Is that right? Hmmmm, Did you do IFO Mode first and then try a DVD Backup without closing the program?
(Im trying to guess what went wrong...)

Ill look into all these things...(I think I fixed the sub de-sync at last)

Cheers,
-Nic

[ffroms]

Thanks for quick replay and GREAT tool ! Maybe in next version will be fixed my problem with not showing subs on DVD player (even though they are visible in PowerDVD and WinDVD). I hope it's somehow related with sub de-sync.

FFS

[Proff]

Hello there,

i have been following Rejig since the original GPL transcoder topic and the name Rejig wasn't even born yet :)

I think Makira and Nic are both very great people and programmers by making all this possible. Tnx guys, u rule!

Anyway, the reason why i'm posting in this thread, besides thanking these guys, is that i might have something to contribute also. A couple of pages back Nic mentioned that he was using the Intel C++ Compiler 8.0 now, because of speedgains. I have always been a fan of this compiler since it is able to generate code specifically for modern processors. The code im talking about is mmx, sse and sse2 and can't be outputted by for instance the Microsoft Compiler without using intrinsics.

While checking the project settings of Nic's workspace, i noticed that the vectorizer that generates this kind of parallel isn't turn on! The compilerswitch im talking about is /Qax. By using this switch, the compiler checks if certain pieces of code can be sped up by using modern processor extension, and if it does, it generates the appropiate code for it. The 'a' in the switch means that also 'plain x86' code is generated so that the resulting executable can be used on older machines also (no benefit for these older machines ofcourse).

I found that using one addional compiler switch: /QaxKWN , Rejig performed up to 10% faster in my small testcases. I added some lines of code that uses GetTickCount() to measure the time Rejig needs to transcode these test files, and than built Rejig with original compiler settings and again with the /QaxKWN switch. I only tested transcoding mpg files via the command line, maybe there are speedgains in the DVD muxing or other parts of the program too.

Tested 2 SVCD files on my P4 2.4 Ghz 533 bus:

1.) 246 Mb (with sound), transcoded to 200 Mb (without sound).
Rejig original: 24581 ms. (average of 5 tries)
Rejig /QaxKWN: 22815 ms. (average of 5 tries)

2.) 771 Mb (with sound), transcoded to 650 Mb (without sound).
Rejig original: 83969 ms. (average of 3 tries)
Rejig /QaxKWN: 81068 ms. (average of 3 tries)

The Rejig build with mxx/sse(2) code always performed faster than the version without this code. Couldn't test any big dvd's since i don't have enough HD space, but why don't u guys report your speed gains?

I don't have any webspace to host the new executable, but i can upload it anywhere or mail it to people. Nic: will u consider turning on this compiler option on by default for the original Rejig? The only disadvantage i see is that the size of the binary went from 900 kb to 1 Mb, but i consider that to be very small still!

[Nic]

(That's very true, I used to use it when I compiled XviD...Forgot about adding that in. For the next release, ill use that switch. Thanks for the reminder :) )

[Amnon82]

I wanted to backup Tomb Raider 2 ...

TR2 has 1 DTS Audio Track.

Original is so:

AC3 Deutsch
DTS Deutsch
AC3 Deutsch
AC3 English
AC3 English

After multiplexing with ReJig 0.5d I got this:

AC3 Deutsch
AC3 Deutsch
AC3 English
AC3 English
MPG Audio ????

IFO Edit didn't mux DTS ...

@Proff: If You want, I can host Your speedy ReJig ...

[mmgrover]

@Nic, I missed 0.5d.. MyBad. I posted it.

Is the Rig_src.zip there for 0.5d?

mike

[TCmullet]

Now that 0.5d became available, I've tried to use the "new" engine again and resume xsvcd authoring. But I've gotten unexpected results. Am hoping someone can explain. (And please forgive if this has been covered but I didn't grasp it.)

(In this note, I'm going to use the 'k' figure displayed in Windows Explorer size column instead of the 'mb' figure in the bottom status line. This will show more precision.)

0.5c took a file (.m2v) that is 760,062k and with old engine and 99.7% made it 757,481k, just right for my XSVCD limit of 795MB w/audio muxed back in. When switching to 0.5d and new engine, I got undersized results. When I tried to increase my percentage to where I'd get the same size as 0.5c-old, I was never able to increase it enough. Here are my figures:

99.7 750,185k
99.8 750,922k
99.9 751,593k
99.95 751,897k
99.98 752,039k
99.99 752,086k
99.999 752,123k
100.0 752,125k

As you can see, even upping it to 100% still compresses over 7MB away! (5MB more than old-engine@99.7%.) How can this be? And why does the "new" new engine behave with less size accuracy than the "old-original"? I tried 105% just for kicks and got an error message.

One more thing-- I thought I was going to ask you all which I should use, the new engine at 100% instead of old at 99.7%, even though the new makes the result 5mb "too" small? But I'm not asking that, because EVERY result with the new engine gives terrible results no matter whether it's 99.7 or 100.0 percent. I get very conspicuous pixelation in every file, but none with the old engine in either 0.5c or 0.5d. Even low-motion looks very bad (pixelated). So it looks like 0.5d with old engine option is the only way for me to go. I'm kinda puzzled as to why anyone can find the new engine useful. Unless perhaps all the movie-type stuff most of you do is less sensitive than my interlaced championship volleyball recorded off of cable-tv.

[TCmullet]

It would help me a great deal if the same kind of message and audio sound would occur when the transcoding has finished that occurs when I enter a percent value greater than 100. That way, I could do something else in the room, and "wait for the beep" so I can continue my mad dash to get caught up on mpeg file processing, and still get other things done in the room while waiting for ReJig to finish. Of course any change like this should not be allowed to impede those who use the program in command-line fashion (if there are any folks like that).

[TCmullet]

For some reason, when ReJig is transcoding, all other tasks on my system (Win98-1st-ed) slow way down. Is there some sort of forced high-priority being given to ReJig? Imho, it should have an average priority so I can "run it in the background" so to speak. Or perhaps user-selectable like in VirtualDub or Tmpgenc. But having it grab top priority causes problems sometimes.

Having a "pause" feature would be real helpful too. Tmpgenc has that via the confirmation message to the 'abort' option. VirtualDub does not, but I wish it did.

[Nic]

http://nic.dnsalias.com/ReJig.zip

Fixes for:
Sup Creation Compatibility fix (adds P-STD buffer info as suggested by jj57 at start of this thread (BTW: submux-dvd does it wrong?!)

Sub timing should be fixed

DTS Support should work now ( oops )

-Nic

[jdobbs]

The .ZIP file is telling me it is corrupt -- and it's only 64Kb long.

[mmgrover]

I'm getting same prob with zip file :-(

mike

[gigah72]

I'm getting an error that "libguide40.dll" is missing.

g.

[robot1]

It's a math lib used by progs compiled by intel compiler
Probably a google search will solve...

[Nic]

Ok, all uploaded and fixed
(Jeeze, my kingdom for a floppy drive or CD-R drive that works...grrr)
(The current version doesn't need the libguide40.dll, so re-download)

[jj59]

@Nic
In which way is submux-dvd doing it wrong?

[Nic]

@jj:It appears to set the wrong values...I could be wrong however, I havent actually tested it.
But:

header[9] = 0x01; // P-STD buffer scale = 1024 bytes (fixed for DVD see spec)


should read = 0x60 in my opinion. Also look at the function mkpesh1, which does set it to 0x60 (correctly). So it depends which parts of the code are being used. But as far as I can see mkpesh1 is only used if cvd or svcd....(From the code)

I could be wrong, but that's my current opinion :)

-Nic

[Proff]

Hello everybody, hope everyone had a nice Christmas and everything?

Nic has added the /Qax compiler switch in the 0.5e release, so there is no need for hosting my version of the executable: just download 0.5e!

Nic, i have another question about the other added switches. The /Qpar_threshold that is set to 100. My documentation of this switch says this:

Sets a threshold for the auto-parallelization of loops based on the probability of profitable execution of the loop in parallel, n=0 to 100. n=0 implies "always." Default: /Qpar_threshold75.

By setting it to 100, doesn't mean that is it always disabled?

[Nic]

Ahh, thats the one bit of the sourcecode I didn't update...having the /Qparallel switch in caused it to need libguide40.dll. So I took that parameter out...
(100 might mean disabled, but I took it to mean it wont only do auto-parallelization if it would be definitely beneficial...either way, doesnt matter now :) )

-Nic

[shoarthing]

Nic - Hi - very impressed by & grateful for your work: might I suggest [re: main-title backup] that you offer an option to set plural drives [ie: temp & target] for the enormous amount of data-shuffling to-and-fro?

- tho' my setup here is all-SCSI, there is a faint impression of the application's speed in this mode being influenced by disk I/O

[TCmullet]

Anyone got any ideas concerns the topics described in my 3 posts earlier? (2 day ago) Not to be impatient, but I kinda expected at least some moderate chatter about all 3. I want to do my part to contribute to the effort by giving carefully documented feedback and suggestions. I hope Nic and/or the other experts will have some things to say about each of the 3 items.

@Nic, Thank you again very much for what is turning out to already be a very helpful tool for me and the rest.

[E-Male]

we're between christmas and new-year
i think we can keep it a lil slow this days :)

[TCmullet]

Guess you're right. I forget that. I don't do much of the xmas thing. This time is the busiest for me in digital video... Plus seeing 11 posts after mine not relating to mine made me think folks were still active and thought something I said was offensive. Will wait a while longer. :)

[Nic]

TC:
You im very limited on time, hence all my posts have been short and kurt recently...Also coding code for 0.6.

1) An audio ping. Yup I can stick in a messagebeep, ill make it a setting though that you need to turn on and will be stored in the registry

2) Slow down, that is probably happening because of disk access more than anything else, so thread priority might not help. But I will try and add something to control that. Also, if you can, upgrade to win2000. It is far more suited to heavy duty processing. Also as a stop gap...look up pstools and find pssuspend. That can pause processes until I can find a way to add it to ReJig.

3) The new engine not being accurate at such ranges will probably be/is looking into by Makira or Dragongodz...I want to get down to the engine last after Ive got 0.6 out. Although I use DVDBackup of ReJig and think it's good, it needs to be made much faster and use less diskspace. 0.6 will solve that (I hope!)

Hope thats enough,
-Nic

[jamesp]

Hi all. I'm using rejig to shrink Mpeg-2 captures and its working fine, but is there a way to mux the m2v back with the mpa after the shrink has happened? Currently i have to mux the streams back together with BBmpeg before authoring with Tmpeg DVD Author. It would be nice if i didn't have to use BBMpeg.

Cheers,

Jim

[TCmullet]

Originally posted by Nic
3) The new engine not being accurate at such ranges will probably be/is looking into by Makira or Dragongodz...I want to get down to the engine last after Ive got 0.6 out. Although I use DVDBackup of ReJig and think it's good, it needs to be made much faster and use less diskspace. 0.6 will solve that (I hope!)
Hope thats enough,
-Nic
That's a great start, Nic. Thanks! However, you may have missed possibly the most serious part of the post, "...EVERY result with the new engine gives terrible results no matter whether it's 99.7 or 100.0 percent. I get very conspicuous pixelation in every file, but none with the old engine in either 0.5c or 0.5d. Even low-motion looks very bad (pixelated)... I'm kinda puzzled as to why anyone can find the new engine useful. Unless perhaps all the movie-type stuff most of you do is less sensitive than my interlaced championship volleyball recorded off of cable-tv."

Comments?

[E-Male]

my comment (that i already stated twice before):
the transcoding has very different quality depending on the source

so if you only tried it with your interlaced volleyball and nothing else your test is useless on it's own

[IvIark]

Currently i have to mux the streams back together with BBmpeg before authoring with Tmpeg DVD Author.

You don't have to do this. Just choose to add the M2V video stream instead of the multiplexed MPG and you can select the appropriate audio stream to go with it.

@Nic: Thanks for this fantastic tool, it's saving me hours with my DVB captures. One thing that I think would be handy though is to be able to choose the display type that is specified in the IFO. 4:3 films are sometimes captured at 528x576 pixels which when displayed as 16:9 pan scan show up as a normal 4:3. At present you have to edit the IFO so that the DVD will use pan scan as default. Is there any chance of adding a selection box to choose this before authoring?

Thanks again
Mark

[jamesp]

Yes i know, but the audio always has some somrt of delay on it (usually -79ms), and there is no way to do this with TMpeg DVD author.

Jim

Originally posted by IvIark
You don't have to do this. Just choose to add the M2V video stream instead of the multiplexed MPG and you can select the appropriate audio stream to go with it.

Thanks again
Mark

[IvIark]

the audio always has some somrt of delay on it (usually -79ms)

Rejig can correct that delay though, or a tool like Ac3 Delay Corrector which can handle MPA. It'd take less time to correct the audio than to remux.

Mark

[DMagic1]

I have to get around to tring the lastest version. Thx to everyone putting time and energy into this project. Just wanted to say it.

[katjarella]

Originally posted by Nic
@katjarella: So, the LPCM was fine when done in IFOMode but when done with DVD Backup it sounded really bad and the files produced were not the same? Is that right? Hmmmm, Did you do IFO Mode first and then try a DVD Backup without closing the program?
(Im trying to guess what went wrong...)
Hello NIC.
The again remuxte PCM sound is bad. And the new files are not the same. I always again started ReJig.

·· Sonic Foundry Soft Encode Test ·······························································

File: DEMUX_ORIGINAL_REJIG_E > AC3_Audio_0xBD_0x81_Delay_0ms.ac3
File size: 10.560.256 bytes
Total frames: 5.892
Start time: 00:00:0,00 *
End time: 00:03:8,54
Status: No errors were found

File: MUX_REJIG_E > DE-MUX_REJIG_E > AC3_Audio_0xBD_0x81_Delay_0ms.ac3
File size: 10.558.464 bytes
Total frames: 5.891
Start time: 00:00:0,00 *
End time: 00:03:8,51
Status: No errors were found

File: MUX_IFOEDIT > DE-MUX_REJIG_E > AC3_Audio_0xBD_0x81_Delay_20ms.ac3
File size: 10.560.256 bytes
Total frames: 5.892
Start time: 00:00:0,00 *
End time: 00:03:8,54
Status: No errors were found

File: MUX_MAESTRO > DE-MUX_REJIG_E > AC3_Audio_0xBD_0x81_Delay_0ms.ac3
File size: 10.560.256 bytes
Total frames: 5.892
Start time: 00:00:0,00 *
End time: 00:03:8,54
Status: No errors were found


Was noticeable to me, ReJig shortens the AC3 sound.

·· CRC Check ·······························································

Original > Demux ReJig v0.5E > Mux ReJig v0.5E > Demux ReJig v0.5E
MPEG_Video_0xE0.m2v ·························· 8F59F8DD ·· 8F59F8DD
LPCM_Audio_0xBD_0xA0_Delay_0ms.wav ·· D8CD3A4A ·· 38B59E19 ·· Play : PCM Sound Bad !!!
AC3_Audio_0xBD_0x81_Delay_0ms.ac3 ····· EB5CBB8C ·· 5474DEFD

Original > Demux ReJig v0.5E > Mux IfoEdit > Demux ReJig v0.5E
MPEG_Video_0xE0.m2v ·························· 8F59F8DD ·· 8F59F8DD
LPCM_Audio_0xBD_0xA0_Delay_20ms.wav · D8CD3A4A ·· D8CD3A4A ·· Play : PCM Sound Ok !!!
AC3_Audio_0xBD_0x81_Delay_20ms.ac3 ···· EB5CBB8C ·· EB5CBB8C

Original > Demux ReJig v0.5E > Mux Maestro > Demux ReJig v0.5E
MPEG_Video_0xE0.m2v ·························· 8F59F8DD ·· 8F59F8DD
LPCM_Audio_0xBD_0xA0_Delay_0ms.wav ·· D8CD3A4A ·· D8CD3A4A ·· Play : PCM Sound Ok !!!
AC3_Audio_0xBD_0x81_Delay_0ms.ac3 ···· EB5CBB8C ·· EB5CBB8C

·· Demux PCM from Original - HEX ·····································

Decrypter : 5249464621A9210257415645666D7420100000000100020080BB000000EE02000400100064617461FDA821*
REJIG_E: ··· 5249464664A1210257415645666D7420100000000100020080BB000000EE0200040010006461746140A121*
VOBEDIT: ·· 524946465CA1210257415645666D7420100000000100020080BB000070170000040010006461746140A121*

* The remainder is identical.

···················································································
The Mutliplexen functions only correctly with Maestro. Everything else is wrong. Now one can also understand, why so many multiplex problems to have. I will times test the video quality later? is you determines more importantly :)

I hope you understand a little.
Greeting Katjarella