DVD Rebuilder (CCE One-Click Beta v0.46) Comments and Suggestions

[Fr4nz]

Just one question: if I select the main movie with dvd-decrypter in file movie, then will DVD-RB be able to make a new DVD from the main movie files?

[Joergen]

Originally posted by Fr4nz
Just one question: if I select the main movie with dvd-decrypter in file movie, then will DVD-RB be able to make a new DVD from the main movie files?

No, you cant do an incomplete/destroyed dvd. If you want only the main movie, use DVDShrink with reauthor mode and store with no compression, then encode that output with dvd-rb :)

[Xitrum]

Open rebuilder.ini, under [options] add this line:

TargetSectors=xxxx , where xxxx are the dimension (in sectors) of how much the DVD will contain.

If I remember correctly the default DVD-RB size is ~2240000 sectors.

The maximum size for a DVD-5 is 2297888, but you have to keep a safety margin. For example, DVD-Shrink makes DVD of 2286112 sectors.
Personally I will try 2286000 for next DVDs.

I added this parameter in the dvd-rebuilder .ini file but it did not make any differences. Did anybody successfully get this parameter to work?


alibert, you are using an ISO source? Right? dvd-rb does not yet support that. jdobbs is looking at it but is convinced it "ain't his code"

I've done about 10 movies so far from an ISO source without this "runtime error 75" errors, so I don't think using a mounted ISO image is the problem.

[unplugged]

Originally posted by RB
unplugged, I remember you telling about cases where you used ConvertToYUY2(interlaced=true) because the original stream was encoded interlaced but it messed up the colors. Can you provide a specific example? I have a hard time believing this is possible...
Some (italian) PAL titles:
Terminator 3, Matrix (1), Harry Potter (1), Random Hearts, Tomb Raider (2)
These and many others all have main movie stream interlaced, but after a simple script like this:LoadPlugin("c:\Avisynth\Plugins\MPEG2Dec3dg.DLL")
mpeg2source("c:\DVD\DVD.d2v",idct=7)
the chroma channel gets already linear, with perfect luma matching.
I have also checked more safely using ConvertToRGB() and VirtualDub for viewing. Then trying with ConvertToRGB(interlaced=true) or ConvertToYUY2(interlaced=true) with the above "affected" movies the chroma not only gets combed but most importantly misplaced (its clearly visible on red sharp edges, zoom 200-300% and look vertically!)
Originally posted by RB
If all components (original encoder, Mpeg2Dec3, AVISynth) play by the rules, then your case 3) cannot exist.
Don't know about, that could mean also I'm not sure about them (wait, that doesn't mean that I does not respect these great works!!), for ex. I don't exclude that could be a bug in mpeg2dec3.
The other case (that I've mentioned yet) may be that chroma was erroneously misplaced directly into original stream at encoding process, so it was encoded as interlaced without a prior conversion to the interlaced YV12 colorspace (funny, eh?), this could explain why in many cases interlaced=true is over the needed, the bad one.
Originally posted by RB
At the risk of repeating myself infinitely :) : an interlaced coded stream must always be upsampled using ConvertToYUY2(interlaced=true). The only bit that matters here is the progressive_frame flag in the MPEG2 picture header.

Progressive frame off == interlaced YV12 -> interlaced=true
Progressive frame on == progressive YV12 -> interlaced=false
Yes, sure and again I understand you and I was one of the first to point this important detail.
I do not disapprove this, what I'm TRYING to communicate is that there MUST be another variable to consider too, possibly for mpeg2dec itself or for us.

Looking at Bitrate View app, after loading an mpeg2 file I see "picture struture" flag on the right, over the panel. With my matrix1 clip it says "frame", where the "frame type" is reported as "interlaced". Is "picture structure" our important flag?

[quantum]

Originally posted by Xitrum
I added this parameter in the dvd-rebuilder .ini file but it did not make any differences. Did anybody successfully get this parameter to work? It works for me. I've been experimenting with changing it and it definitely affects the output size.

CCETargetSectors=2251760

This gives me 30 to 50 megs slack on most disks.

One thing that may or may not trip you up, it seems this value is stored when you save a project, so if you change it in the INI, then load a project, the project will still have the old setting. At least that's the way it looks to me.

[DMagic1]

Originally posted by quantum
It works for me. I've been experimenting with changing it and it definitely affects the output size.

CCETargetSectors=2251760

This gives me 30 to 50 megs slack on most disks.


Yep, this is what I've been using.
RejigTargetSectors=2280000
CCETargetSectors=2250000

[alibert]

@ DDogg: yes, it works fine. thx man...

[hpolimar]

i'm using cce sp 2.67 and eclecc . when i press on encode i've an error on video format. it says 1088x56 is not supported. but video is a standard 720x576

[Xitrum]

Yep, this is what I've been using.
RejigTargetSectors=2280000
CCETargetSectors=2250000

Thanks for the info, I only put in "TargetSectors=2280000" instead of "CCETargetSectors=2280000". I'll try it again tonight.

[jdobbs]

Originally posted by lrosado
I have a question for JDOBBS,

I thought I read somewhere that we no longer needed force film or interlace options with DVDRB? Force Film no longer exists. As for interlace options, it depends what you mean. Are you discusing deinterlacing? If so, that is not "needed" but some people (especially those who play back on a PC Player that doesn't do well with interlaced source) prefer to deinterlace.

[jdobbs]

Originally posted by quantum
It works for me. I've been experimenting with changing it and it definitely affects the output size.

CCETargetSectors=2251760

This gives me 30 to 50 megs slack on most disks.

One thing that may or may not trip you up, it seems this value is stored when you save a project, so if you change it in the INI, then load a project, the project will still have the old setting. At least that's the way it looks to me. It is definitely stored in a .rbd file (project)

[jdobbs]

NEW VERSION POSTED

Just posted a new version (v0.42) -- I recommend everyone download and use this version as there are a couple of very important bugs corrected. To download go to the first post of this thread. Once again, "Thanks!" to all the beta testers. Changes are listed below.


- Important update: Corrected a bug in which TFF/RFF flags could be set incorrectly on in some frames. This error would have resulted in a single frame "stutter" that might randomly occur throughout the video.

- Found and corrected a bug in which streams of pictures that were marked as interlaced in the original VTSs could be mistakenly marked as progressive upon REBUILD. A special thanks to Alex Z for helping identify these two bugs.

- Based upon some good advice from DDogg and others I have changes the default values for bias to 25 quality_prec to 16.

[Joergen]

Originally posted by jdobbs
NEW VERSION POSTED



/me hits cancel on prepare, downloads new version 0.42 ;)

[Joergen]

jdobbs: A quick question (with a complicated answer I bet):

I've never used the "dynamically assign bitrate" function yet. But thinking about how DVD-RB compresses in segments and CCE is thus unable to borrow bitrate from other segments, I guess this would be a good option to use.

Why is it not on by default? :)

[onesoul]

Just turn it on! :)

For any reason you could prefer to have the same average bitrate trhoughout the whole DVD. I am seeing in future expanded option to reduce extras average bitrate :)

@jdobbs
The lpcm information I gave was of any use? I can provide small sample vob if you like (size?).

[Joergen]

Originally posted by onesoul
Just turn it on! :)

If thats the white rabbit from monty python's holy grail then YES SIR!

I'm doing passes now with dynamic enabled and not enabled (of course process again when changing options). The only harm I can think of it doing is giving the end credits mega-bitrate (based on the original compression having the same) or giving the extras huge bitrates if they were stored CBR to speed up the production process.

[onesoul]

Originally posted by Joergen
If thats the white rabbit from monty python's holy grail then YES SIR!

I'm doing passes now with dynamic enabled and not enabled (of course process again when changing options). The only harm I can think of it doing is giving the end credits mega-bitrate (based on the original compression having the same) or giving the extras huge bitrates if they were stored CBR to speed up the production process. Yup, that's the one :)
I understand the "harm" you are refering but imo with dynamic enabled you can rely on having more comparable result (in quality), as a whole, to the original source because every cell bitrate is adjusted proportionally to original cells.

[jdobbs]

Originally posted by Joergen
jdobbs: A quick question (with a complicated answer I bet):

I've never used the "dynamically assign bitrate" function yet. But thinking about how DVD-RB compresses in segments and CCE is thus unable to borrow bitrate from other segments, I guess this would be a good option to use.

Why is it not on by default? :) That, my friend, is a damn good question. I will be the default in version 0.43. Somewhere in my notes It have written down that at some point it may be the only option.

[Joergen]

Originally posted by onesoul
Yup, that's the one :)
I understand the "harm" you are refering but imo with dynamic enabled you can rely on having more comparable result (in quality), as a whole, to the original source because every cell bitrate is adjusted proportionally to original cells.

I'm now doing the same disc with dynamic enabled and the segments are coming out significantly smaller. One segment has a walking-through-forest scene that from a bitrate perspective should be as heavy as it gets and yet that segment is 39MB compared to 51MB without dynamic enabled.

Another fear I have here is that the original mighthave been compressed in segments aswell and one segment got the lions-share of the overall bitrate.

edit: then again I'm working with a worst case scenario here with an extremely long movie with lots of filmgrain (which pays towards a steady bitrate) and about 900MB of menus, audio and extras.

And leaning into the DVD2AVI window on my TFT monitor also gives the sharpest image possible (most revealing of quality flaws).

[Joergen]

jdobbs: ooh I got another one! The source path can be the same as ultimately the rebuild path would be.

I edited a dvd, moved the VIDEO_TS to the root of D: to test it in WinDVD (that needs it like that). Then happily loaded it in dvd-rb, working path to D:, and hit transcode. With glazed-eyes watchin CCE do its thing I realised the rebuild path would be the same VIDEO_TS dir :D

(and it prolly couldnt write over the files its reading)

[Marcelw]

1028x56 error in frame size
1. DVD-R Version: 0.41
2. Encoder being used: CCE 2.66
3. Using eclCCE? Y
4. Bug encountered: Frame size was 1028x56 and is incorrect.:confused:
After prepare of Queen live at wembley, my Encode fase does not start correct. First of all it is complaining of an incorrect frame size. It says that the frame size is 1028x56 and that this is not working for DVD, which I can imagine. Where can I find the format of this size and how can I correct this? But whatever I try, the encoder will not read the item.ecl file. The error 9 on the prepare phase is disapeared now finally.;)

[YaoMing11]

What does it actually fix by changing the VBR_Bias to 16 and the Quality_prec to 25?

Does it improve picture quality or speed up the process?

[the-warriners]

Just to let you know that I have processed movies mounted from an ISO using daemon tools and it has worked fine every time :)

[P3gasus]

Originally posted by Marcelw
1028x56 error in frame size
1. DVD-R Version: 0.41
2. Encoder being used: CCE 2.66
3. Using eclCCE? Y
4. Bug encountered: Frame size was 1028x56 and is incorrect.:confused:


This is not a DVDRB bug. It is related to your avisynth installation. Be sure you installed avisynth 2.54 correctly. Try setting mpeg2dec3dg.dll path in RB setup and checking the add to avs file option. That should solve your problems (after re-preparing).
Bye

[jdobbs]

Originally posted by Joergen
jdobbs: ooh I got another one! The source path can be the same as ultimately the rebuild path would be.

I edited a dvd, moved the VIDEO_TS to the root of D: to test it in WinDVD (that needs it like that). Then happily loaded it in dvd-rb, working path to D:, and hit transcode. With glazed-eyes watchin CCE do its thing I realised the rebuild path would be the same VIDEO_TS dir :D

(and it prolly couldnt write over the files its reading) That ain't good. I'll put in a sanity check for this.

[hpolimar]

testing 0.42:

with this new version (cce sp 2.67) after "prepare" fase i got a different size error while encoding: this time instead of 1088*56 it's 760*56 (format not supported). Even if the program can't fix it, i'd like to do it manually but in the files i've i see no line with the video settings. How can i do?

follows:

.ecl file has

title=V03000000001001
aud_out=0
vaf_file=D:\TEMP6\D2VAVS\V03000000001001.vaf
aud_file=D:\TEMP6\D2VAVS\V03000000001001.mpa
file_focused=0
packet_size=2048
width=720
height=576


so it should be right settings

[lab-one]

This thread will point you in the direction you need to go... (http://forum.doom9.org/showthread.php?s=&postid=478524#post478524)

[Joergen]

jdobbs: Here's a wild and fancy suggestion :D

Since you're encoding in segments.. would it be possible to distribute the encoding of separate segments over the LAN. All I think you'd need is for the source path to be accessible to the LAN puters (easy) and a "reserve" and "completed" function for the batch file where the "client" computers would mark the section they're doing as reserved and then complete (so other cliends would know to skip to the next one).

If you'd really stretch it you could have the server computer create a "dvd-rb droplet" executable and stream the data to the clients, the clients would then only need the same version of CCE and avisynth installed but not need to setup the source path or work path.. or something. :cool:

Just think.. harnessing a few friends' computers in the dorm or computer lab: 15 minute npass CCE DVD-RB copy!

[MedicineMan]

Just think.. harnessing a few friends' computers in the dorm or computer lab: 15 minute npass CCE DVD-RB copy!

Now that's a damm good ideia. Is it possible? It might be, but it has to do with the way jdobbs codes DVD-RB, doesn't it? I guess that this suggestion is going to be in his "to-do" list. And at the speed we see improvements, it's going to be only a couple of weeks before we can do quality encodings that fast.

[kadilak]

That kind of distributed computing model would take a lot of work to implement. Especially since you'd have to have some sort of master machine delegating work to all the nodes, etc. Otherwise the process would be very manual, and probably not worth the effort.

I think the concentration should be on releasing a stable 1.0 that will do a perfect copy of any dvd and work on every dvd player.

[gopalkk]

The non movie VTS_04 is not taking by Nero and complains
about block size being smaller than 2K. I mounted source
ISO on deamon tools. It is Indian movie.

[quantum]

This is probably getting beyond the scope of the original program concept. But that would be for jdobbs to decide. With that said, I have 2 computers on my home LAN and if I could cut the encoding time in half it would interest me.

I've used server farm software for 3dsmax and it can get complex. However I can imagine a simple way for jdobbs to accomplish it. This assumes you don't use the "one click" mode.

Simply let the user select segments in the "encode" phase. Select the first half of files on PC #1 and the second half on PC #2.

All the files are in a single shared directory on the lan, like m:\movie. Remote copies of CCE will be accessing files over the LAN. With fast home networks, I don't think CCE will be bottlenecked by transfer speeds.

[Joergen]

I'm currently compressing on a lan PC off the drive of another on a 100mbit lan. No bottlenecks at all (the HD light on the serving pc hardly flashes and the encoding AMD 64 3200+ is at full steam).

quantum's way of choosing the segments manually would be the easiest way. Then just copy the completed stuff to either PC's D2VAVS dir and hit rebuild.

[quantum]

Originally posted by Joergen
quantum's way of choosing the segments manually would be the easiest way. Then just copy the completed stuff to either PC's D2VAVS dir and hit rebuild. You wouldn't need to copy anything. Assume PC #1 has the shared directory, PC #2 will operate directly on these files over the LAN. The completed files will be on PC #1. You would rebuild on PC #1.

[Joergen]

Originally posted by quantum
You wouldn't need to copy anything. Assume PC #1 has the shared directory, PC #2 will operate directly on these files over the LAN. The completed files will be on PC #1. You would rebuild on PC #1.

Oh yes read and write directly to the shared resource. Its almost what I meant earlier except I suggested the programs would reserve chunks as they chug along instead of manually selecting them. This would help optimize the time used and balancing the chunks, where chunks of various sizes and unbalanced setups on the PC's might have the faster computer completed in 1hour but the slower still chugging along after 2hours.

[quantum]

Originally posted by Joergen
Oh yes read and write directly to the shared resource. Its almost what I meant earlier except I suggested the programs would reserve chunks as they chug along instead of manually selecting them. This would help optimize the time used and balancing the chunks, where This would be better, and may even fit in with whatever technique jdobbs is doing for his suspend/resume code. I assume he must mark sections as being completed in order to resume. So each PC would store a flag indicating a chunk was "started" and pick up the next available one when finished.

Well, I guess it's all settled then. Unfortunately I'm sure it's easier said than done. Now after all this talk we'll have to wait and see if jdobbs will even consider the concept.:)

[Wimpy]

Hi,

Sorry to interupt your discussion about DVD-RB encoding farms, certainly sounds interesting :-)

I have a question, which leads into a suggestion so I think this is the right thread to post in. What is the threshold DVD-RB uses to determine that a given VTS is not suitable for recompression?

I am guessing that this suggestion is already in your mind, but I will ask anyway. It would be great if all Video Title Sets of a reasonable size were made available for recompression, and ideally, that each Video Title Sets can be assigned a diffent level of recompression.

I buy a lot of ex-rental DVDs and they tend to contain a lot of trailers and adverts for upcoming rental titles. It would be great if I could over compress those sections and the balance of the available DVD-5 bandwidth be reapportioned to the main movie, thereby increasing quality. I put my newly purchased ex-rental DVDs on DVD-RW so that I can remove the annoying PUOs because my girlfriend hates being forced to watch trailers for films she isn't insterested in, and I like a quiet life ;-)

Additionally, if DVD-RB could be ehnaced to make the compression level selection by PGC that would increase the flexibility considerably.

As I said some versions ago, thanks muchly for DVD-RB it is a really great tool especially when used inconjuction with QuEnc.

[Joergen]

If the trailers are in a separate VTS, use TitleSetBlanker to get rid of them in a flash.. works great. It can get rid of copyright warnings aswell.

The treshold is 50MB I think.

Hmm.. dvd-rb renderfarms. You could buy some old 1-1.5GHz cpu's or boxes for cheap. Strip them of everything but a small HD, some RAM and a lan adapter.. use PCAnywhere software to control them without a monitor. Mmmm ~8GHz of compression power at your grasp for less than upgrading your main CPU to the best available.

[jdobbs]

Originally posted by quantum
This would be better, and may even fit in with whatever technique jdobbs is doing for his suspend/resume code. I assume he must mark sections as being completed in order to resume. So each PC would store a flag indicating a chunk was "started" and pick up the next available one when finished.

Well, I guess it's all settled then. Unfortunately I'm sure it's easier said than done. Now after all this talk we'll have to wait and see if jdobbs will even consider the concept.:) I think I'll concentrate on getting it working with all sources and out of beta first.;)

[Joergen]

Originally posted by jdobbs
I think I'll concentrate on getting it working with all sources and out of beta first.;)
Of course, a stable core comes before all else. We are just fantasizing about what the future might hold.

Surely you could wreak the benefits yourself, not having to wait all night to see if the output is flawless.. but instead setting up old 2-3 PC's and come back an hour later ;)

[jdobbs]

Originally posted by Joergen
Of course, a stable core comes before all else. We are just fantasizing about what the future might hold.

Surely you could wreak the benefits yourself, not having to wait all night to see if the output is flawless.. but instead setting up old 2-3 PC's and come back an hour later ;) When I'm finally done, you will be able to be expect it to be flawless!:cool:

[quantum]

Originally posted by jdobbs
When I'm finally done, you will be able to be expect it to be flawless!:cool: Good. But we want it flawless, and in half the time ! ;)

[Wimpy]

Originally posted by Joergen
If the trailers are in a separate VTS, use TitleSetBlanker to get rid of them in a flash.. works great. It can get rid of copyright warnings aswell.

The treshold is 50MB I think.

Thanks for the TitleSetBlanker tip, that works great for erasing Video Title Sets I don't want :-) PGC selection for recompression levels would still be a cool addition to DVD-RB though :-)

[Joergen]

Originally posted by Wimpy
PGC selection for recompression levels would still be a cool addition to DVD-RB though :-)

Yes that's coming soon. It will enable you to force extras to as low as you want (even halving the resolution to half-D1), it will get that final edge out of CCE and make backups as good and as complete as possible.

[haggis663]

I am excited by DVD Rebuilder in the sense of the possibilities it offers beyond shrinking a DVD to fit on 4.3gb. If we look at dual layer DVD writers coming up, the current limitations will be removed in time, but the ability to add value will remain. By adding value, I mean the ability to:
Changing of aspect ratios to better fit the viewers needs (4:3LB to 16:9 (done), perhaps 4:3 to 16:9PB, etc.)
Removal of trailers / FBI warnings and other irritants
Conversion from NTSC to PAL and vice versa
Deinterlacing (done)
Merging of episodic DVDs together

Due to DVD Rebuilder doing a reencoding rather than transcoding, it seems to me that it is the only current tool that has the possibility of melding all of this together in a way that an average user can achieve this result.

[P3gasus]

Originally posted by Wimpy
Thanks for the TitleSetBlanker tip, that works great for erasing Video Title Sets I don't want :-) PGC selection for recompression levels would still be a cool addition to DVD-RB though :-)

Titleset blanker has currently some issues regarding tmapti talbes and cell times. Currently the use of titleset blanker could cause problems with some home dvd players due to those issues. I recommend you to use VobBlanker instead.

Bye

[Skinleech]

Just tried my first true test of rebuilder with 3 different disks:

Ed Wood R2
Matrix Reloaded
Usual Suspects R2

I each I used 0.42 plus CCE 2.50 and stripped out 1 language track for each - it works perfectly, with flawless picture quailty on the re-encoded disks. Brilliant tool so jdobbs. Keep up the good work.

[Joergen]

Originally posted by P3gasus
could cause problems with some home dvd players

To titlesetblankers defence, I've used it on at least 15 discs that have been played on about 6 different players (Pioneer, Panasonic, Sony, Euroline) without problems :)

Vobblanker though can blank items inside a VOB too, so its more flexible.

[smlong426]

I just finished encoding Pleasantville with RB 0.42. I noticed that there is still a small glitch at chapter points on my Cyberhome player with this movie. I don't notice that glitch on my Zenith player. Just wanted to post a comment/feedback ..

Stephen

[Wimpy]

Originally posted by P3gasus
Titleset blanker has currently some issues regarding tmapti talbes and cell times. Currently the use of titleset blanker could cause problems with some home dvd players due to those issues. I recommend you to use VobBlanker instead.

Thanks for the info. I will test it now :-)