CCE One-Click Software Beta

[StifflerStealth]

@jdobbs: I hope you don't mind me posting this. If you do, let me know and I'll erase the contents.

Originally posted by jptheripper
@stifflerstealth

thanx for all your insight. Could you please tell me the method you used to calculate the bitrate for the main movie after assigning a bitrate of 1850 to the extras in the .ecl? this is exactly what i want to do.


This is what I do. It takes a lot of HD space, though. I've been using DoItFast4U (DIF4U) for a long while now, so I am most comfortable with that program, and quite frankly it has really good bitrate calculations. DVD-RB is the only other program I've used for CCE that has bitrate calculation as good as DIF4U, so my hats off to both Eyes and Jdobbs.

With that out of the way. Install DIF4U, unless you have it already, and configure it (Offical help fourm (http://forum.doom9.org/forumdisplay.php?s=&forumid=66)). There are more in depth guides on how to configure it, so I will give the basics of what I have. In the Mandatory section I have 1850 for Bitrate for extras, and 100 for Min. MB to Demux.

The view for DVD structure will show all the VTS's if there are any that are unchecked, it is because they are 100 MB or less.

Next, highlight the first checked VTS (the line by the plus or minus not the individual PGC) and make sure that "Create folder for this VTS" (under the working directory on the far left) is checked and also that CellID is checked. You need to do that for each VTS that is checked. Don't uncheck any audio or sub file because you need those to make the calculation accurate.

Take note at which VTS's are unchecked. Then hit the "Do It!" button.

Like I said, check the guides for better configuration.

When the files are demuxed, there will be a CCEdata.txt file, and a REJIGdata.txt file in the Working directory. I always use C:\DVDRecode" as the working directory.

Open the CCEdata.txt file, you will see a bunch of lines that look like: "Some\Path\file.avs",number1, number2, number3,number4,number5.

Num1 is the min vbr rate, num2 is the max vbr rate, and num3 is the average vbr rate.

The name of the avs file is different. It like: VTS__01_V001C001.P16~9_1.nopull.AVS. or simimilar, but you get the idea. It should be easy enough to translate these names to DVD-RB names in your head. Make sure you look at this list carefully, because if a cell that is really short will not be included in it because CCE cannot handle videos that are really small, so you might want to print this file.

Now creat you individual ecl files for the cells listed in the CCEdata.txt file. Use Num3 (avg bitrate) in the place of the one listed. Also, change your VBR pass number to what you want. Load them into CCE under the file menu or in eclCCE.

or you can load up the avs file in BatchCCE and make the corrections, but only load the avs sripts for cells listed in CCEdata.

EDIT: To clear things up, I am talking about the avs and ecl files that were created by DVD-RB. You need to look at the list of cells in CCEdata.txt, and figure out what DVD-RB cell it's refering to.

Remember how I told you to take note of the VTS's that were not checked? Well, back in DIF4U, check all the VTS that were uncheck, and uncheck all the VTS that were checked. Then select the same options as before for all the checked VTS's. Here you can uncheck the audio and subs to make it go faster. the press the "Do It!" button. You don't need to use DIF4U on this step if you want to configure DVD Decypter to extrack the video files by cellid by hand.

Now move all the cells that were not encode by CCE to the DVD-RB working Dir: the cells from the last step, and the ones that have the -nocce flag on them from the first time you demuxed. Next, rename all the files to the name DVD-RB expects. Then, Step three.

I tried to make this breif. The are a number of grammer errors, I know. This guide will be useless when jdobbs implements this functionality, when ever that is.

Stiff

[jptheripper]

thanxfor taking the time to explain all that

I assume if all titlesets are encoded then this is a much easier task as there is no need to copy files over, we are just changing the editing settings in the ecl file.

Any insight on the .inf file?

[SteveV]

Last night i did a first try with 0.16 and i have to say it looks amazing.I did a rip of Piano, i took out all audio except english 5.1 coz if i left in DTS bitrate would be to low.
After a few hours decoding and rebuilding its perfect except for the fact that the bitrate could be even higher if i had the option to take out some trailers/extra`s.
I want to congratulate the coder with an amazing product and i cant wait for the next versions.
Using CCE 2.50 btw.
Thx jdobbs keep on coding :)

Hmm forgot to mention that i burned it and i can fast forward etc with it np.

[StifflerStealth]

I haven't actually looked at the inf file yet. I assumed everything in there was the setting for the program itself.

The method I posted for doing CCE on your own is kinda convoluted, huh? :D

Stiff

[jptheripper]

you might want to use something such as titlesetblanker or dvdstripper to remove extras first, then run rebuilder.

just a thought

[wmansir]

Originally posted by jptheripper
Also can someone explain the addaudio() and resampleaudio for the .avs? or these needed ever if I am preserving the existing audio?
-jp

These are used to workaround a bug in CCE. If the .avs file does not have audio it can cause problems in CCE. In earlier versions (2.5 and below) the program will crash on AMD systems. In newer versions it causes a significant memory leak.

ResampleAudio(44100) was a solution in older versions of AviSynth because it exploited a bug in the function that added a blank mono audio track to the output if no audio was present. That bug has since been fixed. AddAudio() is basically a macro function that does the same thing. It is useful for editing clips by hand, but a program like this could just as easily use the 'long version' for compatibility.

AddAudio() = AudioDub( last , BlankClip() )

[jptheripper]

thanx for that info about the audio, i guess i should add the add audio back in.

@stiffler,
convuluted to say the least, but thank you for taking the time. I love the potential of this program. Especially like the fact that it breaks the titles into chapters, allows for easy assigning of bitrates to credits, etc.

Question, in the dvd spec is there a minimum bitrate? i.e. if i set the credits chapter to a bitrate of 1 (guessing 0 would not work) would this make super crappy credits that take no space and allow for preservation of the dvd structure?

:)

Off to figure an easier way to calculate the bitrates for the .ecl file, im a self ordained excel guru, just gotta fgure the formula. Actually, now that I think of it, the .inf file has everything we need. Given the bitrate assigned in RB, we can back-calculate the total file size of the video to be encoded
jdobbs if you could just add the total GB of the video portion this would make this much easier)
Then using the frame rate code (1 is 23.976 and 4 is 29.97 right?) we could differentially calculate bitrates for each segment quite quickly.

ahh.. this is fun

[redfive19]

@jdobbs

Sorry to bug you, but what should I set my Film Threshold to, it's at 70% now. I had the problem with a jumpy picture on my CCE encode.

[StifflerStealth]

Originally posted by jptheripper
thanx for that info about the audio, i guess i should add the add audio back in.

@stiffler,
Question, in the dvd spec is there a minimum bitrate? i.e. if i set the credits chapter to a bitrate of 1 (guessing 0 would not work) would this make super crappy credits that take no space and allow for preservation of the dvd structure?

.
.
.

this is fun


If you look at the big Rebuilder.ecl file that DVD-RB creates, you will notice that it sets the vbr min rate to "0". I would set the credits with an avg vbr bitrate of 1850 and a max vbr bitrate of 2500

Yes this is fun! I like tearing a movie to peices because it's easier to find easter eggs. I found several so far. Then I back track to figure out how to activate them in the menus and such.

Stiff

[jptheripper]

I was actually thinking that in long movies it might be of value to set the average bitrate dynamically by chapter (specifically, main movie to 3000+ and credits to 1 or whatever) to sneak out extra bits, especially if the credits are long. I am working on the excel spreadsheet that will autocalculate the values for bitrates for you based on the RB output.

man jdobbs you really opened a can of worms, THANK YOU

[PurpleMan]

I have yet to use the program myself, but I did manage to scroll and read this entire thread. And from what I pick up, A few comments/questions to the author:

1) Do you think it's wise to encode cells seperatley? I mean, sure, it helps you keep the original cell/vob ID structures when rebuilding the output, but by doing so you're losing the advantages of VBR encoding since it's ineffective on small segments.

2) I think that your source detection (.d2v creation) algorithm needs tons of work. You can't use a general setting to all VTS sets. Most DVD discs have mixed content framewise. For instance: Main movie (Progressive FILM - will need pulldown), special feature: TV-Spot (Interlaced 29.97i - decomb should be used), etc. Eyes`Only's source preparations algorithms from DoItFast4U are brilliant - it detects properly everything, even field orders, and writes the AVS accordingly. You should consider talking to him about it, or you could correct me if the abovementioned is incorrect.

3) An interface where you could set compression ratios to each titleset seperately, as people mentioned, is obviously a neccesity. However I did gather it's your final goal. So we'll see :)

Other than that, I'd like to thank you for the initiative you took, which is obviously a step towards the right direction for the DVD backup community.

PurpleMan.

[djan]

Originally posted by PurpleMan

1) Do you think it's wise to encode cells seperatley? I mean, sure, it helps you keep the original cell/vob ID structures when rebuilding the output, but by doing so you're losing the advantages of VBR encoding since it's ineffective on small segments.

It's exactly what I wanted to talk about. It would be much better if the vbr is applied to the entire movie. With small cells or plainty high-motion cells, the vbr doesn't make its good work.

Anyway, again thx for this great program, I'll donate as soon as paypal is ready for Europe.

[int 21h]

Originally posted by PurpleMan
2) I think that your source detection (.d2v creation) algorithm needs tons of work. You can't use a general setting to all VTS sets. Most DVD discs have mixed content framewise. For instance: Main movie (Progressive FILM - will need pulldown), special feature: TV-Spot (Interlaced 29.97i - decomb should be used), etc. Eyes`Only's source preparations algorithms from DoItFast4U are brilliant - it detects properly everything, even field orders, and writes the AVS accordingly. You should consider talking to him about it, or you could correct me if the abovementioned is incorrect.

If you have a MPEG-2 that is flagged interlaced, and truly is interlaced, and your aim is to backup or reproduce this DVD, why would you not include the original field format? I am confused as to why we would use Decomb unless we are truly IVTCing...

[jdobbs]

@toolman2k
does your tool only encode the video or also menu's ? tools like instantcopy and dvdsrhink also compress the menu and/or strip languages which is very handy. lately more and more menu vobs are really big like 500mb+ some even around 1gb! True enough. Right now menus are not supported, I decided to release the beta before working on that, but I can see it in a future version.

Also, for those who have mentioned it -- I will take a look this weekend and see how hard it would be to set different bitrates/passes for the main movie and extras. I may even add the ability to make the extras half-d1. The workload is getting a little rich, though, so don't take this as a promise

[PurpleMan]

If you have a MPEG-2 that is flagged interlaced, and truly is interlaced, and your aim is to backup or reproduce this DVD, why would you not include the original field format? I am confused as to why we would use Decomb unless we are truly IVTCing...


To enhance performance on HDTV's, for instance. Progressive should always be a goal. But even when neglecting this option, don't forget about the option that main movie is FILM, and an extra feature is a FILM with PULLDOWN hardcoded as 29.97, where IVTC is needed + a progressive flag.

The bottom line, each D2V needs to be detected as for the best method to use for its encode in similar to what DoItFast4U does.

PurpleMan.

[StifflerStealth]

Originally posted by PurpleMan
1) Do you think it's wise to encode cells seperatley? I mean, sure, it helps you keep the original cell/vob ID structures when rebuilding the output, but by doing so you're losing the advantages of VBR encoding since it's ineffective on small segments.
[/B]

I'm not the author, but I will give a shot at this.

In DVDs, chapter points are always at the start of a new cell. The start of a Cell must always fall on an I-Frame of a GOP sequence. If you have the entire movie as one large video, CCE will change where the I-Frames are, thus changing the position you can have a chapter point. One-click transcoders do not change the GOP at all, hence, you don't have issuses. You can tell CCE where you would like an I-Frame, but usually it's off up to 10 frames (that's the max I've seen it off at least). By breaking the video file into cells you are basically forcing I-Frames in the exact spot you need them.

I actually thought about making a One-click CCE utility, and then I saw this thread. I did think a lot of all the features I needed. None-the-less, I can up with the fact that a one-click CCE Util needs to back up by Cells. It is the only way. Yes you do sacrafice the beautiful nature of VBR, but you can encode each cell with a different bitrate.

TMPGEnc can keep the original GOP as the original file. You can even hand pick a bitrate for each frame of the movie, but I don't think the quality is as good as CCE.

Hope this answers your question and then some.

Stiff

[PurpleMan]

In DVDs, chapter points are always at the start of a new cell. The start of a Cell must always fall on an I-Frame of a GOP sequence. If you have the entire movie as one large video, CCE will change where the I-Frames are, thus changing the position you can have a chapter point. One-click transcoders do not change the GOP at all, hence, you don't have issuses. You can tell CCE where you would like an I-Frame, but usually it's off up to 10 frames (that's the max I've seen it off at least). By breaking the video file into cells you are basically forcing I-Frames in the exact spot you need them.


Yes, I know all that. obviously that's why I said that I know that it's used to help keep the original DVD structure.
But in which case, this program is rather futile as it claims to achieve quality, which it clearly does not in comparison to BIG3 method, and in some cases, in comparison to transcoders. If you take a cell that is a fast action scene (a car chase or such), and try to encode it seperatley at a bitrate of 2350kbit/sec, the result will be inferior to a DVD-Shrink equivalent project.

We obviously needs to achieve a way to control the GOP structure of an encoded MPEG2 file. But I do not think that this is the way.
It ain't a new suggestions either. That's what NUMENU4U uses to keep the structure of menus.

[jdobbs]

@Purpleman
1) Do you think it's wise to encode cells seperatley? I mean, sure, it helps you keep the original cell/vob ID structures when rebuilding the output, but by doing so you're losing the advantages of VBR encoding since it's ineffective on small segments.The effect is minimal. Your statement would only hold true for exceptionally small clips. But, my testing shows that once you are at the frame-count level associated with a typical cell, you pretty much get equivalent results to VBR on an entire stream. But -- one of the reasons I made the .ECL and .AVS files available is so people can play...

My goal is to make this as friendly as possible to people (like me) who want to tinker. Also to those folks, you might note that if you change values in the .ECL file, you can still use the automated (Encode and rebuild) functions. DVD-RB takes the single massive .ECL and pulls file sections individually into a temporary .ECL during encoding. That's because of the well-known memory problems associated with loading large groups of files in one encode session.

2) I think that your source detection (.d2v creation) algorithm needs tons of work. You can't use a general setting to all VTS sets. Most DVD discs have mixed content framewise. For instance: Main movie (Progressive FILM - will need pulldown), special feature: TV-Spot (Interlaced 29.97i - decomb should be used), etc. Eyes`Only's source preparations algorithms from DoItFast4U are brilliant - it detects properly everything, even field orders, and writes the AVS accordingly. You should consider talking to him about it, or you could correct me if the abovementioned is incorrect.I don't make a general setting for all VTS sets. DVD-RB analyzes each cell separately and determines whether is should or should not need pulldown. Then pulldown is automatically applied when appropriate during rebuild. One of the advantages of doing my own scan (instead of calling DVD2AVI) is that I actually get to see all the GOP and frame flags myself.

[quantum]

@jdobbs
Can you explain a little about how dvdrb determines and breaks up clips? When an ntsc extra has mixed telecided and interlaced video is it broken when there is a change from one to the other?

Are you using the patterns in the d2v file to determine telecided / interlaced?

Is TFF and BFF examined and set automatically?

From my experience with CCE, getting these right can be very difficult, and problems may only manifest on a relatively small percentage of titles, where jerky frames can appear at sporadic points. I'm hoping you've exercised some brilliance and figured this stuff out so it works 100% :-)

[StifflerStealth]

Originally posted by PurpleMan
Yes, I know all that. obviously that's why I said that I know that it's used to help keep the original DVD structure.
But in which case, this program is rather futile as it claims to achieve quality, which it clearly does not in comparison to BIG3 method, and in some cases, in comparison to transcoders. If you take a cell that is a fast action scene (a car chase or such), and try to encode it seperatley at a bitrate of 2350kbit/sec, the result will be inferior to a DVD-Shrink equivalent project.

We obviously needs to achieve a way to control the GOP structure of an encoded MPEG2 file. But I do not think that this is the way.
It ain't a new suggestions either. That's what NUMENU4U uses to keep the structure of menus.

I did play around with a way to force the GOP structure in CCE. It's a bit of a lengthly process. This is what I did:

I extracted the entire video file of one PGC. I opened IFO Edit and used the tool "File->Check m2v file". I selected the demuxed file and saved the txt file. This tool ouputs the entire GOP structure of a video. If you look at the IBBP stuff is not all the same length, and the Bs and Ps don't always follow the same pattern. Anyways, I worte a cheap and dirty program to get the I frames out of that file and convert them to frame numbers that BatchCCE supports. I opened the video file in BCCE and imported the chapters. One lone mother of a list. This basically forced CCE to follow the Same GOP, except for the fact that the Bs and Ps would not follow the order, but who cares about them. Since this list forced the same GOP length the chpaters worked out perfectly in DVD Maestro. I used IFOUpdate to get the Adjusted mode Chaptes. I mean they fell right on the money. I think I got lucky on this. I only encoded one small PGC with 8 Cells and a total length of 44min 10secs. I need to do more testing to see if other video files will fall right on the money.

Stiff

[jdobbs]

But in which case, this program is rather futile as it claims to achieve quality, which it clearly does not in comparison to BIG3 method, and in some cases, in comparison to transcoders. If you take a cell that is a fast action scene (a car chase or such), and try to encode it seperatley at a bitrate of 2350kbit/sec, the result will be inferior to a DVD-Shrink equivalent project.Rather futile? That's a little harsh, don't you think? I'm also not in competition with the Big-3 method, nor do I intend to be. I've worked closely with Eyes-Only in the past and as you may recall, one of my tools are a part of that method as well.

[hypo20]

@jdobbs

Again. THANKS FOR THIS PROGGIE.

Tested on LOTR2 with all audio and subs and the picture is better as with any other SW with only one audio lang. :) CCE is CCE. :)

[quantum]

But in which case, this program is rather futile as it claims This is rash of improperly drawn conclusions, especially considering the program has been out in beta for only 4 days.

[Rombaldi]

@jdobbs

I must add to the chorus of cheers for this tool. Absolutely amazing work in such a short time.

as far as your future efforts on this..

- being able to set a specific VTS to half D1 (ie. for extras) would be brilliant, as would being able to specify a given bitrate/pass count per VTS.

- I'm lusting over that "4:3 LB to 16:9" option. I've got more than a coupla of discs I would like to pass that thru (some are plain DVD5 so compression isn't an issue, the anamorphizing (word?) is". obviously being able to set this on a VTS basis is required (ie, main movie 4:3LB>16:9, extras could be real 4:3 full frame, not good to chop them eh?)

- One thing that I haven't heard people speak of yet (perhaps with good reason).. PAL > NTSC, NTSC > PAL conversion??? [ducks and runs for cover]

Brilliant, absolutely brilliant...

[jdobbs]

@Quantum
Can you explain a little about how dvdrb determines and breaks up clips? When an ntsc extra has mixed telecided and interlaced video is it broken when there is a change from one to the other? Kinda, sorta. I break out based upon VOBID/CELLID. For telecining I look at the flags and count the RFFs -- I then use the threshold value set in SETUP to determine (at the cell level) which .D2V to use (I create two, one at 29.97fps and one at 23.976) and whether to apply pulldown on rebuild.

ADDED: I want to be careful about the CELL level part -- there are times when I can break out at a sub-cell level, but that is only for support of VOB/CELL interleaving.

[RB]

Originally posted by StifflerStealth
I did play around with a way to force the GOP structure in CCE. It's a bit of a lengthly process. This is what I did:
Huh? :) The beauty of DVD-RB is that it does not require an identical GOP structure. Though DVD-RB encodes at the cell level, the re-encoded cells will have a different GOP structure already and as you know, DVD-RB muxes this back just fine.

What we want is the ability to encode the many AVSes as a big one, setting I-Frames at the cell boundaries. That would ensure that jdobbs could break out cells from the big MPV.

[PurpleMan]

Rather futile? That's a little harsh, don't you think? I'm also not in competition with the Big-3 method, nor do I intend to be. I've worked closely with Eyes-Only in the past and as you may recall, one of my tools are a part of that method as well.


Well, what is the point of using CCE if you are forced to use segments? a thing that we tried to avoid so many times when trying to achieve 1-click-backup-solutions that'll use CCE.

I might've been a little harsh with that statement and I apologize.

I guess I just expected a breakthrough in DVD backup or hoped you figured out a way to make CCE follow a GOP structure and got disappointed when I saw it's just the same old workaround which I personally (but that's just me) find inappropriate for quality backups. It might be suitable for others.

In any way, any contribution to the DVD backup scene is greatly appreciated. Hence, so are you. Sorry if I unwillingly implied otherwise.

PurpleMan.

EDIT:

p.s.


<FMalibu> I already suggested in an earlier post to set the bitrate for each cell proportionally to the original size of the cell


Would be an excellent idea to override the quality loss in the current state of the program.

[robw]

Hello

Just want to confirm that I am getting the same error reported by Quantum back on page 9 of this string.

I have my DVD mounted from an ISO using Daemon Manager v3.44. When I process the file under DVDRB the first two stages go as expected. Then on third stage, Rebuild, I have an error. The VIDEO_TS folder is produced and a number of files are created quickly. Then I get a

Run-time error '75'
Path/File access error

I copied the VIDEO_TS from the Daemon Manager to a hard disk subdirectory and reset the source to point to this VIDEO_TS in DVDRB. Then after pressing Rebuild all goes well.

Seems that DVDRB is trying to get write access to the Daemon Manager, which of course is not allowed for the virtual drive.

[quantum]

Just want to confirm that I am getting the same error reported by Quantum back on page 9I should add I don't think this happens every time. I'm pretty sure I've done some disks from iso mounted drives that worked, so maybe it's happening only with certain disks.

[StifflerStealth]

Originally posted by RB
Huh? :) The beauty of DVD-RB is that it does not require an identical GOP structure. Though DVD-RB encodes at the cell level, the re-encoded cells will have a different GOP structure already and as you know, DVD-RB muxes this back just fine.

What we want is the ability to encode the many AVSes as a big one, setting I-Frames at the cell boundaries. That would ensure that jdobbs could break out cells from the big MPV.

@RB: First, forgive me for not being clear. I type ahead of my brain sometimes :). PurpleMan (Edit: I originally had PurpleRain there) was asking about encoding video files by the entire video instead of by cells. I replied about an experiment I did that was completely un-related to DVD-RB. I took the video from an entire video, and not on the cells. I used one PGC just to make the video file smaller (44minutes 10 seconds). Then I used IFO Edit and a program I wrote to convert the GOP to frames based on the location os I-Frames. I used cell times generated from IFOUpdate in DVD Maestro, and it worked without complaining about I-Frames.

I think is just as confusing as what I wrote above, but if read the previous post several times, and then this post, you might be able to understand what I wrote. :D

Stiff

[jptheripper]

psuedo proud of myself, i figured out an excel spreadsheet to take the .inf file and set up a KB per titleset sheet that will recalculate the KBs if different avg_bitrates per segment are used. All a need now is a listing of the current avg_bitrates per segment and current avg_bitrate of the whole title (jdobbs is this outputable or is it something i will need to dig into bitrate viewer for based on start and end points). If i had that table, here is what i propose (based on many previous earlier comments).

Take listing, and instead of assigning 1 bitrate based on total and then forcing it to each segment we can weight the bitrate based on length and existing bitrate relative to existing average for the whole title. I.e. if segment 2 has lots of action and the avg_bitrate in that segment is say 20% then the avg_bitrate for the whole title, then we can proportionally assign the bitrate for RB to the same ratio. I am sure i garbled that but i hope you know what i mean.

jdobbs, stifflerstealth, if you want to see the spreadsheet just let me know.

-jp

[jdobbs]

@Purpleman
I guess I just expected a breakthrough in DVD backup or hoped you figured out a way to make CCE follow a GOP structure and got disappointed when I saw it's just the same old workaround which I personally (but that's just me) find inappropriate for quality backups. It might be suitable for others.Again I disagree. I think the number of frames associated with a typical Cell is large enough that it rises above the point at which the quality curve associated with VBR flattens out. Eliminating the obvious single frame cells.

But I'm willing to compromise for the cause. If you can run some tests against commercial DVDs that prove that the bitrate changes dramatically from cell to cell, I'll insert a routine that allocates bitrates proportional to the original allocation.

[FMalibu]

Originally posted by PurpleMan
Yes, I know all that. obviously that's why I said that I know that it's used to help keep the original DVD structure.
But in which case, this program is rather futile as it claims to achieve quality, which it clearly does not in comparison to BIG3 method, and in some cases, in comparison to transcoders. If you take a cell that is a fast action scene (a car chase or such), and try to encode it seperatley at a bitrate of 2350kbit/sec, the result will be inferior to a DVD-Shrink equivalent project.

As I have stated previously in this thread (god knows where though, this thing is getting huge...), I agree with you on the issue of the bitrate distribution. There will be a huge difference in how much bitrate each cell on a DVD needs to get the same average quality. However, I greatly applaud the efforts of jdobbs in creating this program, after all this is a very early beta. The word futile is then not one I would use, and I suppose Purpleman has already revoked this statement.

I have been thinking about this bitrate distribution issue, and if you take the time to read this post, I will attempt to illustrate a method that I feel comes as close to a perfect bitrate distribution as possible when encoding individual cells.

Basically it would involve adding an extra "Prepare" and "Encode" step to the existing three step process. (I know, it's supposed to be a one click program, but bear with me :) )

1. The first prepare step would do the same thing it does now, only it would generate an .ecl file that is set to encode single pass VBR à la roba for every cell.

2. The first encode step would encode the generated .ecl file, resulting in lots of one pass .mpv files.

3. The second prepare step would the use the size of each cell from the one pass VBR encoding to calculate it the size it should be to fill up the DVD, i.e. it would scale up/down the size of each cell with the same ratio. This would also allow for variance between the main movie titleset and extras (e.g. if you want the extras to be 2/3 the relative quality to the main movie or something). This step would generate another .ecl file based on these bitrates.

4. The second encode step has CCE encode the number of specified passes.

5. Rebuilding...(jdobbs black magic :) )

As the single pass VBR encode would determine the compressability of a certain cell as compared to all the other cells, this method would provide a good bitrate distribution, provided that the sizes of the cells are all multiplied by the same ratio.

To jdobbs: Please do not think I'm trying to dictate you how to write your program, I'm merely attempting to give what I think is a good suggestion. I encourage anyone to question, discuss or disprove this method.

-- FMalibu

[jdobbs]

I'm going to have to go back through this thread and start collecting all the suggestions and see how many promises I've made. I'm afraid I might get to a point where my mouth is writing checks my ass can't cash...

[jptheripper]

dont think you need to go that far (my opinion) but i think the need to set per title is a must, but cell would just be nice based on existing.

is there anyway you could add the existing avg_bitrate to the .inf for each cell if you are already calculating it or if it is a small task to do so? i.e. just add an existing_avg_bitrate = xxxx right with the frames etc, that way if we wanted to do it by hand we could.

-jp

[jptheripper]

oh and please dont forget the offers of (free) assistance.

and please take the majority of the criticism as compliments, we are all just so excited and thankful for the tool we want to help you make it be the best it can be

-jp

[jdobbs]

@jptheripper
You bet. If you don't want criticism you kind of defeat the purpose of a beta.

[toolman2k]

Originally posted by Rombaldi
[B]@jdobbs

- being able to set a specific VTS to half D1 (ie. for extras) would be brilliant, as would being able to specify a given bitrate/pass count per VTS.


hmmmm ive tried doing half d1 on extras with some movies. it works very well. although not dvd compliant, most players it ok and because of half of total lines being used, quality looks alot better at low bitrates, less blocks, smoother images etc. so yes please make this happen :)

BUT, what i noticed with these half d1 clips is that they did play ok on my standalone, but not in powerdvd. subitles appeared wrong on the video. probably because the subpicture layer is still full resolution and the clip half causing displacement. but oh well....i play them on my stand alone so....and theyre only xtra so i could live with that :)

but .... maybe you could also adjust the subpictures to match the half d1 video ? now that would be very nice :)

[jdobbs]

@toolman2k
hmmmm ive tried doing half d1 on extras with some movies. it works very well. although not dvd compliant, Half-D1 is very much DVD compliant -- if it wasn't I wouldn't do it.

[quantum]

- being able to set a specific VTS to half D1 (ie. for extras) would be brilliantI messed around with this for some time and finally decided you get about the same results as half-d1 by increasing the softening effect, which is changing the "low pass" filter to a lower number, in CCE.

The result looks almost the same, and maybe better than half-d1 for low bitrate extras, and you don't have to worry about problems with subtitles or compatibility.

[jdobbs]

Hey, quantum -- great idea, and a lot easier to implement...

[FMalibu]

Originally posted by jdobbs
But I'm willing to compromise for the cause. If you can run some tests against commercial DVDs that prove that the bitrate changes dramatically from cell to cell, I'll insert a routine that allocates bitrates proportional to the original allocation.

Now you just got me curious if my claims are true, so I put it to the test. I took movie with some high action scenes, The Matrix R1. I know it can't really be backed up with DVD-RB (yet), but this was one of the few action movies I had availlable.

I demuxed by cell ID and here are the results of the first half of the movie:


V02C01 5429
V02C02 6429
V02C03 5292
V03C01 5199
V04C01 5104
V04C02 5549
V06C01 5702
V06C02 5391
V06C03 5260
V06C04 5166
V06C05 5198
V06C06 5235
V06C07 5348
V06C08 5979
V06C09 5580
V06C10 5356
V06C11 5749
V06C12 6385
V06C13 5830
V06C14 5146
V06C15 5414
V06C16 5684
V07C01 4065
V08C01 4779
V08C02 5372
V10C01 5712
V10C02 5599
V10C03 5342
V10C04 5371
V10C05 5431
V10C06 5336
V10C07 5365
V10C08 4694
V10C09 5213
V11C01 6349
V12C01 7765
V12C02 6579
V14C01 5696
V14C02 6653
V14C03 5179
V14C04 6487
V14C05 6218
V14C06 5239
V14C07 4846
V14C08 5278
V14C09 5273
V14C10 5281
V14C11 6013


You can see that the bitrate varies what I think is a significant amount. I started testing it on Fight Club R1 as well, but then lost the results. The maximum amplitude there between cell bitrates was about 2 Mbit I think, but obviously I can't back that up right now.

The method I proposed is indeed more difficult to implement, but I think it provides the best chances of a fair distribution without a movie, and more importantly between titlesets. I have seen a lot of DVDs that had disproportionate relative bitrates/quality between titlesets.

Also, it provides RoBa support, which rocks. For those not familliar with the merits of RoBa, I have found that it generally produces the same result as multipass encoding, but in one less pass. This means that a 4 pass multipass encode (incidently as many passes as anyone would ever want) can be achieved with RoBa in 3 passes. At least that's my theory according to my findings.

-- FMalibu

[jptheripper]

@fmalibu

can you tell me your method for demuxing by cellid and then determining the bitrates? i have my calculator ready to go but i am missing this one piece of info.

if you dont want to clutter the thread please pm me the steps

-jp

[wmansir]

Originally posted by quantum
I messed around with this for some time and finally decided you get about the same results as half-d1 by increasing the softening effect, which is changing the "low pass" filter to a lower number, in CCE.

The result looks almost the same, and maybe better than half-d1 for low bitrate extras, and you don't have to worry about problems with subtitles or compatibility.

Did you try this out on interlaced material or just progressive?

I find half-D1 is good for interlaced material, which a lot of extras are. But I have never compared the results with filtering, partly because filtering interlaced material can be tricky. Another benefit of half-D1 is that it encodes much faster, as a opposed to filters which slow down encoding.

[jdobbs]

@FMalibu

Okay, Matrix is a good extreme so you convinced me. I'm a little surprised, though. These are average bitrates, right?

[RB]

AFAIK Half-D1 can't be anamorphic though.

[SiXXGuNNZ]

Originally posted by jdobbs
NOTE: NEW VERSION LOADED AFTER 1096 DOWNLOADS ON PAGE 4

wow, I got this forum set to view 30 posts per page, giving this 12 pages, it is not on page 4 :\

edit: found it logged out, page 8 or something, page 6 when viewing 30 posts

http://forum.doom9.org/showthread.php?s=&postid=459634#post459634

and attachment: http://forum.doom9.org/attachment.php?s=&postid=459634

[joaoccc]

Hello

DVD-RB give me the error 9, subscript out of range in shangaii nights.

Anyone had this problem ?

And this movie don't had angles.

Reards

[StifflerStealth]

I was able to do several no thrills DVDs, so I think I'm going to test a DVD that caused Scenarist, DVD Maestro, and even InstantCopy v8 to fail. This DVD has a really strang structure with Hidden Cells. The only way to extract hidden cells is to use Dem(f)ux by CellID. Check this thread for more info (http://forum.doom9.org/showthread.php?s=&threadid=72210). Read all the post so you can get an idea on what's going on with this disc. This will really test your m(f)uxer. If your proggie can m(f)ux this, then it will have done what several thousand dollar software could not.

Stiff

[quantum]

DVD-RB give me the error 9, subscript out of range in shangaii nights.I think I had this on one disk. I'll bet you have a number of cells or chapters or pgcs or something that is going beyond what jdobbs originally allocated. It's a good idea to post the disk name as you did in case jdobbs has the original for testing.

@jdobbs: Yet another thing to consider, since you probably don't have enough to think about, is to cause vb to give the line number when an error occurs. This can help with debugging since you can tell exactly where in your code things like "subscript out of range" occured. This is actually not that simple to implement. I've done it and can help out when you run out of other things to do.