DGMPGDec 1.0.12 released!

[Zep]

Originally posted by neuron2


BTW, I have the Set as Video and Set as Audio buttons working in the detect PID dialogs, but you can't have that until this end of file issue is fixed. :p

haha hey i thought you agreed with me that I was too lazy when
i asked for that feature and you were not gonna add it :D

I'm super glad you went for it!!!


Yes another great version of DGindex out real soon now yum yum

[Guest]

@Zep

Now you know, I'm lazy too. Especially in the pool. Sadly.

Sopranos Season Finale in 18 minutes. I'm outta here!!!!!!!

[len0x]

What just happened with the post for 1.0.12 ? :) (I got the version though anyway).

Don, you might want to swap buttons "Set Audio PID" and "Set video PID" as usually you set the video first and I spent whole 3 min just clicking on the first button and didn't realize that that's audio I'm setting :)

[Cyberia]

Ok this is weird... Try saving a project repeatedly. Save one, let it complete and observe the Elapsed Time in the Info Panel. Then goto File/Save Project again and re-save it. Now look at the Elapsed Time.

My testing on a vob file gives me these results:

1st save: 4 seconds elapsed
2nd+ save: 10 seconds elapsed.

Nothing changed! I have to close DGIndex and reopen it to get the 'fast' mode to work again.

Don, could you make the info panel show tenths of seconds for the elapsed and remaining times?

EDIT: I can now confirm the behavior on multiple vobs. Also, opening a different vob file reactivates the 'fast' mode. But trying to reopen the current vob doesn't get fast-mode back.

[Guest]

@len0x

I found another little issue at end of file that I want to fix, so I withdrew it but you were too quick.

@Cyberia

I'll try to duplicate it.

[Guest]

Something's gone haywire with my web site. It keeps returning cached files even after I replace them. So, to get THE REAL version 1.0.12, use this link:

http://neuron2.net/fixd2v/dgmpgdec1012a.zip

This version has the set PID buttons and the end of file fix. If you have the Set Audio button on the left, you've got the WRONG 1.0.12 and need to tget the one above.

[vispgraedde]

As I was unable to make this new version bug out with my clips :D, I looked for "bugs" in the html files :devil: instead.

It's really minor but

&nbsploadplugin("...\DGDecode.dll")
&nbspmpeg2source("myvob.d2v")


is lacking some ; in the underlying html

in QuickStart.html "What is This Index File and What Do I Do with It?"


(Sorry, but I couldn't resist :D Thanks for improving this nice prog!)

[Cyberia]

Originally posted by vispgraedde
As I was unable to make this new version bug out with my clips :D, I looked for "bugs" in the html files :devil: instead.

It's really minor but

&nbsploadplugin("...\DGDecode.dll")
&nbspmpeg2source("myvob.d2v")


is lacking some ; in the underlying html

in QuickStart.html "What is This Index File and What Do I Do with It?"


(Sorry, but I couldn't resist :D Thanks for improving this nice prog!)

I don't see anything wrong. Please spell it out for us simpletons.

[Guest]

My & n b s p ; spaces are missing their ;.

[eb]

http://neuron2.net/fixd2v/dgmpgdec1012a.zip

as tested so far:

.vob files demuxing all audio streams nicely and simultaniesly
.mpg no problem to demux audio
.ts streams demuxes selected audio no problem

all this above is also working well on "cutted"-selected pieces of the file in DGIndex timebar.

in .ts stream PIDs carring data for teletext/subtitles are wronly named as AC-3 PIDs, generaly for subtitles and AC3 audios Private Streams are used, but for Europe the BD streams are used as I met so far


reported data DELAY after short tests seems be correct

Greeting, nice work

eb

[Guest]

Originally posted by eb
in .ts stream PIDs carring data for teletext/subtitles are wronly named as AC-3 PIDs, generaly for subtitles and AC3 audios Private Streams are used, but for Europe the BD streams are used as I met so far What are the BD streams? How do I distinguish the AC3 from the subtitles if they're both on Private Stream 1? Thank you.

[eb]

neuron2 wrote:What are the BD streams? How do I distinguish the AC3 from the subtitles if they're both on Private Stream 1? Thank you.

Sorry, my foult to be not precise.
In all cases as I met so far in Europe tv AC3 2.0 or 5.1 was always as Private Stream 1 with hex value BD.

eb
edited: see example p7.ts
edit 2 : Yes , it is problem , main descriptions are exactly the same , we must look for the additional byte or flag connected with audio.I have no any detailed specification at the moment

edit 3: I uploaded sample tx.TS with subtitles on:
ftp://www.eb.enterpol.pl
user: www.eb.enterpol.pl
password : eb

edit 4: asked for help Marfi author of program SkyView for SS2 cards

and this is answer: AC3 decsriptor=0x6a ,for teletext 0x56

[Zep]

Originally posted by neuron2
Something's gone haywire with my web site. It keeps returning cached files even after I replace them. So, to get THE REAL version 1.0.12, use this link:

http://neuron2.net/fixd2v/dgmpgdec1012a.zip

This version has the set PID buttons and the end of file fix. If you have the Set Audio button on the left, you've got the WRONG 1.0.12 and need to tget the one above.



UPDATE - I deleted what i said until I can get more info
and the person who capped the stream can look at it with 112a

he is downloading it from me now.

[Cyberia]

Originally posted by neuron2
@Cyberia

I'll try to duplicate it.

I'm home now on an Athlon 2200 system with 1024MB of RAM, and cannot reproduce the problem. But at work on a 2Ghz P4 system with 256MB of RAM, I could make it happen every time.

I am using the same vobs in both cases.

[eb]

Zep wrote:ALL 50+ PIDs are flagged as OTHER

If it is record from ALL data from the given transponder then probably there are many of encrypted channels.

eb

[Zep]

Originally posted by eb
Zep wrote:

If it is record from ALL data from the given transponder then probably there are many of encrypted channels.

eb

I am trying to get what exactly it is now. i do not think
it is a .ts but a PES or some such. I just do not know
and have to wait for him to double check.

the person who sent it to me to test is not even sure and
is double checking now

[eb]

Zep,
Try to play this file with VLC.player,
Right click on screen and set different programs
I am interested with results
eb

[Zep]

Originally posted by eb
Zep,
Try to play this file with VLC.player,
Right click on screen and set different programs
I am interested with results
eb

it wasn't a .ts stream he sent me a CONVERTED .ts to
some strange vob look alike.

wish he would have told me since he is only supposed to send
orginal unaltered .ts streams lol

[buzzqw]

@neuron2

a request:

Would be possible to put in preview info the number of current frame viewed ?
I would use this feature to know when trim/credits in avs script. and, yes, i cannot use the "trim" builded-in

Thanks

BHH

[Guest]

Originally posted by buzzqw
Would be possible to put in preview info the number of current frame viewed ?
I would use this feature to know when trim/credits in avs script. and, yes, i cannot use the "trim" builded-in Sorry but not possible. When you move the timeline slider, the decoder just starts decoding and displays the first I frame it encounters. To know what frame number that is would require parsing/decoding from the beginning to that point.

Why don't you just serve the video into VirtualDub and determine the desired frames in there?

[len0x]

Don, so how about make raw detection available to CLI? And while we on the subject - is it possible to add an option to DGTable to output to file instead of stdout? I have some issues with this as I have to redirect output to a file and this can only be done (afaik) via cmd/command shell, not direct call via CreateProcess(). This complicates procedure a bit coz shell always returns result before file is written by DGTable output (and I have to put artificial delays there). Thanks.

[Guest]

@len0x

Please snd me a PM with your requirements. I'm going to work on CLI stuff tonight (reading files from a text file list). I can do your stuff too but I have to know EXACTLY what you want.

[Cyberia]

Originally posted by neuron2
Sorry but not possible. When you move the timeline slider, the decoder just starts decoding and displays the first I frame it encounters. To know what frame number that is would require parsing/decoding from the beginning to that point.

Oh ok. That was actually on the Enhancement list at #6. I have changed it, hopefully to something more feasible. Here's the new #6:


6) Information window additions (for both Preview and Save Project modes)
a) Info field
b) Vob Cell ID
c) as many other fields as possible


FYI: I added a very minor bug to the list :o

[Guest]

I can't make any more releases for a while, otherwise Gizmotech will mumble. :(

[zettai]

Can't you just call them betas until you get to the 1.1.0?

[Guest]

Yes. Version 1.0.12 appears to be very good, so I'll switch to 1.1.0betaX terminology. That way, it'll be easy to tell what the last stable version is. I can then also release source code for all non-beta releases.

[Zep]

This may be of use

a great page on how TS works and laid out and how to get perfect sync
in your code if you are a dev.


http://erg.abdn.ac.uk/research/future-net/digital-video/mpeg2-trans.html

[Guest]

Originally posted by Zep
how to get perfect sync in your code if you are a dev. I saw some general, well-known stuff, but not what you mentioned above. Where is that, please?

[eb]

quote from above link
They are two types of time stamps:

The first type is usually called a reference time stamp. This time stamp is the indication of the current time. Reference time stamps are to be found in the PES syntax (ESCR), in the program syntax (SCR), and in the transport packet adaption Program Clock Reference (PCR) field.
The second type of time stamp is called Decoding Time Stamp (DTS) or Presentation Time Stamp (PTS ). These time stamps are inserted close to the material to which they refer (normally in the PES packet header). They indicate the exact moment where a video frame or an audio frame has to be decoded or presented to the user respectively. These rely on reference time stamps for operation.

[Guest]

@eb

Your native language must not be English. :)

[eb]

sorry, I tried only to answer

Where is that, please?

eb

[Zep]

Originally posted by neuron2
I can't make any more releases for a while, otherwise Gizmotech will mumble. :(


I hope you are kidding. He can just wait for a point release then.
please don't stop with beta releases :D


thanks!

[Zep]

Originally posted by neuron2
I saw some general, well-known stuff, but not what you mentioned above. Where is that, please?

Oh haha i meant it shows the break down in detail so that
a dev should be able to figure out how to make sure there
is sync :)

How you go about coding it, no it doesn't show that
but the VLC/ffmpeg source does. Oh projectX does also.


thx

[Guest]

Originally posted by eb
sorry, I tried only to answer Don't be sorry, I should be. The thing that you were not aware of is that I know all that basic theory very well, as I work for the market leader in MPEG decoders and integrated set-top box silicon. In fact, I just solved a long-standing AV sync problem for one of our major customers. The problem I have with Zep is that he has supplied a transport stream which presents a strange, unique mystery that we are trying to solve. Believe me, if it were as easy as saying "just present the units when their presentation time arrives", we'd have figured it out long ago.

We already know that DGIndex works out the sync offset only at the beginning and does not add/delete audio to maintain sync if it goes off due to drops, etc., later in the movie. So we know that needs fixing. But we have a case where the calculated offset at the beginning is just off by a whole second, with no obvious reason for it. I don't like to start making code changes with mysteries lingering around, so we are trying to resolve that before deciding what our strategy will be with respect to changes.

I hope that brings everyone up to speed on the context of the discussion.

[eb]

@ neuron2,
I presented this quote traying to say that this piece of iformation is the only one,from the all of the link, that is talking about time dependiences of signal packets.

And the most interesting is last sentence from this quote

These rely on reference time stamps for operation
and this connected with first sentence
This time stamp is the indication of the current time. From my practice with different sattv records, those records with real time stamps are most friendly for further processing.

Time to time i am collecting samples with different strange behaviours for video and audio synchronisations, and here is my ask for the sample from Zep.

Greetings
eb

edited: By underlighting this I am going to say, that if false time stamps for video or audio packets are used by signal suppliers , then most safe way to process this packets is to get time from the refrence time stamps.

[Cyberia]

Originally posted by Cyberia
Ok this is weird... Try saving a project repeatedly. Save one, let it complete and observe the Elapsed Time in the Info Panel. Then goto File/Save Project again and re-save it. Now look at the Elapsed Time.

My testing on a vob file gives me these results:

1st save: 4 seconds elapsed
2nd+ save: 10 seconds elapsed.

Nothing changed! I have to close DGIndex and reopen it to get the 'fast' mode to work again.

Don, could you make the info panel show tenths of seconds for the elapsed and remaining times?

EDIT: I can now confirm the behavior on multiple vobs. Also, opening a different vob file reactivates the 'fast' mode. But trying to reopen the current vob doesn't get fast-mode back.

Don, I tested this here at work again, and it is still happening. Furthermore, TaskManager show DGIndex at 50%-80% cpu usage on the first pass, but only 20%-30% on subsequent passes. The rest of the cpu usage is in the SystemIdleProcess, so nothing is taking cycles away. I can consistently repeat this, so it looks like DGIndex is stalling on the P4 after one pass. Might be a cache issue. My old Athlon box didn't see this behavior.

I spent last night building an Athlon 3200 system for me! 1GB of ram with fast 2-2-2 timings, 10,000rpm drive. I'll happily test anything once I get everything setup :)

[malkion]

Donald, I'd like to add a small request for DGDecode.

Unless it makes absolutely no sense to you, I just happened to rip a DVD by Chapters. Could you add Autoincrement Open to automatically detect chapters? I was using DVDDecrypter and the format is as follows:

VTS_01_CHAPTER_01_1.VOB
VTS_01_CHAPTER_02_1.VOB, etc.

Not sure how SmartRipper names it yet however, sorry.

Also unsure about if there might be instances of VTS_01_CHAPTER... to VTS_04_CHAPTER..., etc., however I suppose this can happen.

[trbarry]

My HiDTV card captures ATSC files using names like Ch025_05_09-07_03pm.0000.tp . Does this conflict somehow with the new DGIndex file naming convention. It seems I have to load them all (dozens) one at a time. Or is there some explanation of Auto Add somewhere that explains this? I couldn't find it.

Also, DGIndex seems to think I have Mpeg audio files instead of AC3. Is there some way to override this?

- Tom

[Guest]

@trbarry

Currently, auto add requires an _ before the number, not a period.

What does the PID detection (PAT/PMT and Raw) show for the PID that you claim is being wrongly determined to be MPEG audio? Can I see a screen grab for each detection dialog?

[Cyberia]

Don, A lot of people are having issues with the Auto-Add feature due to differing naming conventions. I think we might need to revisit this in the future.

I will spend some time thinking about how we should do it. My initial thought is to tie it to the drag-and-drop sorting issue and present a unified auto open method.

In the meantime, one workaround would be to allow multiple files to be selected in the auto-add dialog.

[TripleA]

...regarding Auto-Add:

The way I see it, most people will be opening sets of files created in the same session. Most such file sets will contain numbers that will be incremented at one point or another during the creation of the set, depending on what each number means for whoever wrote the application creating it. With the more significant numbers coming earlier in the name string.

So, the extreme solution would be to disregard all digits(0) for purposes of file-name matching when adding the files. In DOS terms, replace the digits with ?s(1). Then disregard all non-digits for purposes of sorting the list later.

The only problem I see arising from this would be if one puts all files of one type in a single folder with only numbers differentiating them(2). This could be worked around by perhaps having a setting somewhere (maybe on the open dialog itself) to tell DGIndex to treat only the last n sets of digits as numbers.

I know this is rough because I didn't think much about it. But unless I'm missing something major (very likely, I'm afraid), a refined edition of this scheme should put this puppy to sleep.


-=== Break ===-

0- What is a digit? Probable problem with non-standard radices, here. But most probably can be safely ignored: how many people you know count in octadecimal...? ;-)

1- Thus assuming whoever wrote the application would have included leading zeros in the names of earlier files. And in turn assuming she knew how large the numbers are likely to get.

2- Whatever_Year_Month_Day_Time.ext kind of stuff, I mean.

[nanga parbat]

wouldn't it be enough to, as cyberia suggested, allow shift-selecting multiple files (and perhaps leaving the auto-add feature as it is)?
i think most users would know which files to open, so being able to select them all at once would be sufficient.
the other way you would have to deal with quite a variety of naming conventions.

nanga

[Guest]

You'd still have the sorting problem.

[JasonFly]

The question:
Could DGindex or DGDecode.dll be responsible from an error a the end of an encoding process. I explain:
I made a two pass encode of a TV series (1/2 season). So it's quite long (9hours). The source is interlaced so i 've been forced to use Decomb ;) which did a great job, thanks for this. I did the encode 4 or 5 times(each time I thought I had found the cause) because of errors at the end of the encoding process with VduBMod1.4.5.1. The first pass was fine but VdubMod freezed one or two frames before the end of the file(depending on how you interpret VDub Status window).
http://perso.wanadoo.fr/paille/Erreur Vdub-1.png

Analysis:
It is as if DGdecode doesn't give the frames to XviD. Looking at this problem, I looked at my 1st pass XviD statfile and I found that it had 1 frame less than it should (844216 instead of 844217 for the source). So that could be a reason of the crash in the second pass but where does the missing frame come from? XviD? Avisynth? DGdecode? The reason I post in the DGMPGDec thread is that I think that DGDecode could be the responsible, but that's just a suggestion.

Some remarks:
I created one .d2v for each episode and gathered them in avisynth to make the encode. Can this be a cause of my missing frame? I mean:Is it better to make only one .d2v with all the Vobs put in DGindex?

I am thinking writing this post, and I must admit that this would be strange that this error came from DGdecode because the frame(844217) is present in VDub when I go to the end of the file. :confused:


neuron2: I didn't tested the delay issue I talked about earlier in this thread, I'll try to do this later.


PS:Did you know that video encoding could lead you to the edge of depression?

PS2: I'll try to supress two frames before the second pass and I'll see the results. I have already done that previously(and that didn't worked)but I wasn't sure that my d2v and DGdecode version matched so I restarted all the work from the beginning.

[nanga parbat]

You'd still have the sorting problem.

aight, sorry. i thought it was just about auto-detecting, hadn't thought about that...

nanga

[BruceL]

@JasonFly
Have you tried the newest version of VDubMod with latest builds?
The newest version is 1.5.10.1 and the latest build is 2439. You can get it as follows: http://sourceforge.net/projects/virtualdubmod

[trbarry]

I suppose one over simplistic solution would just be to take the rightmost string of consecutive digits, wherever they were and regardless of how many or what (non-digits) surrounded them.

That would take care of all the HDTV cards I'm aware of except the old brain-dead WinTV-HD which doesn't matter anyway.

- Tom

[Guest]

Originally posted by trbarry
I suppose one over simplistic solution would just be to take the rightmost string of consecutive digits, wherever they were and regardless of how many or what (non-digits) surrounded them. I'll look into that. Thanks.

BTW, did you still want help with your audio detection issue? I had asked for some more information about it.

[Cyberia]

Originally posted by trbarry
I suppose one over simplistic solution would just be to take the rightmost string of consecutive digits, wherever they were and regardless of how many or what (non-digits) surrounded them.

I thought about this and can't figure out why we shouldn't just do a straight alpha-numeric sort? This will work for 99.9% of the cases.

I was thinking about different naming conventions and how sometimes I keep vobs and rename them... "bad GOP.vob", "No Doubt - Don't Speak", etc. But then I realized that anytime I would want to use DGIndex to load multiple file, they all have a consistent naming convention and the same digit-count length. So alpha-numeric should work.

If you want to be really flexible provide 'move up' and 'move down' buttons to manually re-order files that don't get sorted correctly.

Handle drag-and-drop the same way (alphanumeric)

Is there a down-side to this obvious solution?

[trbarry]

BTW, did you still want help with your audio detection issue? I had asked for some more information about it.

Sorry. I forgot to respond to that one. It's on a different machine that's currently doing something else. But I'll try to redo it as soon as possible.

- Tom