DGMPGDec 1.0.12 released!

[Guest]

Have you been following my thread in the Avisynth Development thread about Invoke()? :)

That's needed to invoke the conversion filters. So, don't worry, it is in the works.

[zettai]

Ah implementing YUY2 and RGB output via DGDecode? Superb - good to see you don't do things by halves :)

[len0x]

I usually start DGIndex as a minimized application. And on normal DVD sources I never get to see information window, but when dealing with HDTV sources it gets poped up in the beginning (probably due to resizing of the main window or something). Is there a way to avoid that I wonder ?

P.S. I'm still playing with Faith Hill TS file - what sort of audio is there? (doesn't seem to be normal MPA)

[tedgo]

@neuron2

Wow, you are working like a sewing-machine :D
Version 1.0.8 seems to work correctly, and now there is Version 1.0.9 out just a few minutes later ;)
:thanks: for the rapidly updates and your great work

[Guest]

Originally posted by len0x
I usually start DGIndex as a minimized application. And on normal DVD sources I never get to see information window, but when dealing with HDTV sources it gets poped up in the beginning (probably due to resizing of the main window or something). Is there a way to avoid that I wonder ? I'll try to find time to address that.

P.S. I'm still playing with Faith Hill TS file - what sort of audio is there? (doesn't seem to be normal MPA) There's a caveat in the Transport Stream Notes document that says that if the PAT/PMT is absent, the audio type cannot be determined for the specified PID, and it is demuxed as MPEG audio. But that stream has AC3, so it's mucked up.

I need to either add autodetection independent of PAT/PMT or allow the user to configure a fallback type. I'm thinking about the best solution. You'd be better off experimenting with a proper stream.

[Cyberia]

Quod erat demonstrandum.

Watch your language! ;)

Seriously, I vote to require the user to specify the audio format if PAT/PMT is missing. Actually, the user should be able to override the "Auto" feature anyway.

[Guest]

Originally posted by Cyberia
Seriously, I vote to require the user to specify the audio format if PAT/PMT is missing. Actually, the user should be able to override the "Auto" feature anyway. Well, if your specified PID has AC3, there's not much point in overriding it to MPEG audio! So, I am tending toward allowing the user to specify "Fallback Audio Type" to apply when auto detection fails.

[Milkman Dan]

Using the "australia_1.ts" file posted earlier in the thread, I can't get mpeg2source to load the .d2v I made from the .ts. I just downloaded 1.0.9, and the file didn't have the correct tables. I used the raw fuction, guessed correctly on the values, and got working video inside DGIndex. DGIndex also apparently made the .d2v successfully. But trying to load it in VdubMod from the .avs results in an avisynth error message saying that mpeg2source can't decode the video.

The .avs is nothing other than the load statements for dgdecode and undot, and the mpeg2source line itself.

So am I missing something, or did I use the wrong PIDs after all?

EDIT: I must correct myself. The AVISynth error is "Mpeg2Source: Cannot open file." The path is correct.

[Guest]

1.0.9 has a new D2V format. Please remake your D2V, being sure to use the new DGIndex.exe, and be sure to use the new DGDecode.dll, and not pick up an old one by accident.

The software is evolving rapidly, and there will be some stumbles for those who want to stay on the bleeding edge.

[len0x]

Originally posted by Milkman Dan
So am I missing something, or did I use the wrong PIDs after all?


I'm using 200/28a PIDs and have no issues whatsoever. (in fact I just forced AutoGK to produce very nice looking XviD from it :) )

[Milkman Dan]

Yup. I was using the wrong dgdecode.dll file.

Sorry to waste your time with something n00bish. Thanks for the help.

[Guest]

It's my fault, actually. Please do forgive me.

You see, I keep meaning to change the error pop-up to say "Your DGIndex and DGDecode are mismatched!", but I keep putting it off.

[Zep]

Originally posted by neuron2
OK, I've found a VOB that causes it!

Very strange. I'll get right on it.

I'll also add raw PID detection as I am seeing more streams without PAT/PMT tables. That's the cause of your problem, Zep.

Please stick with 1.0.6 until I get this sorted out.

raw pid detection works great! :D


thanks

[Guest]

Originally posted by Zep
raw pid detection works great! Glad to hear it. I'm surprised nobody has asked me to sort the list yet!

I'm wondering, however, where these streams without PAT/PMT are coming from. Are they being stripped of their program guide by some software? Is there a different program guide standard in use? PSIP? Something else? ISO specifies PAT/PMT. Does anybody have any light to shed on this?

[eb]

neuron2 wrote
I'm wondering, however, where these streams without PAT/PMT are coming from. Are they being stripped of their program guide by some software? Is there a different program guide standard in use? PSIP? Something else? ISO specifies PAT/PMT. Does anybody have any light to shed on this?

Generally these streams are from different top set "boxes" of sattv receivers equiped with hard disks to record and then to play back on TVsets.
But people want to have all this records on CDs,DVs and are using different strange metods to rip recods from that build-in hard disks.

Another class are users of sattv PCI cards, and in this case they have rather good knowledge how and what to do with all this "streams"

BTW. Urgently good cutting tool for HDTV .ts records is needed.
May be DarkAvenger can put back resize from DVD2AVI to his MPEG2cut tool to enable acces to the buttons on screens smaller than HDTV picture.
EDITED: or move this buttons to the top like it is done in lattest VLC.player.

As example: try to cut off small .ts (.tp) piece from Australia_1.tp

eb

PS
I can assume that several milions ppl using hammers,screwdrivers and soldering irons are trying to move their hard disks from "boxes" to the computers.
Lattest "boxes" have already build in sockets for cables for USB or 1394 connections to rip this records to the computers.

EDIT 2: @ Dark Avenger
I am sorry but I discovered again already on my computer your Mpeg2Cut2 dated 15.04.2004. It is working very nice with normal HDTV .ts streams,
It is even dealing /but with some trubles / with such difficult files as abnormal Australia_1.tp file

http://www.geocities.com/rocketjet4/
so thank you very much for such very useful tool as Mpeg2Cut2

eb

[Zep]

Originally posted by neuron2
Glad to hear it. I'm surprised nobody has asked me to sort the list yet!

I'm wondering, however, where these streams without PAT/PMT are coming from. Are they being stripped of their program guide by some software? Is there a different program guide standard in use? PSIP? Something else? ISO specifies PAT/PMT. Does anybody have any light to shed on this?

yes i think so.


As it turns out the streams i will send you tommorow
have 23+ PIDS in them. One of them i think is some sort of table
PID with that has that info in it.

it turns out that a bunch of stations are sharing a transponder
and sharing 1 .ts stream. So it is like if abc cbs nbc fox wb upn
all stuck their programs into 1 one broadcast stream and then you
watch whatyou want via pid selection.


(so far that appears to be how it is but I haven't spent much
time looking at teh details yet so...)

[Guest]

Originally posted by Zep
some sort of table That's not a lot of light you are shedding there.

Anyway, I've decided to enhance the raw detection to look at the stream ids to determine what is on each PID.

[TEB]

My understanding of this is that in the DVB world (europe) a given transponder (around 40Mbit/s) has ONE MultiprogramTransportStream (MPTS) wich consists of multiple Singleprogramtransportstreams (SPTS). What we call a channel in normal RL terms is called a SERVICE with its own SID.
IE a spts has its own SID. (Service Identifier)
The SID again has its own elementary streams identified by a PID number.
For example here in Norway the state owned TV channel nr.2 is called NRK2. Its SID is 1502, with the Video PID @ 513 and Audio PID @ 644
In addition comes DVB Teletext (character based and bitmapped, various other streams for MHP and so on.)
AFAIK there are PAT/PMT/SIT tables describing many things like the logical names connected to the SID: NRK2 = 1502. Kind of a DNS resolve :) in the transportstream world.

just my 2 øre! :)

[Cyberia]

@neuron2 - I have found an SVCD that DGIndex will not let me step backwards in at a certain point in the film. I can step forwards across this point and I can use the trackbar back and forth across it.

If you want to see it, I may have to snail mail it to you. The file is 773MB. I'll have to break out the 800MB CDs.

If I can cut the file and reproduce the problem I'll let you know.

[len0x]

Regarding scrolling backwards - I found that if you open avs with just mpeg2source() in VDubMod and go to the last frame and then try to step frame-by-frame back: everything hangs after a couple of frames... I'll do some more tests to see if that's the problem with decode of last frames of a broken source or generic one.

[Guest]

@Cyberia and len0x

I'll of course be happy to look at any problem reports, as long as I can duplicate them. So, you know what is needed: either post a link to the clip or snail mail it to me. I'll provide my mailing address by PM upon request.

[eb]

I was tested on several more then 5GB raw tvsat records, together with this famous Australia_1.tp file. No problems.

DGIndex is quick as the lightning, never met so speedy device.
I think that above mentioned files can be so called contuinity erreors.
My tests were done on DGIndex screen, not on the resulted file /proccesed/.

eb

[Guest]

I've made a version 1.0.10, as follows:

* Raw PID detection now shows the stream type where possible, i.e., you can now identify audio and video.

* Since I can now get the audio type from the raw detection, the limitation about audio demuxing in the absence of full PAT/PMT is removed. Previously, if the PAT/PMT was absent, only MPEG audio could be demuxed. Now, AC3 audio can be demuxed also. This allows you to get the audio from the FaithHill10.ts stream, for example.

http://neuron2.net/fixd2v/dgmpgdec1010.zip

[eb]

neuron2 wrote:
I've made a version 1.0.10

All streams were revealed on my samples.
I think that detection should be limited to the first 5 or 50 MB.
For files more then 5GB we have here wasted time.

Why when analizing .vob files there is hint that it is not transport stream?
Program working nice with .vobs.
As a example .vob from my website in folder Gladia with DTS+2 another audio streams were demuxed exelently to the separated files, even DTS file with .dts extension
Thank you Neuron2,

eb

[Guest]

Originally posted by eb
All streams were revealed on my samples. Good. Thank you for the feedback.

I think that detection should be limited to the first 5 or 50 MB. For files more then 5GB we have here wasted time. Good point. I'll address that in the next release.

Why when analizing .vob files there is hint that it is not transport stream? A VOB is not a transport stream and does not have PIDs, so PID detection does not arise. Are you asking for some new analysis feature for program streams?

BTW, I've decided to add "Set as Audio PID" and "Set as Video PID" buttons to the detect dialogs. You'll select the desired PID line and then punch the button to set that PID.

[eb]

For the .vob /or any others/ my proposal is to change this hint to:

IN THIS FILE YOU have such and such streams/or whatever else.
It is clear that program is dealing with this very well, looking on results with demuxed audios.

eb

EDITED:
I am slowly but gently trying to push you to subtitles problem

[Cyberia]

Originally posted by Cyberia
@neuron2 - I have found an SVCD that DGIndex will not let me step backwards in at a certain point in the film. I can step forwards across this point and I can use the trackbar back and forth across it.

If you want to see it, I may have to snail mail it to you. The file is 773MB. I'll have to break out the 800MB CDs.

If I can cut the file and reproduce the problem I'll let you know.

This is strange. I can't cut the file down. Both TMPGenc and VobCutter hang when I try to cut it. However, DGIndex builds a project file sucessfully and the d2v file looks fine.

LATER: The MPG is corrupt. VCDGear can fix MPEG errors and the fixed file works fine in every way. If you still want to see the file I can snail it to you.

[Cyberia]

I am slowly but gently trying to push you to subtitles problem

Ohhhh. Subtitle demuxing is a great idea!

[Longinus]

Since now DGMPGDec can open TS, and I never played with this before, I decided to give it a try... But I didn't work.. :rolleyes:

It's an Enterprise chapter, of about 4gb..
It plays fine in Media Player Classic, and I can demux the mpeg with HDTVtoMPEG2.

But when I try to open it with DGMPGDec, I get a problem..
I detect the PID as 0x1022 and 0x1023... I try to set the PID as 1022 and 1023 (Or should I use 22 and 23? I tried both ways, same problem). But after I hit ok, DGMPGDec freezes.... consuming 0% processor and just stay dead forever..

The audio is AC3, but it should work with version 1.0.10, right?
If requested, I'll submit a small piece (very small, because of the size) of the stream....

(EDIT) Never mind.. I got it to work. I processed the orignal file with HDTVtoMPEG2 set to re-create the Transport Stream... Then DGMPGDec opened it without problems.. But it said that 'Opening GOP is not closed'.... probably this caused the problem in the first place???

[len0x]

First that scrolling back problem. Sample to work with - P7.TS from eb. How to reproduce:
- create d2v file and open it avs via mpeg2source
- open avs in vdubmod
- press "go to the last frame" button
- three times press "go to the previous frame" button.
- hang indefinitely :)

Second issue. Not sure if that source problem or DGDecode.dll. But anyway - sample named EURO1080_2004-06-03_19-16-25.MPG from eb (if you don't have it I will make it available for you on Monday).
Quite a bit of frames are decoded with garbage in them. First frame like that is 57. When playing in software DVD players I don't see any artifacts and I don't know what else can I use to see the contents of mpeg2 file without frameserving.

P.S. using version 1.0.10.

P.P.S Btw, raw detection now looks really nice! Now all I really need is make use of that detection via CLI: I suggest application called DGdetect which does the same as DGTable , but for raw mode (or may be you can combine them into one).

[Zep]

Originally posted by neuron2

"I think that detection should be limited to the first 5 or 50 MB."
"For files more then 5GB we have here wasted time."


Good point. I'll address that in the next release.


at least 50 megs please. I have some streams that even took OVER 50
but it is rare so even 100 megs may be better. Actually now
that i think about it. Can you add that value in the ini
file? Advanced users could change the amount to suit their needs.

thanks :)

[eb]

,
All my samples are in the state " as recorded", without any cuttings off the ends, that is why DGIndex is making WARNING becase of opened GOPs. I think that the similiar warning about the end will be useful.
But in practice everyone is cutting off /orshould at least/ on the main frames bondary.

zep wrote:
even 100 megs may be better
no protest, it is only few seconds

eb
EDITED: sorry for this ,but the best will be if DGInex will cut off begining of the audio and video,such way that later no DELAY corrections will be needed

[jarthel]

the new d2v format isn't readable by reauthorist. right?

[Trahald]

Originally posted by jarthel
the new d2v format isn't readable by reauthorist. right?

no, it isnt.. i dont think ra will be updated to the new format (since development seems to be stopped on it)

[Guest]

@Cyberia

No need to send it to me.

@Longinus

No, that GOP warning is harmless. If you'd like to provide a small clip, I'd be interested. What does that "Re-Create" function actually do?

@len0x

I found and fixed that hanging issue. I also limited the raw detection to the first 600,000 transport packets (113 Megbytes). Regarding the garbage frames issue, please make the stream available to me ASAP. Maybe eb can get it to me before Monday?

I'll make a release in just a few minutes.

[Guest]

Version 1.0.11 fixes Tony's hanging problem and limits the amount of file read for raw PID detection to 113Mbytes:

http://neuron2.net/fixd2v/dgmpgdec1011.zip

[len0x]

Originally posted by neuron2
I found and fixed that hanging issue.

Great, confirmed - problem's gone. Thanks!

Originally posted by neuron2

Regarding the garbage frames issue, please make the stream available to me ASAP. Maybe eb can get it to me before Monday?


I managed to open the source in Mpeg2Shnitt and saw the same problems there - so I guess its just broken source and DGDecode is doing fine now. Sorry about false alarm :D

[tedgo]

I have a little problem with DGIndex since version 1.0.9:
In every project i created since 1.0.9 there is a single frame out of 50 frames near to end, that will repeat as about 5th frame before the end of the stream. It shows up during playing or encoding the created avs-script in virtualdubmod and then of course in the encoded file. What can cause it?
This problem never occurs in previous versions.

[Guest]

@tedgo

Funny things can happen with transport streams that were not cut cleanly at the end. Is that what you have? If not, please make the stream available for analysis.

[tedgo]

The streams are all VOB's. And it happened with every VOB!
Try it out and encode a trailer with credits at the end for example and you will see.
There will be a single frame that repeats as about 5th frame near to end.

EDIT: Returned to version 1.0.8 solved the problem
EDIT AGAIN: No, it doesn't. The same problem. Strange...

(Sorry, for my very bad english, but my last lesson's 20 years ago)

[eb]

@ neuron2,
http://neuron2.net/fixd2v/dgmpgdec1011.zip

In sample e.7 /with 7 audios/ only 1 one /selected/ is demuxed , besides that DEMUX ALL was checked.

eb

[Guest]

@tedgo

I'll look into it and see if I can duplicate it. Thanks for bringing it to our attention.

@eb

Because only one audio PID can be set at a time, only one audio stream can be demuxed. I can consider allowing multiple audio PID specification.

[eb]

Originally posted by neuron2

Because only one audio PID can be set at a time, only one audio stream can be demuxed. I can consider allowing multiple audio PID specification.

It will be nice to have such possibility.
As for today more and more movies/education programs from satTV are transmited with multi audio and multi subtitles, my target is to be able to prepare from this satv records the CDs and DVDs having all this data.

Why? It is easiest and most pleasure way to learn foreign languages, listening and reading words and sentences not forced by any teacher.

eb

[vispgraedde]

Originally posted by tedgo
The streams are all VOB's. And it happened with every VOB!
Try it out and encode a trailer with credits at the end for example and you will see.
There will be a single frame that repeats as about 5th frame near to end.

EDIT: Returned to version 1.0.8 solved the problem
EDIT AGAIN: No, it doesn't. The same problem. Strange...

(Sorry, for my very bad english, but my last lesson's 20 years ago)

I have seen the same thing. 1.0.9 used, but at the time I thought it was my avs script that bugged out.

Backtracking the versions, 1.0.7 is the earliest that I can reproduce it with.
Always the same frame (with the same clip) (frame 6922 (from vdub timeline) out of 6930 frames of this clip)
I can't reproduce it with 1.0.6 and earlier.

[tedgo]

I made some tests now.
It only seems to occur when the VOB's are demuxed to *.M2V before creating a d2v-project with DGIndex (Sorry, if my previous postings were mistakable). Loading the VOB's directly in DGIndex seems to solve the problem, so it may be a fault by the program i demuxed the VOB's.
But with a *.M2V-Files there is still an "error" frame in the last 5 frames. Sometimes a copy of a frame or a totally green frame.

[Guest]

OK, guys, I am going to thoroughly examine the end of file handling. MPEG is a bit funny because you are always outputting a frame behind what you are decoding, so when you ask for the last frame, there's no next frame to decode. So that funniness has to be handled right. And that last encoded frame may not be last in display order, so a bad frame could occur within several frames of the end if this is not handled right. As I said, I'll analyze it thoroughly and get it sorted out.

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

[eb]

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

You are, like Hichcok , presenting final solution but keeping us in tension up to the last usolved frame.
eb

[tedgo]

I made some more tests (i'm very busy ;) ) and it's not caused by the programs i demuxed the vob's before creating a d2v-project. Some of the vob's i've directly loaded in DGIndex had the same problem, so it must be a little fault in DGIndex.
The vob's and m2v's i tried are definitely ok, cause the problem doesn't occur, when i load them directly in vdubmod.

[eb]

@neuron2 & lenOx,

if from your tests/problems some conclusions are resulting that should be resolved in the recording process, then this hints I can present to authors of different recording programs

eb

[Cyberia]

Originally posted by neuron2
I can consider allowing multiple audio PID specification.

May I make a suggestion? Do the same thing you do in the Audio menu. You can demux a specific audio PID or you can demux all of them.