DGMPGDec 1.5.3 Final

[burfadel]

Seems to be fine now! I don't know why I had to delete it twice, but I can't seem to recreate the problem...?!

[Kurtnoise]

Tiny request: replace the [ and ] in the command line stuff by something more usable (like " " or remove it completely). Not for me really but somebody report a bug for MeGUI. However, it was a path name issue (There was already a [....]).

edit: forgot to check the doc...there is already a command for that. Sorry.

[Guest]

* An option was added to control how HD videos are displayed. You can now shrink by half (the previous behavior) or you can view any quadrant of the video. This allows the interlace structure of HD videos to be inspected.

* The font size of the Info dialog was reduced so that it all stays visible with the new vertical arrangement.

http://neuron2.net/dgmpgdec/dgmpgdec150b9.zip

The INI file was changed again, so open DGIndex and then close it the first time you use it.

[Guest]

Sorry about the quick update, but the fix for emulated audio headers was rejecting some valid headers. I'm pretty sure everything is fully correct now.

http://neuron2.net/dgmpgdec/dgmpgdec150b10.zip

[Guest]

we'll have feedback soon, since i added this version to MeGUI autoupdate... ;) You better get beta 10, because 7-9 may reject valid MPEG audio in program streams.

[Guest]

Please re-download beta 10. A bad build was uploaded.

[burfadel]

You better get beta 10, because 7-9 may reject valid MPEG audio in program streams.

ah ok! that explains the problem I had earlier :)

[Sharktooth]

You better get beta 10, because 7-9 may reject valid MPEG audio in program streams.
ok...

[zurv]

on beta 10, 00011.m2ts for spiderman3 (which plays fine in zoom (+cccp))
but, DGindex says "no video sequence header found!"

It also works in DSmux (Haali), but not mkvmerge. *shrug*

ideas?

thanks

[Guest]

Is it MPEG2 video?

[zurv]

i was about to edit the post.. i assumed it was because the first track worked and was mpeg2.. but i just looked again and the movie itself is avc - didn't expect to see them mix formats :)

hrmm.. now what to do with avc :)
at worst directshowsource would work.. but that feels dirty :)

I guess demuxing the audio would be the big thing...hrmmm

[Guest]

You can try DGAVCDec for the AVC portion.

[burfadel]

You can try DGAVCDec for the AVC portion.

Are you planning to merge Dgmpgdec and DGavcdec in the future? that would resolve people using the wrong one and being confused with mixed tracks etc :) or even just the need to use two different programmes on mixed tracks!

Just thought it would streamline things a bit :), seeing the programme is of similar nature. I know it kind of contradicts the Dgmpegdec name, but technically avc is mpeg4, so the name still works!

[Guest]

See posts 91 and 93 of this thread.

[jeffy]

neuron2, may I please ask, does this really dering (quant=0)?
BlindPP(quant=0,cpu=0,cpu2="ooooxx")
http://forum.doom9.org/showthread.php?p=675252#post675252

Thank you.

[Guest]

I don't know because I didn't write that filter. Generate a clip with and without it in your script and then do a Subtract() to see if there is a difference. You may need Levels() after Subtract to magnify the differences. I use the Levels filter in VirtualDub as it is easier.

[GR]

Thx neuron2
beside that would be wonderful if dgindex could let me to set the begin & end not only on the key frame but also on each frame -v-
I need to clear some tvrip's advertisement(.ts or .m2ts) completely for save certain fullest scene
it just a problem now - -b

[Guest]

* Fixed a problem that caused DGIndex to miss some audio streams. For example, if AC3 substream 0x80 was present, and MPA stream 0xC0 was present, both would be viewed as Track 1, and only the first one encountered would be detected. The concept of track number is removed (as it is a DVD-specific construct), and it is replaced with the concept of audio id. See the users manual for details.

* For MPEG audio streams, the Info dialog now shows full information: audio id, layer, number of channels, sampling rate, and bitrate. This information is also included in the filename of demuxed MPEG audio streams.

http://neuron2.net/dgmpgdec/dgmpgdec150b11.zip

This was a major redesign of audio processing so please beat up on anything audio related and try to break it.

To avoid possible hassles, delete your DGIndex.ini file before running this new version.

[Sharc]

DGIndex returns BT.709 as colorimetry info for unflagged mpeg2 streams. Is there a reason for this? Shouldn't it better return "unspecified" - or similar - for unflagged streams?
Sorry if this has been already discussed.

[kumi]

On the point that Sharc raises, could I request that you allow users to override what colorimetry is assumed by DGindex (and passed via hints) when the sequence display extension fields are absent? At present, when fed SD MPEG-2 input, DGindex assumes Rec.709, but Hank315 believes that the default SD MPEG-2 matrix might in fact be Rec.601. See here (http://forum.doom9.org/showthread.php?p=1061536#post1061536).


Thanks for the new beta :)

[buzzqw]

@neuron2

do you know any affidable command line application able to extract audio track pid/id ?

i used bbsummary, but this utility not extract 80,81,C0 info ...

on command line i usually specify -TN=1,2 for extracting the first and second tracks... but with 150b11 dgindex fails extracting tracks even if i specify 80,81...

this is the old command line
c:\dgindex.exe -FO=0 -OM=1 -TN=1 -YR=2 -AIF=[c:\_aaa.vob] -OF=[C:\movie] -exit -minimize

and this is the vob ( 7.81 mb) http://www.64k.it/andres/data/a/_aaa.vob

how i can translate (or detect TN values) in new command line ?

thanks!

BHH

[G_M_C]

On the point that Sharc raises, could I request that you allow users to override what colorimetry is assumed by DGindex (and passed via hints) when the sequence display extension fields are absent? At present, when fed SD MPEG-2 input, DGindex assumes Rec.709, but Hank315 believes that the default SD MPEG-2 matrix might in fact be Rec.601. See here (http://forum.doom9.org/showthread.php?p=1061536#post1061536).


Thanks for the new beta :)

If i remenber correctly the MPEG2 has Rec.709 as the standard (default) colormetry, as part of the MPEG2-standards. So the fact that DGIndex gives 709 as defeult, is a simple (and correct) consequece of the MPEG2 standard.

If your unflagged stream is in fact Rec.601, than basically, the stream isn't muxed/made properly (is should have been flagged !).

[Guest]

@how i can translate (or detect TN values) in new command line? The audio id's work like this:

PVA and transport: one track always 0

all others:
AC3: 80-87
DTS: 88-8f
PCM: a0-a7
MPA: c0-c7

You could ask for the first two of each category:

TN=80,81,88,89,a0,a1,c0,c1

Or you could just demux them all.

The problem is that the concept of track number is DVD specific but DGMPGDec wants to handle all stream types. You might suggest to number them by the order they are encountered in the source. But that's indeterministic because if you start the project at different points you can change the order in which the streams are encountered.

But obviously I want to support third party applications, so can you take the above information into account and tell me what you need to support your application? Maybe I could make a command line option that just creates the audio stream list (as displayed in the Info dialog) and then exits.

[buzzqw]

Maybe I could make a command line option that just creates the audio stream list (as displayed in the Info dialog) and then exits

that's would be enough!

when i know the audio id (80,81,88,89,a0,a1,c0,c1) is very easy to read a "info.dump" created by dgindex

i my suggest, if possibile to include even more info in this "info" file (maybe all those in info panel, if isn't needed all file parsing)

thanks Neuron2!

BHH

[Guest]

Please redownload beta 11. There was a bug in the CLI operation. With the new one, this works for your sample VOB:

dgindex.exe -FO=0 -OM=1 -TN=80,c1 -YR=2 -AIF=[_aaa.vob] -OF=[movie] -exit -minimize

The problem with the log idea is how far do I parse into the file looking for audio streams? I'll do some experiments with some reasonable size.

Why don't you demux them all and parse the resulting file names? :)

[buzzqw]

The problem with the log idea is how far do I parse into the file looking for audio streams? I'll do some experiments with some reasonable size.

in bbsummary i analyze the first 10000 frames, and it's a matter of two seconds

Why don't you demux them all and parse the resulting file names?

well.. i must admit that wont be a problem... just to search for 80 , 81 , c1 in file names...

BHH

[Guest]

Maybe we can map the bbsummary info to audio_ids. Can you post typical bbsummary output for the audio?

[buzzqw]

bbsummary on _aaa.vob (same sample posted before)

bbSummary - based on BBINFO version 1.9

File C:\Programmi\PureBasic402\AutoMKV\test\_aaa.vob is an MPEG-2 Program Stream

Summary:

MPEG Packs = 4000
System headers = 22
Private AC3 Audio Stream 1 packets = 296, total bytes = 597029
AC3 Audio Stream 0 packets = 296, total bytes = 0
Padding Stream packets = 21, total bytes = 12383
Private Stream 2 packets = 44, total bytes = 43956
MPEG Audio Stream 1 packets = 98, total bytes = 197957
Video Stream 0 packets = 3584, total bytes = 7244864

on a TS files
bbSummary - based on BBINFO version 1.9

File C:\Programmi\PureBasic402\AutoMKV\test\_aaa.ts is an MPEG-2 Transport Stream

Summary:

MPEG Transport Packets = 10000
PID 0000, Program Association Table packets = 3, total bytes = 552
PID 001F, Other packets = 1, total bytes = 184
PID 0113, Other packets = 9, total bytes = 1656
PID 0991, Video Stream 0 packets = 8548, total bytes = 1525328
PID 0992, MPEG Audio Stream 0 packets = 719, total bytes = 129930
PID 0993, MPEG Audio Stream 0 packets = 719, total bytes = 129930


other TS file
bbSummary - based on BBINFO version 1.9

File C:\Programmi\PureBasic402\AutoMKV\test\ateme.luxe.tv.hd.eutelsat.w3a.12.jun.2007.whole.ts.sample.ts is an MPEG-2 Transport Stream

Summary:

MPEG Transport Packets = 10000
PID 0000, Program Association Table packets = 25, total bytes = 4600
PID 0011, Other packets = 2, total bytes = 368
PID 0012, Other packets = 15, total bytes = 2760
PID 0020, Other packets = 24, total bytes = 4416
PID 0065, Video Stream 0 packets = 9391, total bytes = 1720842
PID 0100, MPEG Audio Stream 0 packets = 271, total bytes = 49204
PID 0101, MPEG Audio Stream 0 packets = 271, total bytes = 49204

on other vob file


bbSummary - based on BBINFO version 1.9

File C:\Programmi\PureBasic402\AutoMKV\test\_telecine.vob is an MPEG-2 Program Stream

Summary:

MPEG Packs = 10000
System headers = 34
Private AC3 Audio Stream 1 packets = 1855, total bytes = 3740719
AC3 Audio Stream 0 packets = 415, total bytes = 0
AC3 Audio Stream 1 packets = 415, total bytes = 0
AC3 Audio Stream 3 packets = 209, total bytes = 0
DTS Audio Stream 2 packets = 816, total bytes = 0
Padding Stream packets = 34, total bytes = 39246
Private Stream 2 packets = 68, total bytes = 67932
Video Stream 0 packets = 8110, total bytes = 16382960

last vob

bbSummary - based on BBINFO version 1.9

File C:\Programmi\PureBasic402\AutoMKV\test\movimentato.vob is an MPEG-2 Program Stream

Summary:

MPEG Packs = 4840
System headers = 22
Private AC3 Audio Stream 1 packets = 273, total bytes = 548625
AC3 Audio Stream 0 packets = 273, total bytes = 0
Padding Stream packets = 24, total bytes = 29997
Private Stream 2 packets = 44, total bytes = 43956
Video Stream 0 packets = 4544, total bytes = 9175284

BHH

[Guest]

I'll implement the log file, as there is not enough information there to map to audio id.

[Sharc]

If i remenber correctly the MPEG2 has Rec.709 as the standard (default) colormetry, as part of the MPEG2-standards. So the fact that DGIndex gives 709 as defeult, is a simple (and correct) consequece of the MPEG2 standard.

If your unflagged stream is in fact Rec.601, than basically, the stream isn't muxed/made properly (is should have been flagged !).
The background is that there were quite some controversial discussions on standards and on what (unflagged) DVDs (using mpeg2) comply with, and what (most) players do expect, in this thread:
http://forum.doom9.org/showthread.php?p=1061205#post1061205.
I don't know what are opinions and what are facts, and how much room for the interpretation of standards is left to the studios and player/chipset manufacturers. Still it appears that many SD players expect 601.

[kumi]

If your unflagged stream is in fact Rec.601, than basically, the stream isn't muxed/made properly (is should have been flagged !).
Exactly, that's what I thought, too. But there is some good evidence that Rec.601 IS the default, and not Rec.709. And the fact that many (most?) SD players assume 601 is icing on the cake.

[Guest]

The later spec says that if it is unspecified then the colorimetry used is "determined by the application". I could just report "unspecified" but I have to use something for RGB upconversion. Do you want it as an option?

[Guest]

I have the log file working and will release a beta tonight.

Whenever the Info dialog is closed, either manually or automatically by DGIndex, the Info dialog contents are written to a file. For CLI operation, if the new -preview parameter is given, the first 100 frames will be decoded and then the Info dialog will be killed, causing the log file to be generated. Thus, the way to get a silent dump of the data is this:

dgindex -AIF=[zdf.mpeg] -OF=[zdf] -preview -minimize -exit

The output file looks like this:

Stream Type: MPEG2 Program
Profile: main@main
Frame Size: 720x576
Aspect Ratio: 4:3
Frame Rate: 25.000000 fps
Video Type: PAL
Frame Type: Interlaced
Colorimetry: BT.709
Frame Structure: Frame
Field Order: Top
Coded Number: 99
Playback Number: 99
Frame Repeats: 0
Field Repeats: 0
VOB ID:
Cell ID:
Bitrate: 5.016 Mbps
Bitrate (Avg): 5.738 Mbps
Audio Stream: 80: AC3 2/0 384
Audio Stream: c0: MPA L2 2ch 48 256
Audio Stream: c1: MPA L2 mono 48 128
Timestamp: 0:00:05
Elapsed: 0:00:03
Remain: 0:00:10
FPS: 24.96
Info:

[buzzqw]

Excellent!

thanks Neuron2!

BHH

[kumi]

but I have to use something for RGB upconversion. Do you want it as an option?
Yes, but only if it's possible to get the user's choice of default matrix passed into the stream hints. That's the reason for my request, really.

Currently, the 2nd line of this script is a no-op:
mpeg2source("foo.d2v", info=3) #foo.m2v is a MPEG2 SD clip w/ no sequence display extension fields
colormatrix(mode="Rec.601->Rec.709", hints=true)

I wish DGdecode would hint my source with Rec.601 coefficients. Can that be done? Thanks for listening :)

[Sharc]

I second this.
The user's choice should only kick in when the source has no sequence display extension.

If the source comes with with sequence display extension I think it should override the user's choice.

[kumi]

If the source comes with with sequence display extension I think it should override the user's choice.
I agree.

[Guest]

* Whenever the Info dialog is closed, a log file is created that contains the same information. This may be invoked by the CLI to silently query the nature of the input file. Please see the users manual for details.

* When a stream does not declare the colorimetry, matrix_coefficients=1 is assumed for HD video and matrix_coefficients=5 is assumed for SD video.

http://neuron2.net/dgmpgdec/dgmpgdec150b12.zip

I'll wait and see who cries foul about the second bullet item. I'd prefer not to make an option for it.

[kumi]

Thanks a lot! Tested and working A-OK.

I wonder if you might also consider adding an "icing on the cake": how about some sort of indicator in the Info|Colorimetry area, to show when DGindex is applying it's own matrix coefficients? Perhaps something simple, like displaying *BT.470-2 B,G instead of BT.470-2 B,G.

[Boulder]

I'll wait and see who cries foul about the second bullet item. I'd prefer not to make an option for it.I'll try this later today when I get home from work. All my DVB captures show 'unknown' for colorimetry in DGIndex, but I must not use ColorMatrix(mode="rec.601->rec.709") when encoding with CCE or HC. If I use that line, the colors in the resulting video are different from the original video when looking in VDub. I'll need to check whether the difference exists when playing the video on the standalone player.

[buzzqw]

http://www.64k.it/andres/data/a/ateme.luxe.tv.hd.eutelsat.w3a.12.jun.2007.whole.ts.sample.ts (7 mb)

dgindex.exe -preview -OF=[aaa] -AIF=[C:\ateme.luxe.tv.hd.eutelsat.w3a.12.jun.2007
.whole.ts.sample.ts] -minimize -exit

in this case i must know the pid ? since dgindex prompt "No data. Check your PIDS"

... i think i will continue to use bbsummary to extract information and then use dgindex with right command line :o

thanks!

BHH

[G_M_C]

I'll try this later today when I get home from work. All my DVB captures show 'unknown' for colorimetry in DGIndex, but I must not use ColorMatrix(mode="rec.601->rec.709") when encoding with CCE or HC. If I use that line, the colors in the resulting video are different from the original video when looking in VDub. I'll need to check whether the difference exists when playing the video on the standalone player.

Hmmm it would seem that is allready in rec.709 than ?

A theoretical question though;
Is it possible to detect what colormetry is used, simply by scanning the stream ?

[SeeMoreDigital]

;)http://www.64k.it/andres/data/a/ateme.luxe.tv.hd.eutelsat.w3a.12.jun.2007.whole.ts.sample.ts (7 mb)

dgindex.exe -preview -OF=[aaa] -AIF=[C:\ateme.luxe.tv.hd.eutelsat.w3a.12.jun.2007
.whole.ts.sample.ts] -minimize -exit

in this case i must know the pid ? since dgindex prompt "No data. Check your PIDS"

... i think i will continue to use bbsummary to extract information and then use dgindex with right command line

thanks!

BHHHi Buzz,

The sample you provided contains MPEG-4 AVC video and not MPEG-2 video (along with 3No MP2 audio tracks) :o


Cheers

[buzzqw]

:stupid:

DOH!!!

thanks and sorry!!!

BHH

[Boulder]

Hmmm it would seem that is allready in rec.709 than ?Exactly. Which is why I'm not sure if unknown colorimetry really means Rec.601:confused:

[buzzqw]

this sample has problem

the "info" report only 1 audio track but pid detection reveal 2 audio
and dgindex is able to extract the second track

http://www.64k.it/andres/data/a/film_Ac3_5_2.ts (37mb)

is it a stream problem ?

BHH

[Guest]

@buzzqw

It's currently a limitation for transport streams that only the streams whose PIDs are set are processed. Item 16 of the Development List sticky:

"16) Process multiple audio streams at once for transport streams (as is currently done for program streams)."

As a workaround, you can use a CLI table parser to find the audio PIDs (I have one based on the PID detection code in DGIndex), and then retrieve info for each PID in turn.

[Guest]

I wonder if you might also consider adding an "icing on the cake": how about some sort of indicator in the Info|Colorimetry area, to show when DGindex is applying it's own matrix coefficients? Perhaps something simple, like displaying *BT.470-2 B,G instead of BT.470-2 B,G. Will do.

[buzzqw]

@Neuron2
As a workaround, you can use a CLI table parser to find the audio PIDs (I have one based on the PID detection code in DGIndex), and then retrieve info for each PID in turn.

i used bbsummary... :)

but if it possibile to add that cli capability to dgindex is even better (or you want to spare your cli app..)

thanks!

BHH

[Guest]

@buzzqw

I'll do something so that no other external CLI app is needed.

@kumi

Please redownload beta 12. I added your * idea (but at the end of the string).