Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > DVD2AVI / DGIndex

Reply
 
Thread Tools Search this Thread Display Modes
Old 20th August 2010, 15:59   #2421  |  Link
krosswindz
Registered User
 
Join Date: Feb 2006
Location: kernel space
Posts: 245
Out of curiosity does any one have a Parallels Workstation Extreme and can try DGNV tools in the parallels VM. AFAIK Parallels Workstation Extreme should be able to see Nvidia GPU and should support CUDA as well. I dont have a copy of it, it would be interesting to see if some one could try to use DGNV tools in it.
krosswindz is offline   Reply With Quote
Old 20th August 2010, 22:34   #2422  |  Link
LeXXuz
21 years and counting...
 
LeXXuz's Avatar
 
Join Date: Oct 2002
Location: Germany
Posts: 716
Quote:
Originally Posted by Sharktooth View Post
that's it. you can use RD while frameserving but you cant start frameserving within a remote desktop session.
Makes sense. I will use a VNC session for starting frameserving jobs. Thanks Sharktooth.

I have another problem with DAR of source files. I don't know if it is a problem of DGDecNV or Megui. But since it never occured with DGindex, FFMSindex nor DGAVCdec I guess it might be a problem with DGDecNV.

I'm transcoding a couple of older mpeg2 DVB captures and it seems that Megui does not recognize the correct DAR when DGdecNV is used. F.e. a source mpeg2 file (just the video stream as mpv, no program stream) PAL 720x576, correct DAR is ITU 4:3 (1:1.367521) for this file, but when serving it with DGdecNV instead of DGindex the DAR is simply set to 1,25 (720/576).

Also opening the .dgi file in Meguis Avisynth script creator takes VERY long with mpeg2 source files. Looking at hdd activity it seems that the whole source file is scanned somehow before the preview window is shown and the script generator can be used.
If this is a megui issue, I apologize and will post it in the megui-thread.

Last edited by LeXXuz; 21st August 2010 at 10:49.
LeXXuz is offline   Reply With Quote
Old 20th August 2010, 22:45   #2423  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Quote:
Originally Posted by LeXXuz View Post
I have another problem with PAR of source files. I don't know if it is a problem of DGDecNV or Megui. But since it never occured with DGindex, FFMSindex nor DGAVCdec I guess it might be a problem with DGDecNV.
Can you please post a link to a sample that does this?

Quote:
Also opening the .dgi file in Meguis Avisynth script creator takes VERY long with mpeg2 source files. Looking at hdd activity it seems that the whole source file is scanned somehow before the preview window is shown and the script generator can be used.
If this is a megui issue, I apologize and will post it in the megui-thread.
Does it open quickly in bare DGIndexNV?
Guest is offline   Reply With Quote
Old 20th August 2010, 23:53   #2424  |  Link
LeXXuz
21 years and counting...
 
LeXXuz's Avatar
 
Join Date: Oct 2002
Location: Germany
Posts: 716
Quote:
Originally Posted by neuron2 View Post
Can you please post a link to a sample that does this?
Here it is.

Sequence header look fine to me, DGIndexNV also seems to get correct ar:

MeGui's Avisynth script creator shows DAR 1:1,2222 (704/576) with DGIndexNV and ITU 4:3 PAL with DGIndex as it should.

Quote:
Originally Posted by neuron2 View Post
Does it open quickly in bare DGIndexNV?
When I drag&drop the .dgi file to DGIndexNV the related .m2v loads almost instantly. So yes.

Last edited by LeXXuz; 21st August 2010 at 10:48.
LeXXuz is offline   Reply With Quote
Old 21st August 2010, 09:05   #2425  |  Link
Sharc
Registered User
 
Join Date: May 2006
Posts: 3,997
Quote:
Originally Posted by LeXXuz View Post
Here it is.

Sequence header look fine to me, DGIndexNV also seems to get correct ar:

MeGui's Avisynth script creator shows 1:1,2222 (704/576) with DGIndexNV and ITU 4:3 PAL with DGIndex as it should.
PAR of your 704x576 picture is 12:11 (1.0909) as one can estimate from the circle distortion of the ARD logo. DAR (display aspect ratio) is 4:3. So when you playback the file at 4:3 it will be resized correctly (undistorted). So everything is fine I would say.
Sharc is offline   Reply With Quote
Old 21st August 2010, 10:01   #2426  |  Link
LeXXuz
21 years and counting...
 
LeXXuz's Avatar
 
Join Date: Oct 2002
Location: Germany
Posts: 716
I think it's not. MeGUI sets the wrong SAR to x264. The aspect ratio of the output is slightly too small and the picture looks slightly narrowed in width. With DGIndex everything is fine.

EDIT: It is the DAR Megui is having trouble with. Sorry for any confusion.

Just opened a 16:9 anamorphic encode and MeGUI sets DAR to 1,25 instead of ITU 16:9 PAL. As a result, x264 encoder is fed with SAR setting of 5:4 instead of 31:17 as it should.

Last edited by LeXXuz; 21st August 2010 at 10:47.
LeXXuz is offline   Reply With Quote
Old 21st August 2010, 11:05   #2427  |  Link
Sharc
Registered User
 
Join Date: May 2006
Posts: 3,997
Quote:
Originally Posted by LeXXuz View Post
I think it's not. MeGUI sets the wrong SAR to x264. The aspect ratio of the output is slightly too small and the picture looks slightly narrowed in width. With DGIndex everything is fine.
Your sample is mpeg2, not AVC. The PAR is not explicitely signalled in the stream. So you might have to add --sar 12:11 in the x264 CLI when you re-encode it with x264.
Nevertheless your sample should display undistorted (here it does) when you play it back at forced 4:3, e.g. with VLC.
When I play it back with DGindex the ARD logo looks slightly squeezed (narrowed) as expected for a 12:11 or 128:117 PAR (ITU).
Possibly your playback device assumes a par of 16:15 (generic PAR for PAL 4:3), means it will stretch your picture by 6.7% rather than about 9%. Hence a slight (but almost unnoticeable) distortion may persist.

Just a note: In case you intend to author the clip to blu-ray format at some later stage you would have to add 8 pixels black borders to the right and left each, such that the picture width becomes 720.
Sharc is offline   Reply With Quote
Old 21st August 2010, 11:35   #2428  |  Link
LeXXuz
21 years and counting...
 
LeXXuz's Avatar
 
Join Date: Oct 2002
Location: Germany
Posts: 716
Quote:
Originally Posted by Sharc View Post
Your sample is mpeg2, not AVC. The PAR is not explicitely signalled in the stream.
But DAR should. The question is why does MeGUI set the correct DAR with DGIndex (I guess it gets this information from the sequence headers of the mpeg stream) while it simply sets hight/width as DAR when using DGIndexNV, which is not correct.

In short, I think MeGui should get the same results (and for some reason it does not), no matter what indexing method is used, no?
LeXXuz is offline   Reply With Quote
Old 21st August 2010, 11:40   #2429  |  Link
Sharc
Registered User
 
Join Date: May 2006
Posts: 3,997
^^^^
I can't comment on MeGui as I am not very familiar with it.
Maybe someone else .....
Sharc is offline   Reply With Quote
Old 21st August 2010, 11:49   #2430  |  Link
stax76
Registered User
 
stax76's Avatar
 
Join Date: Jun 2002
Location: On thin ice
Posts: 6,837
This is clearly a MeGUI and not a DGDecNV problem. Detecting the correct PAR for every possible source and source filter is very difficult, remember there are countless possible source formats and numerous possible source filters. I've changed and tweaked my detection probably more then 100 times in the last 8 years.
stax76 is offline   Reply With Quote
Old 21st August 2010, 11:51   #2431  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
I could pass hints if it's thought to be a significant enough issue.
Guest is offline   Reply With Quote
Old 21st August 2010, 12:11   #2432  |  Link
Groucho2004
 
Join Date: Mar 2006
Location: Barcelona
Posts: 5,034
Quote:
Originally Posted by LeXXuz View Post
The question is why does MeGUI set the correct DAR with DGIndex (I guess it gets this information from the sequence headers of the mpeg stream) while it simply sets hight/width as DAR when using DGIndexNV, which is not correct.
I just looked at megui's source code - it reads the d2v file that DGIndex creates and gets the value from the entry "Aspect_Ratio".

DGI files (from DGDecNV) do not provide this property.
Groucho2004 is offline   Reply With Quote
Old 21st August 2010, 12:17   #2433  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Quote:
Originally Posted by hydra3333 View Post
Sure.
Please remove the SetMTMode() and report results.
Guest is offline   Reply With Quote
Old 21st August 2010, 12:17   #2434  |  Link
stax76
Registered User
 
stax76's Avatar
 
Join Date: Jun 2002
Location: On thin ice
Posts: 6,837
@Groucho2004

That's not robust anyway because 16:9 is sometimes defined in IFO only but not in VOB.

Code:
            If Filepath.GetExt(p.SourceFile) = ".d2v" Then
                Dim content = File.ReadAllText(p.SourceFile)

                If content.Contains("Aspect_Ratio=16:9") Then
                    p.SourceAnamorphic = True
                Else
                    Dim ifoFile = GetIfoFile()

                    If File.Exists(ifoFile) Then
                        Dim dar2 = MediaInfo.GetVideo(ifoFile, "DisplayAspectRatio")

                        If dar2 = "1.778" Then
                            p.SourceAnamorphic = True
                        End If
                    End If
                End If

Last edited by stax76; 21st August 2010 at 12:23.
stax76 is offline   Reply With Quote
Old 21st August 2010, 12:22   #2435  |  Link
Groucho2004
 
Join Date: Mar 2006
Location: Barcelona
Posts: 5,034
Quote:
Originally Posted by stax76 View Post
@Groucho2004

That's not robust anyway because 16:9 is sometimes defined in IFO only but not in VOB.
I know, I just answered a question.
Groucho2004 is offline   Reply With Quote
Old 21st August 2010, 12:30   #2436  |  Link
Groucho2004
 
Join Date: Mar 2006
Location: Barcelona
Posts: 5,034
Quote:
Originally Posted by neuron2 View Post
I could pass hints if it's thought to be a significant enough issue.
Actually, that would be a nice feature.
Groucho2004 is offline   Reply With Quote
Old 21st August 2010, 12:39   #2437  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Thinking more about it...

Hints are needed only for things that can change on a per-frame basis. I should just add the aspect ratio information to the DGI file.
Guest is offline   Reply With Quote
Old 21st August 2010, 13:21   #2438  |  Link
hydra3333
Registered User
 
Join Date: Oct 2009
Location: crow-land
Posts: 540
Quote:
Originally Posted by neuron2 View Post
Please remove the SetMTMode() and report results.
Commented it out like this
Code:
#SetMTmode(mode=5,threads=1) # start with mode=5 forAVIsource http://forum.doom9.org/showthread.php?p=1067216#post1067216

and it works fine now.
I'd thought mode=5 was supposed to be the most "compatible" however leaving it off seems to be even more so Hurrah !

Out of interest, is that the end solution or a workaround ?

Last edited by hydra3333; 21st August 2010 at 13:27.
hydra3333 is offline   Reply With Quote
Old 21st August 2010, 14:52   #2439  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
You can't multithread the NV-based DGSource(). I don't know for sure the technical reason or if there would be some way to allow it. I believe that the CUDA context must be associated with a single thread. Whether floating contexts changes anything I simply don't know at this time and haven't had time to look into it. So for now, just don't multithread DGSource.

It's probably a bad idea in general to multithread any source filter.

Last edited by Guest; 21st August 2010 at 14:55.
Guest is offline   Reply With Quote
Old 22nd August 2010, 01:47   #2440  |  Link
hydra3333
Registered User
 
Join Date: Oct 2009
Location: crow-land
Posts: 540
ok yes, someone recommended mode=5 for source filters however looks like not this one.

if multi-threading is "out" when using NV then multi-core boxes like my 4-core effectively become single-core boxes inside avisynth... hence it must be swings and roundabouts for overall speed.

in this specific case there is no problem, since not much is being done and x264 multi-threads anyway - however if there's a lot of processing to be done inside avisynth then NV's single-threading requirement may end being a "chokepoint" for overall performance.

unless there's a way around it I don't know of (i'd thought to use MT, the 1st line of a script had to contain a setMTmode).

oh well, at least we know. thanks.

Last edited by hydra3333; 22nd August 2010 at 13:03.
hydra3333 is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 14:32.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.