ffdshow tryouts project: Discussion & Development

[Cheesegeezer]

What splitter are you using for mkv? Using Haali I had issues, using the MPC internal splitter they seem to work well for me.

Aha, I will try using the internals as I see Haali load up in the task bar.

Thanks for the advise.

[SamuriHL]

Definitely ATI driver supports 24 bit as confirmed with foobar2000 and J. River MC. This (http://www.avsforum.com/avs-vb/showthread.php?p=16305663#post16305663) is a bit old result with HD 4670, but the same is true for HD 5xxx. There might be a compatibility issue between MPC Audio Renderer and ATI HDMI driver.

That would mean reclock has the same issue. I'm not convinced yet.

[Andy o]

Realtek R2.35 driver:

- 16 bit integer: Works with correct sample rate, but channel count does not match (always the same as the one selected in Windows' Sound control panel).
- 24 bit integer, 32 bit integer, 32 bit fp: ffdshow does not connect to MPC Audio Renderer.

Then it doesn't seem like it's the same problem as with ReClock, cause ReClock can take in 24 and 32-bit, only that it can't output it with the ATI drivers. The MPC renderer it seems can't take in anything but 16-bit. This seems like it can be fixed fairly easily, no?

IOW, with the MPC audio renderer, higher-than-16 bit is failing before even reaching the driver.

[Andy o]

Does the MPC audio renderer only work with these betas of ffdshow? Cause I can't get it to connect with another ffdshow build (3126) even though I only enable 16-bit output.

[Skinleech]

Aha, I will try using the internals as I see Haali load up in the task bar.

Thanks for the advise.

Good luck! Haali doesn't see TrueHD stream either, so pretty useless for our purposes.

[albain]

If absolutely necessary I can grab my dev tools and throw it on that machine. It'll likely break TMT3 because of their stupid protection, but, if it helps fix this issue I can grab VS. Should I go for VS 2010? I'll definitely need some help with the setup of the project, though. I haven't compiled anything in VS in a while. The stuff I do for a living is all Java using Eclipse these days.

Well if nobody else can do it (is there ?), you can set up a development enviromnent but there is some work :

1/ VS 2008 (not 2010 it is crappy and so sloooow)
2/ DirectX SDK
3/ Grab last SVN release from bitstreamHD branch

[SamuriHL]

Well if nobody else can do it (is there ?), you can set up a development enviromnent but there is some work :

1/ VS 2008 (not 2010 it is crappy and so sloooow)
2/ DirectX SDK
3/ Grab last SVN release from bitstreamHD branch

Alright. I'll start getting that set up. It'll likely be sometime this weekend before I have everything set up. Also, I don't do anything with SVN, so, I will need assistance on how to get the latest SVN release code. (Where it's located, etc) But I've got experience debugging in VS, so, once I get it set up I should be able to help figure out what's going on.

[albain]

Then it doesn't seem like it's the same problem as with ReClock, cause ReClock can take in 24 and 32-bit, only that it can't output it with the ATI drivers. The MPC renderer it seems can't take in anything but 16-bit. This seems like it can be fixed fairly easily, no?

IOW, with the MPC audio renderer, higher-than-16 bit is failing before even reaching the driver.


The MPC renderer would accept anything in condition the audio driver would accept it. This is how it works

However : LPCM has a special mediatype so this is the reason why you won't get 24/32 bits for now with the new ATIs.

This has to be fixed on FFDShow : there is a LPCM checkbox in output section but I don't know what it is used for (DVD audio from the code). I wonder if it shouldn't be used in conjonction with 16/24/32/32 floats checkbox instead of exclusive (checking LPCM will disable the 16/24/32 checkboxes) ?

Anyway, LPCM support is a second step, let's focus on HD audio bitstream : we need to confirm that THD/DTSHD are okay (EAC3, I am working on it) and that now with the last build the xonar is also working correctly

[SamuriHL]

@albain

The latest DirectX SDK I see is from November 2007. Is that the right one to grab? I've got VS 2k8 downloading now and the SDK will download after that.

[Snowknight26]

Can we split the ffdshow bitstreaming discussion/development off to another thread? Seems like anything not relating to bitstreaming goes unnoticed and gets lost (bye bye, bug reports) in the 2/3 new pages made each day.

[tetsuo55]

@albain

The latest DirectX SDK I see is from November 2007. Is that the right one to grab? I've got VS 2k8 downloading now and the SDK will download after that.woah, classic lol.

Get the latest version here : http://www.microsoft.com/downloads/details.aspx?FamilyID=b66e14b8-8505-4b17-bf80-edb2df5abad4&displaylang=en#dx

[albain]

@albain

The latest DirectX SDK I see is from November 2007. Is that the right one to grab? I've got VS 2k8 downloading now and the SDK will download after that.

No, there is a more recent one
http://www.microsoft.com/downloads/details.aspx?FamilyID=B66E14B8-8505-4B17-BF80-EDB2DF5ABAD4&displaylang=en

[SamuriHL]

Thanks, guys. Figures MSDN isn't up to date on that. :rolleyes: :) The download of VS is still going. I found a subversion client to integrate into it. So, getting there.

[albain]

Can we split the ffdshow bitstreaming discussion/development off to another thread? Seems like anything not relating to bitstreaming goes unnoticed and gets lost (bye bye, bug reports) in the 2/3 new pages made each day.

No problem with that. However, this subject is nearly over, a few things apart (latest feedbacks) before merging modifications to the trunk.

I propose to start over with a new thread this monday if necessary

[Skinleech]

Hi albain & fellow testers.

I'm sorry to say that I'm still not getting great results with this, certain files give me issues. One thing to note is that I'm playing back over a network currently - I don't think it will make a difference, but I will also try a problematic file from the local drive to be sure.

DTS-HD: All show was 96000hz with 8 channels in the Ffdshow tray icon, with varying bitrates. I get stutters & audio drop outs.
TrueHD: Seems to read the file data correctly, but I get audio stutter and drop outs on some files.

I have tried both in m2ts and mkv with no luck. Using MPC-HC 1373, the internal splitter, beta 46 and the Arcsoft renderer.

I'll post a sample for both problem files - do you want logs too? I guess yes.

Thanks.

[sub24ox7]

hello i have started testing now too and xonar with ffdshow beta 46 and arcsoft audio renderer works with dolby-true hd from the hangover blu-ray remuxed to m2ts with tsmuxer no drop outs or anything i can tell. I have A dolby true-hd blu-ray with 7.1 and 96000hz track if you would like me to try. i am playing these files over a network as well i don't think that will cuase any problems at all null

[sub24ox7]

alright the 7.1 dolby true-hd 96000hz bitreams perfect! no problems at all, same setup as the previous post WAIT EDIT there are some short dropouts but it is straight from disc with anydvd decrypting and playing the main movie m2ts. I don't think its because i am playing it from disc and not the hard drive so there are still a few things to be worked out. This is so wonderful i am not even taking the time to type a correct sentence or spell. Now i am going to try dts-hd and hr with the xonar

[sub24ox7]

mpc-hc is crashing on me with dts-hd discs

[SamuriHL]

mpc-hc is crashing on me with dts-hd discs

And chance you can look in the windows event viewer and post the offset that it's crashing at? This is an issue I'll be looking into as soon as I can get my dev env built up. Just want to see if the offset matches any of mine.

[sub24ox7]

yes i will do that now

[whurlston]

SHL, if you haven't set up the dev environment yet, I've got to run out and I can pick up a rental copy of Monsters Inc to test it. I've got the environment set up to test. It will be 3 or 4 hours before I get back though.

[sub24ox7]

xml view
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
- <System>
<Provider Name="Application Error" />
<EventID Qualifiers="0">1000</EventID>
<Level>2</Level>
<Task>100</Task>
<Keywords>0x80000000000000</Keywords>
<TimeCreated SystemTime="2009-12-04T18:38:30.000Z" />
<EventRecordID>5244</EventRecordID>
<Channel>Application</Channel>
<Computer>showroom-PC</Computer>
<Security />
</System>
- <EventData>
<Data>mpc-hc.exe</Data>
<Data>1.3.1333.0</Data>
<Data>4b16cf28</Data>
<Data>ffdshow.ax</Data>
<Data>1.0.7.3280</Data>
<Data>4b17fc74</Data>
<Data>c0000005</Data>
<Data>0024da7a</Data>
<Data>e2c</Data>
<Data>01ca7510c5a71d11</Data>
</EventData>
</Event>

mpc-hc.exe
1.3.1333.0
4b16cf28
ffdshow.ax
1.0.7.3280
4b17fc74
c0000005
mpc-hc.exe
1.3.1333.0
4b16cf28
ffdshow.ax
1.0.7.3280
4b17fc74
c0000005
0024da7a
e2c
01ca7510c5a71d11
01ca7510c5a71d11


offset: 0024da7a

[SamuriHL]

SHL, if you haven't set up the dev environment yet, I've got to run out and I can pick up a rental copy of Monsters Inc to test it. I've got the environment set up to test. It will be 3 or 4 hours before I get back though.

That'd be awesome, too. Since you have it set up and know the environment better than me if you don't mind that'd be great. I'm working on getting it set up now. Got VS installed, doing updates. Then I'll install the DX sdk. Then the subversion plugin I found. Then I gotta get the project set up in VS. Clearly I have a ways to go and I'm going out tonight. But I'd like to learn this environment so I can help in the future. But having both of us look at it would be a good thing.

[SamuriHL]

xml view
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
- <System>
<Provider Name="Application Error" />
<EventID Qualifiers="0">1000</EventID>
<Level>2</Level>
<Task>100</Task>
<Keywords>0x80000000000000</Keywords>
<TimeCreated SystemTime="2009-12-04T18:38:30.000Z" />
<EventRecordID>5244</EventRecordID>
<Channel>Application</Channel>
<Computer>showroom-PC</Computer>
<Security />
</System>
- <EventData>
<Data>mpc-hc.exe</Data>
<Data>1.3.1333.0</Data>
<Data>4b16cf28</Data>
<Data>ffdshow.ax</Data>
<Data>1.0.7.3280</Data>
<Data>4b17fc74</Data>
<Data>c0000005</Data>
<Data>0024da7a</Data>
<Data>e2c</Data>
<Data>01ca7510c5a71d11</Data>
</EventData>
</Event>

mpc-hc.exe
1.3.1333.0
4b16cf28
ffdshow.ax
1.0.7.3280
4b17fc74
c0000005
mpc-hc.exe
1.3.1333.0
4b16cf28
ffdshow.ax
1.0.7.3280
4b17fc74
c0000005
0024da7a
e2c
01ca7510c5a71d11
01ca7510c5a71d11


offset: 0024da7a

Then I can tell you what renderer you were using. :D Default DirectSound, non-ATI version. :) Thanks.

[sub24ox7]

yes that log is for the xonar with mpc-hc and ffdshow beta 46 and arcsoft audio renderer the crash that happens with dts-hd ma. Oh and i forgot to mention this same setup with dolby true-hd works great but if you skip with the mouse cursor to a different spot the audio is gone and never comes back in

[SamuriHL]

yes that log is for the xonar with mpc-hc and ffdshow beta 46 and arcsoft audio renderer the crash that happens with dts-hd ma

That was with the arcsoft renderer?? Ohhhh, fascinating. It crashed in the same place as mine did for me with the Default DS renderer. Hmmmmm. You *SURE* it's using the arcsoft renderer and not the Default DS one??

[Skinleech]

That was with the arcsoft renderer?? Ohhhh, fascinating. It crashed in the same place as mine did for me with the Default DS renderer. Hmmmmm. You *SURE* it's using the arcsoft renderer and not the Default DS one??

It has to be - the xonar won't bitstream without the Arcsoft renderer.

Sounds like the crash I get with DTS-HD files, though they play 2nd or 3rd try.

[SamuriHL]

It has to be - the xonar won't bitstream without the Arcsoft renderer.

Sounds like the crash I get with DTS-HD files, though they play 2nd or 3rd try.

I get the same thing with the default DS renderer with my 5870. However, it'll keep crashing. Now if I switch to the DirectSound: ATI HDMI renderer then everything plays fine. But the crashes I've seen reported are ALL with DTS-HD MA tracks. This is what I'm going to attempt to debug tomorrow after I get my system set up for debugging. Whurlston will hopefully beat me to it but I'll go through the exercise anyway as I want to learn so I can help in the future. I loathe putting a dev env on that machine due to TMT3 but I'm keeping my fingers crossed it'll be ok. TMT3 shouldn't care if VS isn't running. It will, however, exit immediately if VS is running. :mad: In any case, we're going to see if we can figure out why these crash.

[sub24ox7]

that is 100% for sure using the arcsoft renderer i hope you can find the reason for the crashes soon . bye and thanks

[Skinleech]

I get the same thing with the default DS renderer with my 5870. However, it'll keep crashing. Now if I switch to the DirectSound: ATI HDMI renderer then everything plays fine. But the crashes I've seen reported are ALL with DTS-HD MA tracks. This is what I'm going to attempt to debug tomorrow after I get my system set up for debugging. Whurlston will hopefully beat me to it but I'll go through the exercise anyway as I want to learn so I can help in the future. I loathe putting a dev env on that machine due to TMT3 but I'm keeping my fingers crossed it'll be ok. TMT3 shouldn't care if VS isn't running. It will, however, exit immediately if VS is running. :mad: In any case, we're going to see if we can figure out why these crash.

Good luck. All DTS-HD tracks here too - HR and MA. Mkv is better though, no crashes there, but I've only tested one so far.

[SamuriHL]

I'll do my best. I haven't touched VS in several years so don't expect miracles out me. :D I'm hoping I can at least pinpoint where it's crashing so albain can take a look at it. It doesn't happen on all DTS-HD MA tracks, so, something has to be common to the ones it's crashing on.

[SamuriHL]

One of you SVN people...help. I can't locate the bitstream branch for ffdshow in svn. Anyone? :)

[clsid]

http://ffdshow-tryout.svn.sourceforge.net/viewvc/ffdshow-tryout/

[SamuriHL]

Well if nobody else can do it (is there ?), you can set up a development enviromnent but there is some work :

1/ VS 2008 (not 2010 it is crappy and so sloooow)
2/ DirectX SDK
3/ Grab last SVN release from bitstreamHD branch

I don't see the bitstreamHD branch under ffdshow. I must be doing something wrong.

[SamuriHL]

http://ffdshow-tryout.svn.sourceforge.net/viewvc/ffdshow-tryout/

Thanks! I forgot it's the tryout not the main ffdshow project. sigh. :)

[SamuriHL]

I'm starting to not like SVN one bit. It keeps telling me the repository has been temporarily moved and won't allow me to open it. WTH???

Nevermind. I grabbed the tarball and called it good. Of course I get a bunch of errors when trying to compile it. Awesome. :)

EDIT: YASM, eh? No one warned me about that or having the source code in a path with spaces! Curse you all! :p

[clsid]

MinGW32 + YASM:
http://sourceforge.net/projects/ffdshow-tryout/files/Tools/mingw32_gcc441_20091010.exe/download

[SamuriHL]

Sorry, I didn't mean to suggest it was still an issue. Just that I was blindsided by it. Got that figured out but it can't find the DX SDK stuff. Not sure where to add the include dir for the headers.

EDIT: The rust is coming off. I'm getting there. Working through it all slowly but making progress. :D

[TripleH]

Hi everyone,

My setup is the following:

ASUS Xonar HDAV 1.3 Deluxe
Windows 7 Ultimate 32-bit
FFDShow Bitstream Beta 46
MPC-HC SVN 1378

I managed to get bitstreaming of HD Audio codecs working, but I have a problem.

I don't know why it happens, but if FFDShow video decoder is loaded, then the Arcsoft Audio Renderer refuses to load, in instead MPC-HC loads the Default DirectShow Device renderer, which of course can't bitstream with the Xonar.

What do you think can cause this problem ?

At first I thought that it is maybe the madVR renderer, but then I tried CoreAVC decoder instead of FFDShow and it worked fine. I also tried FFDShow with EVR and EVR Custom Pres. and the problem returned.

[SamuriHL]

I don't know, this doesn't seem like it's going to work. Every time I fix an error I get more and more and more each time. I added some Boost library and now I've got 1000's of errors. :mad: I think this is simply an exercise in frustration.

KSDATAFORMAT_SUBTYPE_IEC61937_DOLBY_DIGITAL undeclared identifier

[SamuriHL]

You know...I tell ya. I finally get the thing compiled (yea, about that error...I commented that code out :p) and I can tell you the debug version is NOT what you want. :D It skips, slows down, drops out. Yea, anyway, so I bring up Monsters Inc with direct sound. Runs fine. sigh. I give up. :)

[SamuriHL]

Holy snikes I have a headache now. albain, buddy. I know it seems like I'm an idiot when it comes to this dev stuff based on my trials and errors here. :D However, I rebuilt the damn thing in release mode to make sure the error is there. Indeed, KABOOM. Sweet! I figure now I'll just attach to MPC-HC and trap the error. Yeaaaaa, about that. :) I attach the debugger to it and it works fine. sigh. :) I don't have any idea at all how to trap this error!

[albain]

Thanks SamuriHL for all your efforts

This is a big problem when you can't reproduce the crash in debug mode. Have you tried to run it without vs debugging into ?
I mean there are 2 ways to debug : either attach to the process and run it, or let the player run without visual studio and when it will crash it will ask you if you want to debug

Last resort is to add DPRINTF but this is very difficult to narrow the problem down. I will perform some deeper analysis of the code, but I need a max of info

Could you send me logs whith working and "crashing" samples in that case ?

Otherwise I have opened a new thread for HD audio bitstream.

We have taken this one into hostage enough :)

http://forum.doom9.org/showthread.php?p=1349575#post1349575

[SamuriHL]

Moving the discussion to the new thread....

[Spec-Chum]

One of the main causes for something running in debug builds but not in a release build is variable/memory initialisation. This is because in release builds nothing is initialised unless you do it explicitly but in a debug build any allocated variable/memory is initialised automatically to set values.

In Visual C++ these are:

0xCD – Clean - Memory on the heap that has been allocated by new but not yet written to
0xDD – Dead – Heap memory that has been deleted
0xFD – Fence – AKA "no mans land". Automatically placed at boundaries of Clean memory. If you overwrite this you've probably gone off the end of an array
0xCC – All variables created on the stack are initialised to 0xCC until explicitly initialised in the code

There's also 0xFEEEFEEE which is Fill Heap Memory, which is memory marked for usage by the OS, but wasn't allocated by HeapAlloc() or LocalAlloc, or it could just have been freed by HeapFree().

Hope this helps, SamuriHL :)

Spec-Chum

[rica]

Moving the discussion to the new thread....

Yea, it's better i think.

[SamuriHL]

One of the main causes for something running in debug builds but not in a release build is variable/memory initialisation. This is because in release builds nothing is initialised unless you do it explicitly but in a debug build any allocated variable/memory is initialised automatically to set values.

In Visual C++ these are:

0xCD – Clean - Memory on the heap that has been allocated by new but not yet written to
0xDD – Dead – Heap memory that has been deleted
0xFD – Fence – AKA "no mans land". Automatically placed at boundaries of Clean memory. If you overwrite this you've probably gone off the end of an array
0xCC – All variables created on the stack are initialised to 0xCC until explicitly initialised in the code

There's also 0xFEEEFEEE which is Fill Heap Memory, which is memory marked for usage by the OS, but wasn't allocated by HeapAlloc() or LocalAlloc, or it could just have been freed by HeapFree().

Hope this helps, SamuriHL :)

Spec-Chum

Ohhh, nice. See, I've been in java too long and forgot all my C++ training. sigh. :) Yea, that's actually a very good thought. I'll walk through the code. I'd like to pinpoint where exactly it's failing so I can concentrate on whatever module it is. However, you may be on to something. I'll look in the morning. Too tired to deal with it tonight and my wife wants to watch some shows. Thanks for the tip!

[Leak]

I'm starting to not like SVN one bit. It keeps telling me the repository has been temporarily moved and won't allow me to open it. WTH???
You can't check out code from the nicely formated SVN Browse URLs that usually get posted around here, you'll need to use the URL that's listed on the "Code / SVN" page of the project:

https://ffdshow-tryout.svn.sourceforge.net/svnroot/ffdshow-tryout/branches/bitstreamHD/

np: Markus Guentner - Meer Der Luegen (Doppelgaenger)

[SamuriHL]

You can't check out code from the nicely formated SVN Browse URLs that usually get posted around here, you'll need to use the URL that's listed on the "Code / SVN" page of the project:

https://ffdshow-tryout.svn.sourceforge.net/svnroot/ffdshow-tryout/branches/bitstreamHD/

np: Markus Guentner - Meer Der Luegen (Doppelgaenger)

Ah, very nice. I'm new to the whole SVN thing so I had no clue at all how that worked. I ended up grabbing a tarball of the source and extracting that to my drive. I was able to find the crashing bug and offer up a fix so that worked out very well. Now I'm set up for any future debugging that albain needs done. :) Thanks for this tip!!

[[ReX]]

I would be asking too much if I said that register.bat shouldn't run automatically after the compilation?

Or I could just delete the
<Tool
Name="VCPostBuildEventTool"
Description="Registering output..."
CommandLine="bin/register.bat"
/>

lines? :confused: