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. |
8th October 2010, 11:56 | #12601 | Link | |
Registered User
Join Date: Jun 2005
Posts: 278
|
Quote:
If the FLAC file encodes more than 16 bit per sample libavcodec will always produce 32 bit per sample output. What then happens in ffdshow I do not know (it might not be able to play at all, it might reduce to 16 bit, ...). |
|
8th October 2010, 12:19 | #12602 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,346
|
FLAC is actually one of the few codecs that does decode to the proper bitdepth in avcodec.
What ffdshow does with it is unknown to me, however. FWIW, TrueHD/MLP is also properly decoded to the original bit depth. it would be awesome if ffdshow would pass it to the renderer in the same bitdepth as it was decoded, but sadly thats not possible.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
8th October 2010, 21:14 | #12603 | Link |
Registered User
Join Date: Dec 2007
Posts: 652
|
In testing with a FLAC 24 Bit (in MKV Container) I get the following results when "Output format for uncompressed or decoded streams" has "16 and 24 bit interger" checked.
1) FFDSHOW decdoing = 16 Bits output 2) MADFLAC (decodes to 24BIT PCM) --> FFDSHOW (Uncompressed) = 16 Bits ... This one gets me, If you insert FFDSHOW back into the filter graph to do post processing it still reduce the bitdebth to 16 Bit? However, if I check "32 bit integer" I get 32 bit output for it all. ================================ I'm not sure how these 24-Bit streams are being processed internally in FFDSHOW, so I have: 1) Unchecked 16 Bit Integer in Processing 2) Checked 16, 24, and 32 Bit Integer in the Output Format for uncompressed or decoded streams I "think" I may get 24bit input --> ??? 32 Bit ???? internal processing --> 32 bit output, but couple of Q: - Is this assumption correct or is something still reducing the bitdepth to 16-bit at some point in the chain? - Why does it output at 16Bits if I have only 16 and 24 bit output checked instead of 24-bit? Thanks Nathan Last edited by jmone; 8th October 2010 at 22:10. Reason: Edit - added info on Output Format + more clarification |
8th October 2010, 21:56 | #12604 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,346
|
ffdshow will always expose all output formats that you checked, and the renderer will pick whatever it likes. Maybe it doesn't like ffdshows 24 bit, for some arcane reason? I don't know.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
8th October 2010, 22:54 | #12605 | Link | |
Registered User
Join Date: Dec 2007
Posts: 652
|
Quote:
Anyway I'm using Reclock - so the full Audio chain is now more like: MKV (FLAC 24-Bit) Input --> FFDSHOW 32-Bit Output --> Reclock 24bit int padded to 32 (max the format the card with accept) --> WASAPI --> ATI HD5450 I'm just trying to avoid multiple conversions (esp at lower bit debths / sampling rate) unless there are other suggestions to FFDSHOW Config I'll leave it at 32-Bit Output with: 1) Uncheck 16 Bit Integer in Processing 2) Check 16, 24, and 32 Bit Integer in the Output Format So is it confirmed (and againts prior assertions), that FFDSHOW using libavcodec is able to decode, process, and output 24-Bit FLAC at higher than 16-Bit? Thanks Nathan Last edited by jmone; 8th October 2010 at 23:35. Reason: Clarity |
|
9th October 2010, 11:08 | #12606 | Link | |
Registered User
Join Date: Jun 2005
Posts: 278
|
Quote:
But since ffdshow obviously can do the 32 -> 16 bit conversion by itself, there's probably some code missing that converts 32 bit to 24 bit in case the decoder creates 32 bit output and you have only 16 and 24 checked. |
|
9th October 2010, 11:16 | #12607 | Link |
4:2:0 hater
Join Date: Apr 2008
Posts: 1,302
|
ffdshow doesn't output 24-bit unless everything else is unchecked. If the decoder outputs 24-bit, it'll use 32-bit int, with the extra bits zeroed (not sure about this though). So, if you have 24-bit sources, just be happy with 32-bit int because all your info is there untouched.
|
9th October 2010, 15:55 | #12608 | Link | |
German doom9/Gleitz SuMo
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,782
|
Quote:
|
|
9th October 2010, 17:46 | #12609 | Link | |
Registered User
Join Date: Jun 2005
Posts: 278
|
Quote:
|
|
10th October 2010, 21:21 | #12611 | Link |
Registered User
Join Date: Oct 2009
Location: France
Posts: 616
|
Hi STaRGaZeR,
It seems that when you remove mp3lib, it break something about FFDShow.Interfaces.IffdshowBase.getParamStr3(UInt32 paramID, IntPtr& bufPtr) when use FFDshow in MediaPortal, r3602 is OK but not 3604. I try to see why. Thanks, Seb. Edit : In FFDshowAPI.cs : i comment out line 2173 : ffdshowBase.getParamStr3((uint)param, out result); I don't know exactly is doing but it fix the issue.
__________________
HTPC : i7 920 6Go Win10(x64) / Nvidia 1050Ti / P6T Deluxe / Harman-Kardon AVR-355. Last edited by Sebastiii; 10th October 2010 at 21:32. |
11th October 2010, 20:11 | #12614 | Link |
4:2:0 hater
Join Date: Apr 2008
Posts: 1,302
|
Compile a debug build, put breakpoints in TffdshowBase::getParamStr3 (TffdshowBase.cpp, line 164) and run debug. When something calls the function the execution will be stopped at the first breakpoint. You can see variable names, values, etc. By pressing the play icon you'll go to the next breakpoint. You just have to look for what's in the variables when it crashes, that will pretty much tell you where the problem is. You know the issue is in getParamStr3 but knowing the exact line and variable values helps a lot. You see that the get(...) function is called from getParamStr3, if it crashes in this line you can put breakpoints in get(...) to further track down the crash. Maybe it's an incorrect param, I don't know
|
11th October 2010, 20:42 | #12615 | Link | ||
Registered User
Join Date: Oct 2009
Location: France
Posts: 616
|
Quote:
So i have compil debug version of FFDShow, put break point line 164. Start debug -> select -> mediaportal.exe Open a movie and try to select subtitle. MVS2008 -> ask me for FFDShowAPI.cs (the one from MePo Sources). After that, it stop on line 2254 (from this FFDShowAPI.cs) line for the one from FFDShow source is line 2173. I have this value for param : FFDShow.FFDShowConstants.FFDShowDataId.IDFF_subTempFilename Quote:
I really don't know why Thanks. Ps : Thx to Underground78 to try to help me
__________________
HTPC : i7 920 6Go Win10(x64) / Nvidia 1050Ti / P6T Deluxe / Harman-Kardon AVR-355. |
||
12th October 2010, 14:45 | #12616 | Link | |
4:2:0 hater
Join Date: Apr 2008
Posts: 1,302
|
Quote:
This crash is related to subtitles, and I didn't change anything regarding subtitles in the mp3lib commit, so I'm out of ideas. What you can do is revert file by file until you find what makes it crash. BTW, why does MP have a different FFDShowAPI.cs? Really a noob on this I am |
|
12th October 2010, 17:07 | #12617 | Link |
Registered User
Join Date: Oct 2009
Location: France
Posts: 616
|
Thanks,
I will try this this evening I have look @ each file and yes nothing related to sub (and it's only happen for external sub). I don't know exactly why the API (cs) is different, i know that i have add modification in FFDShowAPI.cs but the bug appear after r3602. I'll and notice you Thanks again.
__________________
HTPC : i7 920 6Go Win10(x64) / Nvidia 1050Ti / P6T Deluxe / Harman-Kardon AVR-355. |
12th October 2010, 19:09 | #12618 | Link |
Registered User
Join Date: May 2005
Posts: 395
|
raised a new issue on bug-tracker regarding subtitle (start play with subs hidden) interaction between Zoom Player 7.1a5 and clsid's newest 32-bit/x86 ffdshow build (3611)
https://sourceforge.net/tracker/?fun...41&atid=867360 also CC'd ZP msg board thanks, m |
12th October 2010, 22:41 | #12619 | Link | |
Registered User
Join Date: Oct 2009
Location: France
Posts: 616
|
Quote:
Debug version is working (register .ax) but the issue is always here and i didn't find what happen, i have try each file but it's hard because all is link between them (not all but a big part). And i didn't see why is related to sub section +++
__________________
HTPC : i7 920 6Go Win10(x64) / Nvidia 1050Ti / P6T Deluxe / Harman-Kardon AVR-355. Last edited by Sebastiii; 14th October 2010 at 08:34. |
|
14th October 2010, 21:31 | #12620 | Link |
Registered User
Join Date: Feb 2004
Posts: 399
|
It seems libavcodec's VP6F decoding has a slight bug. Here is a sample video:
http://videoff7.free.fr/FF13_fn.avi (15megs) When decoded by ffdshow, there is some disturbance at the very bottom of the screen, all along. When played via activex (on website), there is no such artefact. I tried every renderer and no changes, if anyone has a clue on how to solve this? ^^; Last edited by TheShadowRunner; 14th October 2010 at 22:11. |
Tags |
ffdshow, ffdshow tryouts, ffdshow-mt, ffplay, icl |
|
|