LAV Filters - DirectShow Media Splitter and Decoders

[nevcairiel]

Quite possibly i used 0x3f because Windows 7 can generally deal with both just fine, but XP had some occasional issues with 0x60f (0x3f was still the preferred mode in XP), honestly i don't remember anymore why i choose to do it like this. There haven't been any issues really. Even if LAV would change to 0x60f, or even better, change to output the value in the file untouched, you will still need to handle 0x3f.

[Thunderbolt8]

got some problem with getting LAV splitter to work with some kind of file types.

got .flv files from twitch.tv which I can open just fine with the splitter.

Video
Format : AVC
Format/Info : Advanced Video Codec
Format_Profile : Main@L3.1
Format_Settings_CABAC/String : Yes
Format_Settings_RefFrames/String : 1 frame
CodecID : 7
Duration/String : 22mn 10s
BitRate_Mode/String : Variable
BitRate/String : 1 947 Kbps
BitRate_Maximum/String : 2 011 Kbps
Width/String : 1 280 pixels
Height/String : 720 pixels
DisplayAspectRatio/String : 16:9
FrameRate_Mode/String : Constant
FrameRate/String : 30.000 fps
Standard : NTSC
ColorSpace : YUV
ChromaSubsampling : 4:2:0
BitDepth/String : 8 bits
ScanType/String : Progressive
Bits-(Pixel*Frame) : 0.070
StreamSize/String : 317 MiB (95%)

Audio
Format : AAC
Format/Info : Advanced Audio Codec
Format_Profile : HE-AAC / LC
CodecID : 10
Duration/String : 22mn 10s
BitRate/String : 92.7 Kbps
Channel(s)/String : 2 channels
ChannelPositions : Front: L R
SamplingRate/String : 44.1 KHz / 22.05 KHz
Compression_Mode/String : Lossy
Video_Delay/String : 10ms
StreamSize/String : 15.4 MiB (5%)


but when I remux some of these files to .mkv (by joining some parts via append), then I cannot open the .mkv file with lav splitter alone any more and need to have "file source (async)" running as well as external filter in mpc-hc. when I block file source then I get the message that the file cannot be rendered.

stats for the .mkv file dont seem to be too different:

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : Main@L3.1
Format settings, CABAC : Yes
Format settings, ReFrames : 1 frame
Codec ID : V_MPEG4/ISO/AVC
Duration : 4h 52mn
Bit rate mode : Variable
Maximum bit rate : 2 011 Kbps
Width : 1 280 pixels
Height : 720 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 30.000 fps
Standard : NTSC
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Default : Yes
Forced : No

Audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : HE-AAC / LC
Codec ID : A_AAC
Duration : 4h 52mn
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 44.1 KHz / 22.05 KHz
Compression mode : Lossy
Delay relative to video : 46ms
Default : Yes
Forced : No

so why is this?

[nevcairiel]

Probably because you installed something after LAV which destroyed the registry entry. Run the LAV installer again.

Note that the other way isn't really any worse.

[Reino]

1. add support for chapters(cuesheets) in ape/flac/wv?Please add TAK to that and maybe there are more lossless audioformats that have support for cuesheets.

For those interested: the initial cuesheet support request in MPC-HC (https://trac.mpc-hc.org/ticket/1673)
After 22 months still no support in MPC-HC, while MPC-BE has added support for cuesheets for their FLAC- and WavPack source filter some time ago already.
I've also got it on my wishlist for DC-Bass Source Mod, but I don't have the know-how and I can't find people who do :(.

P.s. Nev, good job on the support for Matroska xml-chapters. Unlike MPC-HC/BE (https://trac.mpc-hc.org/ticket/1673#comment:8), they look exactly the same as ogm-chapters with LAV Splitter.

[Aleksoid1978]

Hi Nevcairiel

Can you modify LAV Splitter, that it for .aiff format return "normal" MediaType for audio output. Now it for int24be(AV_CODEC_ID_PCM_S24BE)/int32be(AV_CODEC_ID_PCM_S32BE) return:

subtype: Unknown GUID Name {00000000-0000-0010-8000-00AA00389B71}
wFormatTag: 0x0000


And no any external Audio Decoder can't connect, only LAV Audio + WAVEFORMATEXFFMPEG.

Modify subtype:
AV_CODEC_ID_PCM_S24BE = MEDIASUBTYPE_PCM_IN24
AV_CODEC_ID_PCM_S32BE = MEDIASUBTYPE_PCM_IN32

[Thunderbolt8]

Probably because you installed something after LAV which destroyed the registry entry. Run the LAV installer again.

Note that the other way isn't really any worse.yep, that was it. I installed a bunch of other stuff after LAV filters, also the haali media splitter among other things. wasnt there said to be a problem with that when installing haali afterwards?

[nevcairiel]

Can you modify LAV Splitter, that it for .aiff format return "normal" MediaType for audio output. Now it for int24be(AV_CODEC_ID_PCM_S24BE)/int32be(AV_CODEC_ID_PCM_S32BE) return:

What about int16be?

[Aleksoid1978]

What about int16be?

I think yes - but i do not have such .aiff files.

[Aegwyn11]

@nevcairiel

When are you planning on a next release?

How much trouble would it be to build the next version of LAV filters based on current (today) FFmpeg code?

I'm asking because there was a regression introduced in FFmpeg around UTvideo decoding that just got fixed, and as a result LAV Filters after 0.55.3 has a broken UTvideo decoder.

Details on the FFmpeg bug here:
https://ffmpeg.org/trac/ffmpeg/ticket/2661

[nevcairiel]

The latest version of LAV already includes this ffmpeg fix.
I always try to keep it fairly up to date, as long as there aren't any big changes shortly before my release which would require more extensive testing.

[nevcairiel]

I think yes - but i do not have such .aiff files.

Done.
16-bit uses the TWOS subtype.

[Aleksoid1978]

Done.
16-bit uses the TWOS subtype.

Thanks.

[Octo-puss]

Is LAV supposed to decode XviD? I thought it was, but I can't get Sony Vegas to open such video at all. Only after downloading the codec from their site it started to work.

[VipZ]

I found a very odd bug, i happened to drag the MPC-BE exe file onto itself and it played with LAV Splitter, not that it sounded any good :)

Any chance you could block this from occurring.

[nevcairiel]

Not really, its a feature that it tries to detect media in all files, and this can in some cases lead to false positives i suppose.
Could limit it to exlcude files probeb with a very low score i suppose, but not that much gained from that.

[VipZ]

No problem, guess one would normally not try play an exe :p

[Aegwyn11]

The latest version of LAV already includes this ffmpeg fix.
I always try to keep it fairly up to date, as long as there aren't any big changes shortly before my release which would require more extensive testing.

I don't understand.... The latest version shown is 0.57 and definitely has this problem. The issue was only fixed in ffmpeg a couple of days ago.

[nevcairiel]

The latest version is not the latest release. :P

[Warlock]

Guys, I know the question is silly, but I need to solve this question: In output formats, 8 bits, which option is the best: NV12 or YV12? Some people say that YV12 is best for 8 bits, is this true?

[wanezhiling]

@Warlock
http://forum.doom9.org/showpost.php?p=1574634&postcount=6
http://forum.doom9.org/showpost.php?p=1600852&postcount=15088

[Warlock]

@Warlock
http://forum.doom9.org/showpost.php?p=1574634&postcount=6
http://forum.doom9.org/showpost.php?p=1600852&postcount=15088

Thanks, my question was completely solved.

[Mercury_22]

How to use the new RTMP ?
Something like rtmp://npsaflivefs.fplive.net:1935/npsaflive-live/stream50 it's working fine with LAV in MPC-BE BUT how do I play this rtmp://62.115.64.212/live?_fcs_vhost=cp116697.live.edgefcs.net playpath=BnazlkNDpCIcD-QkfyZCQKlRiiFnVa5I_640_360_1000@18679 pageUrl=http://www.bloomberg.com/tv/ app=live?_fcs_vhost=cp116697.live.edgefcs.net swfUrl=http://player.ooyala.com/static/cacheable/6086bd175c4cf1d14cbd4c9aa54d019d/player_v2.swf/[[DYNAMIC]]/3 swfVfy=true live=true timeout=60which is playing OK in XBMC as a strm file

[nevcairiel]

There is no rtmp support in LAV, and even if someone enables it in the ffmpeg build, there is no official support for it, so you figure it out.

So if anyone wants to write a parser for such URLs and pass all the appropriate settings to ffmpeg, then i can add it. :p

[zerowalker]

Will VP9 be added?
As if i am correct, the bitstream is frozen, making a decoder that works now, works forever.

[Mercury_22]

There is no rtmp support in LAV, and even if someone enables it in the ffmpeg build, there is no official support for it, so you figure it out.

So if anyone wants to write a parser for such URLs and pass all the appropriate settings to ffmpeg, then i can add it. :p

Looks like "Someone":rolleyes: has remove it now :( It was in the git earlier today

[clsid]

Edit: it seems that FFmpeg is not able to parse the parameters given after the rtmp:// URL. The calling app need to parse them and pass them to the demuxer as options (see rtmpproto.c).

[nevcairiel]

Looks like "Someone":rolleyes: has remove it now :( It was in the git earlier today

I accidentally pushed a test branch, it was up for maybe 5 minutes or so, and if you had read the commit message that came with it, it would've explained that anyone testing this is on their own. :p

[kasper93]

@clsid @Mercury_22

LAV doesn't support rtmp so it's pointless to discuss how to pass additional parameters. RTMP support can be easily enabled by changing ffmpeg build script, but this only support simple streams. Basically only those which can be described by singe link
rtmp://server[:port][/app][/instance][/playpath]

[ryrynz]

Has anyone else experienced random glitches with what I assume is avcodec decoding? Sometimes I'll have a small square patch that displays properly while the rest of the video freezes.
It happens on both my PCs, if you rewind and replay that particular sequence there is no problems, completely random.

[dukey]

nevcairiel, does lav correctly set the average time per frame on dynamic format change ?
http://i.imgur.com/87xfqHu.png

[nevcairiel]

You should never really rely on that value, its notoriously inaccurate.
But sure, when the source sends a new time, it should end up at LAVs output as well. Unless of course its one of those cases where LAV decides to override it with a bitstream value.

[dukey]

It seems not to propagate the value, and if it has overridden it, then it's done so incorrectly I think.

[nevcairiel]

Like i said, simply don't rely on it, its unreliable.

[dukey]

I need it to properly create a dxva type, otherwise I have to guess some default rate

[nevcairiel]

LAV Filters 0.58

General
- LAV Filters now includes a manifest file to control DLL loading
- Fixed the shortcuts to the configuration dialogs created by the installer in the start menu

LAV Splitter
- Handle the "Default" flag in MKV files with multiple video streams
- Fixed a regression in VC-1 demuxing which caused some streams to fail decoding with the WMV9 DMO decoder
- Improved transition between ordered chapters in certain (badly formed) MKV files
- Improved handling of AVC1 in AVI files
- Improved accuracy of IAMStreamSelect::get_CurrentMarker
- Several potential crash fixes and stability improvements

LAV Video
- YADIF deinterlacing is now multi-threaded, significant performance improvement on multi-core CPUs
- Improved detection of progressive streams to ensure deinterlacing is disabled, especially when "Aggressive Deinterlacing" is used
- Restored Fraps constant frame rate output by proper handling of repeated frames
- Support for the new UtVideo sub-types, and fixed a regression in UtVideo decoding
- cuvid: fix double-rate deinterlacing on certain streams without timing information
- Updated the Intel QuickSync Decoder to 0.44


Download: Installer (both x86/x64) (http://files.1f0.de/lavf/LAVFilters-0.58.exe) -- Zips: 32-bit (http://files.1f0.de/lavf/LAVFilters-0.58.zip) & 64-bit (http://files.1f0.de/lavf/LAVFilters-0.58-x64.zip)

This message has been reduced in length due to heat.

Standard disclaimer applies, report regressions/issues, include samples, because without reproducing a bug, no fixing a bug.

So, take care, and enjoy the summer while it lasts - well, if the summer reached you yet.

[Qaq]

Thanks for the new version.
And yes, its hot on Baltic sea these days.

[DragonQ]

Heh, I'm wearing a jumper indoors!

Thanks for the new release. Would you care to expand a bit on the "Improved detection of progressive streams to ensure deinterlacing is disabled, especially when "Aggressive Deinterlacing" is used" fix?

[nevcairiel]

It does what it says, it improves the detection of 100% progressive streams to ensure deinterlacing is turned off.
Previously it would rely on the media type information for this, but when watching Live TV for example, there typically is no such information available, and it could deinterlace some progressive broadcasts which didn't need it - and even cause some issues when double-rate deinterlacing a stream which already was 59.94 progressive (as i'm told some ATSC broadcasts in the US are)

[DragonQ]

I see. I assume it's not based on looking at the actual frame images though?

[nevcairiel]

No. Just the info in the bitstream.

[Keiyakusha]

- Handle the "Default" flag in MKV files with multiple video streams
- Improved transition between ordered chapters in certain (badly formed) MKV files
Thanks! Exactly what I needed... though this means one of my milti-edition files is "badly formed". This is what I thought but couldn't figure out how to fix it. >__<

[nevcairiel]

"badly formed" in this case just means that the chapter does not *exactly* match a Cue point, however if the new version fixed it, then there is a cue point very close by at least, so the chapter work was just "sloppy" (ie. a few milliseconds off target)

[cyberbeing]

There seems to be a LAV Splitter regression regarding MKV Segment Linking which was introduced between LAVFilters-0.55.3-73-g3f9ce87 and LAVFilters-0.56.1.

LAVFilters-0.55.3-73-g3f9ce87 (working) <- LAVFilter test builds inbetween (crash) -> LAVFilters-0.56 (crash) -> LAVFilters-0.56.1 (issue) -> LAVFilters-0.58.0 (issue)

I'll PM you a sample.

[nevcairiel]

Its not a regression if it never worked in a release. :D

[pOpY]

Dear nev.

v0.58 doesnt solve my issue.
I have updated Issue 353 on bugtracker.

https://code.google.com/p/lavfilters/issues/detail?id=353

Ill think now you should be able to easily reproduce the problem.

thx
pOpY

[pOpY]

Also i had to edit build_ffmpeg.sh line 21-23 from.

strip_libs() {
x86_64-w64-mingw32-strip lib*/*-lav-*.dll
}

to

strip_libs() {
strip lib*/*-lav-*.dll
}

to work with my MinGW.

[nevcairiel]

Thats because your mingw is incomplete. :)

[pOpY]

hmm, thats not good :(
could you please provide me your MINGW directory or pint me in the right direction :o

Ill have used this one: http://nuwen.net/mingw.html
with all components and msys.

thx
pOpY

[nevcairiel]

Thats apparently the "old" mingw32, while around here we switched to mingw-w64 a long time ago.
Its curious though, GCC 4.8.1 shouldn't even build with the old mingw, but hey, what can you do.

My mingw-w64 based toolchain is always available here (without MSYS, you still need that from somewhere else):
http://files.1f0.de/mingw/

Edit:
The mingw-w64 project has a complete MSYS package which should work:
http://sourceforge.net/apps/trac/mingw-w64/wiki/MSYS

Unzip MSYS, and then put my mingw-w64 toolchain into MSYS\mingw
To build ffmpeg you'll need some additional tools like YASM, though.

[kasper93]

You can also grab XhmikosR's package http://xhmikosr.1f0.de/tools/