[Ffmpeg-devel-irc] ffmpeg.log.20150111

burek burek021 at gmail.com
Mon Jan 12 02:05:01 CET 2015


[02:25] <hann> greetings. what do you do when you want/need to -c:v libx264 -preset veryslow or alike?
[02:25] <hann> do you actually have a crazy single-core computer?; do you EC2?; is there some cluster project I'm unaware of?
[02:25] <hann> my pretty decent build still has to run overnight. I'm not keen on GPU rendering, but maybe I should be?
[02:25] <c_14> put my computer in a corner and try not to touch it for 20-30 hours
[02:26] <hann> bah :P
[02:26] <hann> yeah. well, I guess 3 days ain't that much in a wider perspective.
[02:27] <PovAddict> can I use my computer's audio output as ffmpeg's audio input?
[02:27] <PovAddict> without using some analog trick like connecting my line out to my line in with a cable :P
[02:28] <c_14> PovAddict: Windows/Mac/Alsa/Pulse ?
[02:28] <PovAddict> Linux, Alsa
[02:28] <c_14> hann: more powerful cores will help, graphics is still very meh
[02:29] <c_14> PovAddict: https://trac.ffmpeg.org/wiki/Capture/ALSA#Recordaudiofromanapplication
[02:29] <PovAddict> I want to also hear the output - I see that's the next section :)
[02:29] <hann> c_14: so this is not favorable? http://www.phoronix.com/scan.php?page=news_item&px=MTcxMDc
[02:29] <PovAddict> but it looks scary :o
[02:30] <c_14> PovAddict: it's not as scary as it looks (I hope)
[02:32] <c_14> hann: graphics card video encoding doesn't do high compression very well, for technical reasons I keep forgetting
[02:33] <hann> c_14: thanks
[02:36] <PovAddict> c_14: do I need to restart something after editing .asoundrc?
[02:36] <c_14> nop
[02:37] <PovAddict> [alsa @ 0x18ad240] cannot open audio device loopout (Invalid argument)
[02:37] <PovAddict> ah got it
[02:38] <PovAddict> or not
[02:40] <PovAddict> now ffmpeg -i loopout works, but programs can't output audio
[02:41] <PovAddict> what am I supposed to enter in 'card'?
[02:44] <PovAddict> found what to put in 'card', now audio output works and ffmpeg doesn't fail, but doesn't get any audio either D:
[02:44] <PovAddict> ffmpeg spews lots of "ALSA buffer xrun."
[02:45] <c_14> try using speaker-test -c2 -Dloopin as the source
[02:45] <PovAddict> "Unable to set buffer size 570652 for playback"
[02:46] <PovAddict> wait you meant stopping ffmpeg?
[02:47] <c_14> You shouldn't have to stop it, but try it anyway.
[02:47] <c_14> /restarting it
[02:47] <PovAddict> let's see, nothing is running now
[02:47] <PovAddict> I start speaker test
[02:48] <PovAddict> now I run ffmpeg, and I get "ALSA buffer xrun." messages
[02:48] <c_14> The xrun isn't that drastic
[02:49] <c_14> As long as it gets audio input
[02:49] <PovAddict> ...and now vlc can't open the audio output device so I can't check if my capture has audio or not
[02:49] Action: PovAddict tries mplayer instead
[02:49] <PovAddict> doesn't seem to have audio
[02:49] <PovAddict> no, mplayer couldn't open the audio output device either
[02:50] <c_14> ye, cause I overwrote the default Alsa output device
[02:50] <PovAddict> 'device or resource busy'
[02:50] <c_14> Or wait
[02:50] <c_14> You used the multi config?
[02:50] <PovAddict> "Record audio from an application while also routing the audio to an output device"
[02:50] <PovAddict> I used that section
[02:51] <c_14> Did you copy-paste it?
[02:51] <PovAddict> copied the .asoundrc verbatim except for the 'card' line
[02:51] <c_14> ok
[02:53] <PovAddict> so now what? how do I listen to anything? ._.
[02:53] <c_14> I'm testing on my system
[02:53] <PovAddict> youtube in chromium still works o.O
[02:54] <c_14> oh
[02:54] <c_14> kill chromium
[02:54] <c_14> And all other applications that could be using audio that you don't want to use audio
[02:55] <c_14> That config has no dmix support, therefore only one application will be allowed to output audio at a time.
[02:55] <c_14> And web browsers love stealing audio access and never releasing it.
[02:59] <PovAddict> it looks like the buffer xruns *are* a problem... I hear bits and pieces of the audio
[02:59] <c_14> hmm
[02:59] <c_14> shouldn't be happenning
[03:01] <PovAddict> http://stuff.povaddict.com.ar/capture.mp4
[03:03] <PovAddict> is this easier in pulse?
[03:05] <c_14> I've personally never tried it in pulse, but you do get graphical thingies you can click on.
[03:28] <PovAddict> gah... ffmpeg captures from x11grab just fine when recording video, but when streaming to ustream it streams blackness
[03:28] <PovAddict> avconv "works", but leaks memory and eventually dumps core
[04:08] <xreal> I want to build stereo from AC3's front left and right only. Does it work this way? -filter_complex "channelsplit=channel_layout=5.1[FL][FR]" -ac 2 ?
[04:09] <shadowchancellor> Hey I've got a pcm file that contains a small portion of non-pcm "mystery" data. Is there a way I can tell which bytes are the non-pcm ones, and potentially what format they may be?
[04:10] <xreal> I am stupid...  pan audio filter helps me.
[04:11] <xreal> ffmpeg -i 6channels.wav -af "pan=stereo|c0=FL|c1=FR" output.wav <- isn't it?
[04:32] <shadowchancellor> Has anyone demuxed raw PCM before?
[04:32] <klaxa> no, nobody has ever done that
[04:33] <klaxa> wait... what?
[04:33] <klaxa> you have some random bytes in your pcm data?
[04:33] <klaxa> is there a regularity to it? is there some sort of rule?
[04:34] <shadowchancellor> I have a pcm file that contains a small portion of non-pcm data
[04:35] <shadowchancellor> It's part of a developer challenge to separate that non-pcm data and "watch" it
[04:35] <klaxa> well does the pcm data make sense when listened to?
[04:35] <shadowchancellor> When I convert the pcm to wav, the last ~.036 seconds sound like fuzz
[04:36] <klaxa> you could take a look at its spectrogram or other graphs with graphical software like audacity or use a videofilter to produce graphs
[04:36] <shadowchancellor> I've analyzed it in audacity and have a fairly close of idea of where the non-pcm begins
[04:36] <shadowchancellor> approx the last 600 bytes
[04:37] <shadowchancellor> ~550-600
[04:37] <klaxa> well take them out and look at it in a hex editor
[04:37] <shadowchancellor> sure I've got it open in sublime text
[04:38] <shadowchancellor> can't really glean much from the binary
[04:39] <shadowchancellor> Only oddity is there are a lot of 0's starting out
[04:40] <shadowchancellor> 0x00
[04:41] <shadowchancellor> about ~5kb is like that
[04:41] <shadowchancellor> in the beginning
[04:42] <shadowchancellor> would those bytes equate to silence in a wav file?
[04:42] <klaxa> yes
[04:42] <klaxa> at least as far as i know
[04:44] <shadowchancellor> so then I guess it's mostly just a matter of somehow detecting the binary format
[08:40] <meorom> hi Mavrik: finally, I can build success FFmpeg with MSVC! Thanks for your help!
[08:43] <meorom> problem here is I use wrong Visual developer command prompt (2012 instead of 2013), so I can't link libs in VS 2013 project
[10:53] <Fjorgynn> Sun
[13:20] <MK> Hey guys, is there a known problem with encoding lossless h264? Everytime I do "-crf 0" or "-qp 0" with libx264 the final video file shows the wrong frame width/height in its file properties (Windows). It shows as 16x96 instead of 1920x1080. I'm using the latest precompiled build (Jan 9 2015).
[13:21] <MK> using "-preset veryslow" instead shows 32x96, any value for crf or qp > 0 shows correctly as 1920x1080
[13:24] <JEEB> MK, I wouldn't expect whatever is producing that information to be able to parse the extradata properly
[13:25] <JEEB> I mean, there are close to zero commercial software decoders that handle lossless AVC
[13:25] <JEEB> (coreavc is the only one I remember)
[13:26] <MK> hmm so lossless avc is a no-go? MPC-HC properly plays the video but my intent was to edit it in Premiere Pro, but this isn't a good way to go then?
[13:27] <JEEB> of course since mpc-hc contains a lavc-based decoder :P
[13:28] <JEEB> I think it's workable if you use a mode that lets you use an open source decoder
[13:28] <JEEB> usually it means sticking (lossless) AVC (without b-frames) into AVI and then using a VFW decoder for it
[13:28] <ribasushi> MK: also are you sure it is a codec issue as opposed to container issue? tried different -f ?
[13:29] <JEEB> ribasushi, it's just a failing parser :P
[13:29] <JEEB> too bad the only lavc-based VFW decoder is by now ancient
[13:29] <JEEB> (ffdshow-tryouts)
[13:30] <ribasushi> JEEB: it is a failing parser yes - my point was perhaps the parser will fail less with a different mux
[13:32] <JEEB> well you can try mpeg-ts and mp4, since that's what Windows supports :P
[13:32] <JEEB> but I think the result will be very similar
[13:37] <MK> hmm thanks for your answers :P The more I dig into this the less useful Premiere Pro is :( I've used it for years and only now realised it doesn't support variable framerate and some common formats/containers...
[18:48] <H2O1> I want to keep only the first 2940 seconds of  myfile.mp3 (3167 seconds) (Linux and Windows) with this "ffmpeg -loglevel error -t 2940 -i myfile.mp3 out-myfile.mp3".  The problem is that this keeps too much time (3018 seconds) (I use WindowsMediaPlayer to evaluate the time, which should be alright). it always keeps too much time, that was just one example
[18:49] <c_14> What does ffprobe say?
[18:49] <H2O1> i don't know what is ffprobe
[18:50] <c_14> Just do an `ffmpeg -i file' then
[18:50] <c_14> What does start and end say?
[18:50] <H2O1> just a moment...
[18:54] <H2O1> for the original file it says http://pastebin.com/cjJtjAmY
[18:56] <H2O1> for the output file it says http://pastebin.com/4AxDkc2R
[18:57] <c_14> According to that, it's 3 microseconds longer than 49 minutes and those 3 microseconds are because of the start offset
[18:58] <H2O1> could the problem be with WindowsMediaPlayer, because the times I mentioned are what it shows
[18:59] <c_14> could be
[18:59] <H2O1> ok, I'll try it with VLC
[19:12] <Jiippe_> hello. any ideas how i could convert dvb-subs from mpeg transport stream to dvd compliant format? i was able to extract the video and audio but can't get the dvb-subs stream out of it. Any ideas? It needs to be scaled down, it is hdtv material..
[19:23] <xreal> When using  -t 03:00:00 before -i, ffmpeg complains "-t is not an input option" - why?
[19:26] <xreal> fflogger: oh man...
[19:26] <xreal> c_14: same, 1 sec
[19:27] <xreal> c_14: http://pastebin.com/2SE2tEhY
[19:31] <c_14> Because using -t as an input option was merged on Aug 6 2013
[19:31] <c_14> And your build is from march
[19:32] <Darby_Crash> hi to all
[19:32] <xreal> c_14: 90% of the tutorials on the web blame users, not using -t on an input :D
[19:33] <xreal> c_14: that's why I hate the internet. 1000 of outdated and wrong tutorials, noone cares about :(
[19:33] <Darby_Crash> thanks for your wonderful ffmpeg!
[19:33] <xreal> c_14: yeah, this was the wrong ffmpeg, fixed the path, thx
[19:33] <Darby_Crash> i've a question
[19:33] <c_14> ask, if someone can help you they will
[19:33] <Darby_Crash> i use this script for compile ffmpeg on Debian
[19:34] <xreal> c_14: But when pusing it at the end, will it compress the complete input? Normally, this should be an input function, which tells to use 3 hours of the *input* only.
[19:36] <Darby_Crash> http://pastebin.com/ghgntsE6
[19:36] <Darby_Crash> how can i compile also ffplay?
[19:36] <c_14> xreal: the result is the same, the way it works is a bit different
[19:37] <xreal> c_14: I see. But it's confusing :D
[19:37] <c_14> Darby_Crash: install sdl
[19:37] <Darby_Crash> i have do it
[19:38] <c_14> xreal: -t as an input option means decode only time, -t as an output option means encode only time
[19:38] <xreal> Darby_Crash: "never" use make install on Debian :D
[19:38] <Darby_Crash> in the script i compile it from source
[19:38] <c_14> Darby_Crash: pastebin config.log
[19:38] <xreal> Darby_Crash: yeah, but enter a prefix when doing "make install".
[19:38] <xreal> ah, you did : --prefix
[19:39] <xreal> I'm quiet.
[19:43] <Darby_Crash> is too large
[19:44] <xreal> Darby_Crash: use this https://clbin.com/
[19:44] <Darby_Crash> thanks
[19:47] <xreal> c_14: "-preset help" seems as it was removed? http://pastie.org/9825895
[19:50] <c_14> xreal: seems so
[19:50] <xreal> c_14: okay, let's crawl through the sourcecode and find the settings :)
[19:50] <xreal> Darby_Crash: any update?
[19:51] <Darby_Crash> sorry but the page freeze
[19:51] <xreal> c_14: A bitrate of 4000 kbit/s means, about 40% of my 100 MBit/s LAN network will be used?
[19:52] <xreal> Darby_Crash: just do this and give us the link: cat config.log | curl -F 'clbin=<-' https://clbin.com
[19:53] <c_14> xreal: 4% afaik
[19:53] <Darby_Crash> https://clbin.com/LElI5
[19:54] <xreal> c_14: A normal 100 MBit/s network can do 10 MB/s only :(
[19:54] <crschmidt> xreal: yes, and 4kbit/s is .5MB/s
[19:55] <xreal> crschmidt: 4000 kbit/s !
[19:55] <crschmidt> Sorry, yes. 4000Kbps is 4Mbps is .5MB/s
[19:55] <xreal> crschmidt: ok. Enough for my Raspberry for streaming over lan :) Thx.
[19:56] <xreal> Darby_Crash: did you look in the building directory if there's ffplay?
[19:56] <xreal> Darby_Crash: just try ./ffplay in the directory you compiled the code.
[19:57] <c_14> Darby_Crash: SDL detection is failing.
[19:57] <c_14> Darby_Crash: SDL 1.0 or 2.0 ?
[19:58] <Darby_Crash> take a look in my bash
[19:59] <c_14> mhm
[19:59] <Darby_Crash> 1.2.15
[19:59] <Darby_Crash> compiled from source
[20:00] <Darby_Crash> i have ffplay.c
[20:00] <Darby_Crash> in target/bin
[20:00] <c_14> Darby_Crash: what does `pkg-config --libs --cflags sdl' return
[20:02] <Darby_Crash> root at localhost:/sdcard# pkg-config --libs --cflags sdl
[20:02] <Darby_Crash> -D_GNU_SOURCE=1 -D_REENTRANT -I/usr/include/SDL  -lSDL
[20:03] <c_14> export PKG_CONFIG_PATH=$TARGET_DIR/lib/pkgconfig
[20:03] <Darby_Crash> in my bash?
[20:04] <c_14> yep
[20:05] <Darby_Crash> is this the solution?
[20:05] <Darby_Crash> :)
[20:07] <c_14> It should be
[20:10] <Darby_Crash> ok
[20:10] <Darby_Crash> i try it
[20:10] <Darby_Crash> :)
[20:25] <Darby_Crash> anyway i want ffmpeg on Wheezy again :)
[20:26] <Darby_Crash> avconv is not good
[20:36] <Darby_Crash> i'm compiling ffmpeg now
[20:50] <IRCmonkey> Hey, I was wondering. How do I configure ffmpeg. I was given a list of things to configure
[20:50] <IRCmonkey> http://pastebin.com/4pzNFYr9
[20:51] <IRCmonkey> I'm not sure how to input this into cmd in order to configure ffmpeg
[20:53] <c_14> ./configure --foo --bar --baz --klunk
[20:54] <JEEB> of course, you actually have to have all of the libraries already built and ready for that
[20:55] <JEEB> (that thing has a lot of kitchen sink in it and stuff you would pretty much never use)
[21:11] <Darby_Crash> c_14
[21:12] <Darby_Crash> don't compile ffplay
[21:12] <Darby_Crash> :(
[21:17] <Darby_Crash> i can't compile ffplay
[21:23] <c_14> Darby_Crash: config.log ?
[21:29] <Darby_Crash> ok c_14
[21:32] <IRCmonkey> 'configure' is not recognized as an internal or external command, operable program or batch file.
[21:32] <IRCmonkey> Any more help?
[21:32] <Darby_Crash> https://clbin.com/xKNeh
[21:32] <Darby_Crash> c_14
[21:33] <iive> IRCmonkey:  ./configure
[21:33] <IRCmonkey> '.' is not recognized as an internal or external command, operable program or batch file.
[21:34] <PovAddict> you're on Windows?
[21:34] <IRCmonkey> I'm using cmd if that makes a difference
[21:34] <IRCmonkey> yes
[21:34] <PovAddict> you need to run it in a msys shell
[21:34] <IRCmonkey> I don't know how to do that
[21:34] <PovAddict> did you install mingw?
[21:34] <IRCmonkey> I'm gonna guess no?
[21:34] <iive> mingw64 (it can compile 32 bit programs too).
[21:35] <IRCmonkey> uh, I'm not sure if I'm compiling anything
[21:35] <PovAddict> yes you are
[21:36] <PovAddict> 'configure' prepares to compile
[21:36] <IRCmonkey> oh
[21:36] <IRCmonkey> I didn't realize
[21:37] <IRCmonkey> can I do this inside of cmd?
[21:37] <PovAddict> also your paste already expired
[21:38] <IRCmonkey> This is part of the instructions I was given
[21:38] <IRCmonkey> http://pastebin.com/hfgbn3EN
[21:38] <Darby_Crash> https://clbin.com/xKNeh c_14 please help me
[21:38] <IRCmonkey> it's the part that tell me how to configure ffmpeg
[21:38] <IRCmonkey> but it's not really helpful
[21:39] <Darby_Crash> in the config.log file nothing is changed
[21:39] <c_14> Darby_Crash: what does `pkg-config --libs --cflags sdl' output? Assuming you set PKG_CONFIG_PATH
[21:40] <Darby_Crash> root at localhost:/sdcard# pkg-config --libs --cflags sdl
[21:41] <Darby_Crash> -D_GNU_SOURCE=1 -D_REENTRANT -I/usr/include/SDL  -lSDL
[21:41] <Darby_Crash> root at localhost:/sdcard#
[21:42] <c_14> Are you sure you set your PKG_CONFIG_PATH correctly ?
[21:43] <Darby_Crash> take a look to this
[21:46] <Darby_Crash> http://pastebin.com/LkMsxRXB
[21:47] <Darby_Crash> c_14 with this script ffplay is compiled
[21:47] <Darby_Crash> but i want to fix the other script
[21:48] <Darby_Crash> because this is unstable
[21:54] <Darby_Crash> c_16 maybe the problem is here
[21:54] <Darby_Crash> export PKG_CONFIG_PATH="$TARGET_DIR/lib/pkgconfig"
[21:55] <c_14> Rebuild SDL with --disable-rpath
[21:55] <c_14> That's what the internet is telling me.
[21:56] <PovAddict> IRCmonkey: you need to install the mingw compiler first
[21:56] <PovAddict> you're being told to compile ffmpeg with that same configuration?
[21:57] <PovAddict> oh and as JEEB said you have to compile a ton of other things first
[21:57] <PovAddict> if you want to --enable-lzma you need to compile lzma first
[22:11] <Popara> Hello,i would like to see a small patch (improvement) in the upcoming verison of FFmpeg or even a private patch if it is allowed. I'm willing to pay
[22:11] <Popara> if you are interested please message me
[22:14] <IRCmonkey> uhh
[22:14] <IRCmonkey> I'm using the static build of this http://ffmpeg.zeranoe.com/builds/
[22:14] <IRCmonkey> does that help?
[22:14] <PovAddict> IRCmonkey: you can't reconfigure an existing build of ffmpeg, in that sense
[22:15] <IRCmonkey> then, I'm lost
[22:15] <PovAddict> that thing you pasted is compile-time configuration, if you want to configure it that way you have to compile it from source
[22:15] <IRCmonkey> uh
[22:16] <IRCmonkey> How would I do that again?
[22:17] <PovAddict> on windows?
[22:17] <IRCmonkey> Yes
[22:17] <PovAddict> first you install mingw, then you compile from source all the libraries you need for your configuration (help with which is outside the scope of this channel), then you ./configure ffmpeg
[22:18] <PovAddict> why can't you use those static builds?
[22:18] <PovAddict> why do you need that exact configuration?
[22:18] <IRCmonkey> I don't know. I installed a static build
[22:19] <IRCmonkey> but I figured it was configured differently
[22:19] <IRCmonkey> than what I needed
[22:19] <PovAddict> it is
[22:19] <IRCmonkey> So I have mingw
[22:19] <PovAddict> "configured differently than what I needed" why do you need that configuration exactly?
[22:19] <IRCmonkey> So that I can use it in conjunction with an irc bot
[22:20] Action: PovAddict picks a random switch
[22:20] <IRCmonkey> ?
[22:20] <PovAddict> you need a ffmpeg with libcaca support for an irc bot?
[22:20] <IRCmonkey> idk
[22:20] <PovAddict> and frei0r?
[22:20] <IRCmonkey> I didn't build the bot
[22:21] <PovAddict> so you don't know what you're doing and you don't know why you're doing it :P
[22:21] <IRCmonkey> Well, I know what I intend to use the bot for
[22:21] <IRCmonkey> I just don't know how to set it up
[22:21] <PovAddict> an irc bot needing ffmpeg sounds weird
[22:21] <IRCmonkey> it's really the other way around
[22:22] <IRCmonkey> ffmpeg plays something, and the bot tells the channel what's it streaming via irc
[22:23] <PovAddict> have you tried if the static build works?
[22:23] <IRCmonkey> Not yet, no
[22:23] <PovAddict> start there :)
[22:23] <IRCmonkey> k
[22:26] <IRCmonkey> uh, weird question
[22:26] <IRCmonkey> how do I run a python script from cmd
[22:26] <PovAddict> do you have python installed?
[22:26] <IRCmonkey> yes
[22:26] <IRCmonkey> in the instructions, it says do ./streambot.py -i <ini file>
[22:26] <PovAddict> yeah that's unixy instructions
[22:26] <IRCmonkey> but ./ is for linux I think
[22:26] <IRCmonkey> yeah
[22:27] <PovAddict> try "python streambot.py ..:"
[22:27] <IRCmonkey> k
[22:27] <PovAddict> otherwise use full path to python.exe
[22:27] <PovAddict> C:\Python\python.exe streambot.py or wherever it is
[22:32] <IRCmonkey> this is what I got
[22:32] <IRCmonkey> http://pastebin.com/jH3uvZLM
[22:46] <IRCmonkey>  Does any of that make sense, PovAddict ?
[22:46] <PovAddict> no, I'd have to know about the bot
[22:53] <IRCmonkey> what would you need to know?
[23:39] <xreal> Have the sources of the zeranoe-builds been released anywhere?
[23:40] <xreal> I mean, the build bot.
[23:41] <xreal> Is there a fast way to "re-transcode" MPEG2? I remeber DVD-shrink and other applications in the past.
[23:48] <BtbN> xreal, i don't think the sources are modified at all
[23:49] <BtbN> don't the builds even have the git commit in their name?
[23:49] <xreal> BtbN: I didn't mean ffmpeg sources. I mean the building script :D
[23:49] <xreal> BtbN: I've created my own in cygwin, but I'd like to see a professional one, like this one.
[23:50] <BtbN> Well, won't do much more than running configure and make
[23:51] <xreal> BtbN: yeah, but isn't he using VC++ or Intel as compiler?
[23:52] <xreal> BtbN: might perform better on Intel CPUs.
[23:52] <BtbN> just mingw as far as i'm aware
[23:53] <BtbN> doesn't matter for performance at all
[23:53] <BtbN> all performance critical parts in ffmpeg are hand written in yasm anyway
[23:53] <xreal> BtbN: I've tested several applications, which are CPU intensive and Intel optimized often has more speed when using pure C/C++ code.
[23:53] <xreal> BtbN: ah okay, yasm doesn't count :)
[00:00] --- Mon Jan 12 2015


More information about the Ffmpeg-devel-irc mailing list