[Ffmpeg-devel-irc] ffmpeg.log.20121118

burek burek021 at gmail.com
Mon Nov 19 02:05:01 CET 2012


[00:38] <strange0ne> such a weird question
[00:38] <strange0ne> is ther a gui for ffmpeg or is the cli just better?
[00:41] <frecel> hey I need some help with x11grab, anyone willing to help?
[00:43] <klaxa> what kind of help do you need?
[00:44] <klaxa> frecel: your request for help is too vague... really...
[00:46] <frecel> I'm trying to do a screen capture and I set fps at 30 but it records at about 20 when I'm just on the desktop and at 10 fps when I start teeworlds
[00:46] <klaxa> what kind of cpu do you have?
[00:46] <frecel> I started htop during recording and none of the cores goes over 50%
[00:47] <klaxa> did you add -threads 0 ?
[00:47] <frecel> amd phenom 2 black 3.4ghz
[00:47] <klaxa> try adding -threads 0
[00:49] <frecel> heres the script I'm using
[00:49] <frecel> http://pastebin.com/DMkebZKe
[00:49] <klaxa> also what resolution are you recording with, what codec, to what medium... would be easiest to provide the command you run actually
[00:49] <klaxa> ah great
[00:49] <frecel> where do I add that
[00:51] <klaxa> it's already set to 4 threads that should actually be enough... hmm...
[00:53] <klaxa> i assume the machine has 4 cores?
[00:53] <klaxa> what is your upload speed?
[00:54] <cbsrobot> frecel: see https://ffmpeg.org/trac/ffmpeg/wiki/StreamingGuide
[00:54] <cbsrobot> just before Latency
[00:56] <frecel> yes its a 4 core cpu and I get about 1Mbps upload
[00:59] <klaxa> maybe just lower the bitrate? i'm not sure that that's it though
[00:59] <klaxa> and replace the 4 in the thread argument by 0?
[01:02] <frecel> what will the thread argument changed to 0 do?
[01:02] <klaxa> ffmpeg will be like: OMG I CAN HAS AS MANY THREADS AS I WANT?!
[01:03] <klaxa> so basically, it'll increase threads to whatever the amount is that suits best
[01:05] <frecel> neat
[01:09] <frecel> now the rtmp server won't auth me, it looks like I have to email the server fairies about that
[01:16] <frecel> I got some hiccups but on average I managed to make about 21 fps
[01:16] <frecel> thanks
[01:16] <frecel> any idea how to make it up to 30?
[01:19] <klaxa> if your cpu is not the limiting factor i actually don't know
[01:20] <frecel> I'm wondering if turning compositing off would help
[01:26] <frecel> i got about the same results without compositing
[01:26] <frecel> do you think this is watchable? http://www.twitch.tv/polishedperspective/b/340947212
[01:26] <klaxa> yeah i think compositing is offloaded to gpu in most cases
[01:26] <frecel> keep in mind this is just a video test, I'll start adjusting audio levels after I'm happy with the video
[01:28] <klaxa> frecel: i'd say so... (also i barely play the standard mod, but dynamic camera? seriously?)
[01:28] <cbsrobot> frecel: try ultrafast
[01:28] <cbsrobot> instead of fast
[01:33] <frecel> I've been playing with dynamic camera since way before they changed the name to teeworlds
[01:34] <frecel> it's just a force of habit now
[01:34] <klaxa> frecel: kk... well i play instagib, really hard to keep up with dynamic camera these days... dynamic camera binds to switch between static and dynamic are pretty popular by now
[01:35] <frecel> the framerate is about the same on ultrafast
[01:36] <klaxa> get rid of kde maybe? :D
[01:37] <frecel> klaxa: I used to play a lot of instagib some time ago, I don't want to be one of those people who say that it requires less skill but you have to admit there are a lot of lucky frags in insta
[01:37] <klaxa> well... the thing is, you can't really steal kills...
[01:38] <klaxa> and it's way faster paced
[01:39] <frecel> you can have fast paced vanilla game, you just have to find 10 skilled players and somehow convice them to play on the same server :D
[01:39] <klaxa> :D
[01:42] <frecel> also, not ditching KDE, I'm in love with it now
[01:42] <klaxa> well you could like... start a plain X server just for streaming
[01:44] <frecel> good idea
[01:52] <josefnpat> Oh gods thank you
[01:52] <josefnpat> I'm running ubunut 10.04 and I can't seem to get the --enable-x11grab flag to work.
[01:53] <josefnpat> I've cloned from the sites repo, and if I grep the configure output, all I get is "x11grab"
[01:53] <klaxa> the binary package distributed with 10.04 should have x11grab enabled
[01:53] <josefnpat> but when i make install it, i keep getting the error "Unknown input format: 'x11grab'
[01:54] <josefnpat> ok, that's cool. Maybe I'm doing something wrong with the line of code then.
[01:54] <klaxa> i'm actually not that sure now...
[01:54] <josefnpat> http://hastebin.com/raw/jilejifago
[01:54] <klaxa> but put your command and ffmpegs output on pa-- perfect thanks
[01:55] <josefnpat> :)
[01:55] <klaxa> output would be great
[01:55] <josefnpat> from what?
[01:55] <klaxa> oh wait
[01:55] <josefnpat> tghe configure?
[01:55] <klaxa> nvm
[01:56] <klaxa> yeah looks like that wasn't copiled in in that avconv build
[01:56] <klaxa> (also why again did ubuntu replace ffmpeg with avconv?)
[01:56] <josefnpat> so i need to recompile avconv?
[01:56] <josefnpat> lol
[01:57] <josefnpat> I was wondering the same thing. I thought avconv was a part of ffmpeg
[01:57] <josefnpat> http://hastebin.com/raw/xonawaxudo
[01:57] <josefnpat> There's my compile
[01:57] <josefnpat> I see the indev x11grab
[01:57] <josefnpat> but not the x11_grab_device
[01:58] <klaxa> maybe a dumb question, but you did run make install or ran the compiled binary specifically?
[01:59] <josefnpat> I did `sudo make install`
[01:59] <josefnpat> I made sure to `apt-get remove ffmpeg` beforehand
[01:59] <josefnpat> along with apt-get autoremove
[01:59] <josefnpat> shoudl i have --purge'd it?
[01:59] <klaxa> nah should have been enough...
[02:00] <klaxa> hmm... sorry no idea, maybe wait for someone else that knows more to show up
[02:00] <josefnpat> I appreciate the help.
[02:00] <josefnpat> Can you point me to the source code for avconv?
[02:00] <josefnpat> found it!
[02:00] <josefnpat> Thanks anywa.
[02:00] <josefnpat> Thanks anyway.
[02:00] <klaxa> you're welcome
[02:01] <josefnpat> I'm going to rebuild libav if you think that's the culprit
[02:01] <klaxa> i actually don't know the specific differences between avconv and ffmpeg, but at some point i read in here that avconv ships broken ffmpeg binaries, so you might want to try to compile ffmpeg instead :X
[02:01] <josefnpat> I did compile ffmpeg from the git remove they had :(
[02:02] <frecel> I tried playing Quake Live, I got about 6fps :(
[02:02] <klaxa> also this information could be totally wrong, but i think i also read that avconv is frowned upon kinda...
[02:02] <josefnpat> it is?
[02:02] <klaxa> oh... you did replace the avconv command by ffmpeg then, did you?
[02:02] <josefnpat> Wait, can I do that?
[02:02] <klaxa> i mean in the script
[02:02] <josefnpat> I will do that.
[02:02] <josefnpat> Let's see what happens
[02:03] <klaxa> if you installed ffmpeg with "make install" it should install ffmpeg and not avconv
[02:03] <josefnpat> iiiii
[02:03] <josefnpat> it did something.
[02:03] <josefnpat> something I think I cna fix!
[02:13] <josefnpat> klaxa: that was the issue! Thanks!
[02:13] <klaxa> nice :)
[02:14] <josefnpat> Unfortunatley, it seems to be spitting out nothign other than black
[02:14] <josefnpat> perhaps you can tell me how to output this to device instead of URL?
[02:16] <josefnpat> When i run the output of this, I get this error many times, "Reference 2 >= 2
[02:16] <josefnpat> error while decoding MB 67 5, bytestream (111)
[02:18] <klaxa> replace the rtmp url by output.flv
[02:18] <josefnpat> ok
[02:19] <josefnpat> 50.116.63.25/public/test.flv
[02:19] <josefnpat> This is what I'm getting
[02:20] <josefnpat> http://This is what I'm getting
[02:20] <josefnpat> http://50.116.63.25/public/test.flv
[02:21] <klaxa> hmm... the video looks fine, that test.flv you play later looks not that good
[02:23] <josefnpat> ok, i was able to get the video!
[02:23] <josefnpat> By commenting out lines
[02:24] <josefnpat> it's a matter of deduction at this point.
[02:32] <josefnpat> klaxa: http://www.twitch.tv/josefnpat
[02:32] <josefnpat> I did it!
[02:33] <klaxa> heh, nice
[02:33] <josefnpat> <3 much love, thanks
[02:33] <josefnpat> Now to write this down ...
[02:33] <josefnpat> lol
[02:40] <CalimeroTeknik> hum, excuse me, is concat:file1\|file2 compatible with multi-track files?
[02:40] <CalimeroTeknik> it stops right after the first file
[02:40] <CalimeroTeknik> using ffmpeg -i concat:part1.mkv\|part2.mkv\|part3.mkv -vn -acodec copy -scodec copy out.mkv
[02:41] <klaxa> i think concat only works for files that actually are desinged to be concanated (e.g. mpeg1) see if mkvmerge works better for you
[02:41] <CalimeroTeknik> the result is equivalent to the first file without video
[02:41] <CalimeroTeknik> ah, good idea, thanks
[03:01] <josefnpat> https://gist.github.com/4102798#file_justin.tv.txt
[03:02] <josefnpat> If anyone was intrested in my attempt to document my journey
[03:03] <frecel> josefnpat, was it necessary to build ffmpeg from source?
[03:03] <josefnpat> maybe
[03:03] <josefnpat> Lol
[03:03] <josefnpat> "Note: some of this may be superflous"
[03:03] <josefnpat> I was going to get a VM
[03:04] <josefnpat> and repeat the process
[03:04] <josefnpat> superfluous
[03:08] <frecel> can you go to http://www.twitch.tv/polishedperspective and tell me how is the overall quality of the stream?
[03:09] <klaxa> frecel: desktop looks good :P
[03:10] <josefnpat> what v of linux is he running?
[03:10] <klaxa> mint 13 kde he said
[03:11] <josefnpat> ok, cool
[03:12] <frecel> http://pastebin.com/DMkebZKe
[03:12] <klaxa> framerate is a bit low... no idea how to fix that though
[03:12] <josefnpat> nice
[03:12] <josefnpat> gotcha
[03:12] <josefnpat> thank you!
[03:15] <klaxa> i don't really think changing codecs helps :S
[03:16] <klaxa> frecel: it appears that it's really capped to 20 fps
[03:17] <klaxa> are you sure you didn't accidentally cap it at 20 instead of 30?
[03:18] <klaxa> frecel: you can drop the '"' around "$FPS" within the command line right after the -r
[03:20] <klaxa> weird...
[03:22] <klaxa> it actually looks like less than 24 fps, or that's flash lemme try to grab the rtmp directly
[03:24] <klaxa> oh lol yeah pretty much was flash
[03:24] <klaxa> looks watchable at least
[03:29] <josefnpat> are you the raptor from #ludumdare raptor67782 ?
[03:30] <klaxa> ha lol the stream is seconds behind
[03:31] <klaxa> well yeah doesn't really matter, but the thing with the low fps is really weird
[03:31] <klaxa> can't explain it
[03:32] <klaxa> baaaaad idea frecel :P
[03:33] <klaxa> yeah that gets converted automatically
[03:33] <klaxa> not really
[03:34] <klaxa> your computer screen is in rgb colors, h264 wants yuv though, therefore ffmpeg converts it
[03:34] <josefnpat> god damn smurfs
[03:35] <klaxa> switching the second monitor off might help, see if you can get a plain X session for streaming too (that would get rid of the KDE overhead i guess)
[03:36] <klaxa> yeah... maybe it's just monitor I/O, i bet you can easily get 30 fps if you change your resolution to 720p
[03:37] <klaxa> since you resize anyways it wouldn't make a lot of a difference for the stream
[03:39] <klaxa> frecel: ^
[03:41] <klaxa> frecel: you have to adjust the script to record 720p
[03:41] <klaxa> then restart
[03:41] <klaxa> the script i mean
[03:43] <klaxa> before you start teeworlds, edit your config for it to use 720p frecel
[03:44] <klaxa> frecel: lower resolution means less stuff to process, means more time for more frames, means more fps :P
[03:45] <klaxa> ah nice, very smooth frecel
[03:47] <klaxa> you could even cap it to 30 fps then you won't have the dropped frames
[03:48] <klaxa> perfect too, very smooth
[03:56] <frecel> thanks for the help
[03:56] <frecel> now I can start to make millions as a professional game caster :D
[03:56] <klaxa> :D
[04:08] <Xgates> hi guys
[04:09] <Xgates> does 0.11.2 & 1.0 use libavcodec54?
[04:11] <klaxa> just run ffmpeg with no arguments, it should tell you
[04:11] <klaxa> after the configuration output it shows for me:
[04:11] <klaxa>   libavcodec     54. 59.100 / 54. 59.100
[04:11] <klaxa>   libavformat    54. 29.104 / 54. 29.104
[04:11] <klaxa> this is version 1.0
[04:26] <Wolfed> hmph. Google fails to assist me with an error while compiling. Something regarding libavutil.
[04:29] <Xgates> klaxa: I know how to check it, I wanted to know what versions before I compile one of them which I should pick...
[04:29] <Wolfed> Can I paste it here by any chance? Its one line I think
[04:30] <Xgates> I need 54 wasn't sure if it's in both versions...
[04:30] <klaxa> ah... dunno
[04:30] <Xgates> k
[04:31] <klaxa> Wolfed: to be sure paste all of the log i guess?
[04:31] <klaxa> on pastebin or something
[04:31] <Wolfed> ./usr/bin/ld: libavutil/intfloat_readwrite.o: relocation R_X86_64_32 against `a local symbol' can not be used when making a shared object; recompile with -fPIC
[04:31] <Wolfed> Thats the error. I can pastebin the screen though.. or what I can see from it. But thats mainly it
[04:32] <klaxa> did you try to add -fPIC to the CC flags in the makefile or something?
[04:32] <klaxa> i mean that's what that line suggests, no?
[04:32] <Wolfed> Well.. yeah... hum.. I guess
[04:32] <Wolfed> I have no idea what its suggesting really :P
[04:32] <klaxa> heh
[04:32] <klaxa> in the makefile there should be some declaration for CC
[04:33] <klaxa> i.e. something like CC = "some stuff"
[04:33] <Wolfed> ffmpeg's or libavutil's?
[04:33] <klaxa> uhh...
[04:33] <klaxa> i'd add it to libavutils's
[04:33] <Wolfed> Alrighto
[04:33] <klaxa> *-s
[04:33] <Wolfed> I'll give it a try
[04:33] <klaxa> i'm no expert though, wild guess here
[04:34] <Wolfed> Nope, nothing like that in libavutil
[04:34] <Wolfed> I mean no CC
[04:34] <Wolfed> I'll check out ffmpeg's
[04:34] <Wolfed> Not there either.. Well there is one $(CC)
[04:35] <Wolfed> But I dont think I cans et something there
[04:37] <Wolfed> I'll try adding -fPIC to --extra-cflags
[04:37] <Wolfed> See if that works
[04:38] <klaxa> yeah
[04:41] <Wolfed> Alright, that didnt work. Hmph
[04:41] <klaxa> :S
[04:41] <klaxa> different errors?
[04:42] <Wolfed> Nope. same one
[04:42] <Wolfed> How annoying...
[04:42] <klaxa> in that case it's not really recompiling, heh
[04:42] <klaxa> um... i just thought of something different
[04:42] <Wolfed> Is it possible to make ffmpeg reload the codecs database?
[04:42] <Wolfed> Like available codecs?
[04:43] <klaxa> cd indo the libavutil directory and run gcc -c -o intfloat_readwrite.o intfloat_readwrite.c
[04:43] <klaxa> *into
[04:43] <Wolfed> Alright
[04:43] <klaxa> um... no idea, run ./configure again?
[04:43] <klaxa> ah wait
[04:43] <klaxa> add the -fPIC of course
[04:43] <Wolfed> lol
[04:44] <Wolfed> Done
[04:44] <Wolfed> Well, that didnt work
[04:44] <klaxa> no idea then
[04:44] <Wolfed> gcc -c -fPIC -o intfloat_readwrite.o intfloat_readwrite.c
[04:44] <Wolfed> Right?
[04:44] <klaxa> maybe wait for the devs to show up
[04:44] <klaxa> well that would be my guess :P
[04:45] <Wolfed> Should I try Dev channel? Or thats not user-compiling support?
[04:45] <klaxa> just idle in here, ask again when you see other people than me talking
[04:45] <Wolfed> nvm. Topic there answered
[04:45] <Wolfed> :P
[04:45] <Wolfed> Will do. Thanks :)
[04:48] <klaxa> here is some stuff on the -fPIC flag if that's relevant to your interests http://gcc.gnu.org/onlinedocs/gcc/Code-Gen-Options.html#Code-Gen-Options
[04:49] <Wolfed> Thanks. I had read it though
[04:49] <Wolfed> :P
[04:49] <klaxa> ah kk :P
[05:05] <Wolfed> Well. I got a new error no
[05:05] <Devon> i get this error:  Can't write packet with unknown timestamp av_interleaved_write_frame(): Invalid argument
[05:05] <Devon> what does this mean
[05:28] <Wolfed> Gosh, if it didnt take so long for it to compile everytime -_-
[05:32] <Wolfed> FFS Finally
[05:32] <Wolfed> -_-
[05:33] <Wolfed> Wow.. how gay...
[05:37] <retard> do you mean gay in the good way or gay in the bad way?
[05:38] <Wolfed> In the way a non-living object should receive it when not wanting to work.
[05:38] <retard> i'm not sure i understand
[05:39] <Wolfed> I guess that was the point :P
[05:53] <Wolfed> What'd this mean: [libvpx @ 0x86df20] Failed to initialize encoder: ABI version mismatch
[05:54] <Wolfed> I compiled both versions...
[05:54] <Wolfed> I mean both, libvpx and ffmpeg
[05:55] <klaxa> i think ABI version is something hardware specific
[05:55] <klaxa> incorrectly compiled binaries would be my guess
[05:55] <klaxa> but that's only a guess
[05:55] <Wolfed> hmph...
[05:56] <Wolfed> As in compiled for x86 when it was x64? O.o
[05:56] <juanmabc> ABI is aplication binary interface
[05:56] <juanmabc> happens with wrong versions linked
[05:57] <Wolfed> Oh I see... Not that I have any idea what are you talking about, though :P
[05:58] <juanmabc> are you sure 1) libvpx installed is ok for ffmpeg build, and 2) you are linking to own libvpx and not old distro one?
[05:59] <Wolfed> I never installed libvpx through yum. As for if its ok for ffmpeg. I just installed the latest version available I think
[05:59] <Wolfed> *Double checks*
[06:00] <Wolfed> Ok, no idea which version I installed. Just noticed I have two of them
[06:00] <juanmabc> worst case scenario: abi mismatch, he
[06:00] <juanmabc> check that
[06:01] <Wolfed> Should I just try to reinstall whichever the latest it?
[06:01] <Wolfed> is*
[06:03] <Wolfed> There we go again, recompiling
[06:05] <Wolfed> Apparently I installed one from Aug 2011
[06:08] Action: Wolfed jumps off a cliff
[06:08] <Wolfed> Ok, recompiled both libvpx from the git, did the same with ffmpeg
[06:08] <Wolfed> Did not work
[06:09] <Wolfed> I'll drop by tomorrow I guess, way too late and I dont seem to be going anywhere
[06:09] <Wolfed> Thanks.
[09:40] <K-Rich> Hi all i have a small issue when recording my desktop, it records, but the output flickers when some things happen, i assume this is cause i'm using a compositing wm, is there a way to compensate for this, my current command is, ffmpeg -f x11grab -s $size -r 30 -qscale 1  -i :0.0 $video
[13:22] <madabang> hi, all. I read http://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuideLucid, and I'd like to know , when installing x264 with checkinstall , why the pkgversion takes column 4 and column 5 from version.sh? Someone pls let me know of this. TIA.
[13:27] <JEEBsv> version.sh creates the version entry via git or otherwise
[13:27] <JEEBsv> so it runs it and sets the package version
[13:27] <JEEBsv> i would think so at least
[14:26] <schreber> Is it possible to do a x11grab and encode directly to libx264 or would it be better (or should I say less taxing) to capture in another format then re-encode to the specific codec + container?
[14:55] <JEEBsv> schreber: libx264's lossless can be fast, but I usually use less taxing encoders like ffvhuff/huffyuv for a quick lossless
[14:56] <schreber> JEEBsv, do you re-encode after the fact or do you find that solution acceptable ?
[14:57] <JEEBsv> usually the end result I need is lossy so yes, I encode to lossy with relatively high compression settings after the fact
[14:59] <schreber> JEEBsv, okay, thanks
[16:51] <Wolfed> Apparently ffmpeg is attempting to use 0.9.7 instead of the latest version I installed. Is there any workaround for this?
[16:52] <JEEBsv> try setting --extra-cflags="-I/your/prefix/include" and --extra-ldflags="-L/your/prefix/lib" in configure?
[16:52] <Wolfed> I have
[16:52] <Wolfed> Didnt seem to work.
[16:54] <JEEBsv> well, I guess then you'd have to either have gcc forget about its built-in paths for headers and libraries, or just remove the older version if possible. That is, if you are doing it all correctly. With static libraries and libx264 f.ex. I've so far been able to build with my own newly built library into a prefix into my home folder f.ex.
[16:56] <Wolfed> Should both, cflags and ldflags point to where libvpx.a is?
[16:56] <JEEBsv> no
[16:57] <JEEBsv> ldflags should be pointing to the /lib of the prefix you installed to
[16:57] <JEEBsv> (the -L flag adds it to the library search path9
[16:57] <Wolfed> Yeah, its pointing to /lib while cflags is pointing to /include
[16:57] <JEEBsv> the -I compiler flag adds the folder to the header search path (/include in your install prefix)
[16:57] <JEEBsv> yes, that's correct
[16:57] <Wolfed> however libvpx doesnt seem to install anything into /include
[16:57] <JEEBsv> o_O
[16:58] <JEEBsv> then you wouldn't be able to use it :D
[16:58] <Wolfed> lol
[16:58] <Wolfed> Alright... so apparently its a libvpx install issue...
[16:58] <JEEBsv> because the header defines the functions given by the library
[16:59] <JEEBsv> and then you link the library in so that every symbol needed to make a working binary then is there
[16:59] <Wolfed> I get that. I am not a pro at this though, just minimum knowledge :P
[16:59] <JEEBsv> and yes, that would be the reason why it isn't finding your custom install
[16:59] <JEEBsv> if it can't find your newer header
[16:59] <Wolfed> So wait, include should be pointing to binaries?
[16:59] <JEEBsv> (I will guess the symbols are versioned9
[16:59] <JEEBsv> Wolfed: no
[16:59] <Wolfed> Gosh...
[16:59] <Wolfed> Alright
[16:59] <JEEBsv> to /include is correct, of the install prefix
[17:00] <Wolfed> include is the sources, right?
[17:00] <Wolfed> I mean, sources go in /include
[17:00] <JEEBsv> it is the headers
[17:01] <Wolfed> Checking out the configure script to see if I can figure out a way to get those installed
[17:01] <Wolfed> Can I set more than one path to the headers?
[17:01] <JEEBsv> headers define functions and possible other needed code, and then libraries in lib have stuff you can link to then most probably gain the needed symbols to use those functions defined in the headers
[17:01] <JEEBsv> yes --extra-cflags="-I/herp/derp/include -I/hjurpa/djurpa/include" f.ex.
[17:02] <JEEBsv> anyways, your libvpx install probably got fucked up somewhere if you didn't get all of the needed stuff installed
[17:02] <JEEBsv> also you did actually set a prefix and install, right?
[17:03] <JEEBsv> and not only do 'make' and set the cflag/ldflag to the source folder? :D
[17:03] <Wolfed> on ffmpeg?
[17:03] <JEEBsv> ...
[17:03] <Wolfed> No, on libvpx I just went ./configure && make && make install
[17:04] <Wolfed> Although, it only installed binaries and libvpx.a at /lib
[17:04] <JEEBsv> ok, so you used make install at least
[17:04] <JEEBsv> binaries should've gone to /bin
[17:04] <Wolfed> Well they did
[17:04] <JEEBsv> libraries go to /lib
[17:04] <Wolfed> binaries at /bin and libvpx.a to /lib
[17:04] <Wolfed> And there one another file that went to /lib
[17:04] <JEEBsv> and nothing to /include , not even a folder possibly?
[17:04] <JEEBsv> a .so?
[17:04] <JEEBsv> you generally don't want that with custom builds :D
[17:04] <JEEBsv> (shared libs)
[17:06] <JEEBsv> seems like libvpxenc is including vpx/vpx_encoder.h
[17:06] <Wolfed> Yeah, no .so (I thought that was a bit weird, but then again, I dont know much)
[17:06] <Wolfed> http://pastebin.com/F91PMpgG
[17:06] <Wolfed> Thats what install did
[17:06] <Wolfed> make install*
[17:07] <JEEBsv> .so are usually more of a PITA if you have no idea about how to deal with them
[17:07] <JEEBsv> o____O
[17:07] <JEEBsv> seriously no headers o_O
[17:07] <JEEBsv> wtf
[17:07] <Wolfed> At least I am not the only one with that face
[17:07] <Wolfed> :D
[17:07] <JEEBsv> yeah, no wonder it doesn't get found
[17:08] <Wolfed> I was checking the configure script and I couldnt find anything about setting header's path for libvpx
[17:09] <JEEBsv> check ./configure --help
[17:09] <Wolfed> I seem to have a lot of header files around vp8/common, vp8/encoder
[17:10] <JEEBsv> yeah, but those should then be installed
[17:10] <JEEBsv> f.ex. you should have something like /include/vpx/vpx_encoder.h
[17:11] <Wolfed> Yeah No include
[17:11] <Wolfed> --enable-install-srcs       control whether sources are installed
[17:11] <Wolfed> Thats the closest thing to perhaps, getting something new installed
[17:11] <Wolfed> lol
[17:12] <JEEBsv> that might not be it... anyways, I'm deeply surprised that the library gets installed and the headers don't...
[17:12] <Wolfed> The only include I have comes from libvpx/nestegg, libvpx/vpx_mem and libvpx/vpx_scale
[17:13] <Wolfed> I mean only ./include/
[17:13] <JEEBsv> of the prefix?
[17:13] <Wolfed> No, onlie include directories :P
[17:13] <Wolfed> within libvpx...
[17:13] <JEEBsv> lol
[17:13] <JEEBsv> yeah, you shouldn't be looking at that too much
[17:14] <JEEBsv> how a project sets its source tree and how it installs its headers don't usually have much similarity
[17:14] <JEEBsv> bug a libvpx channel I guess
[17:15] <Wolfed> lol
[17:15] <Wolfed> I'll give it a try there
[17:15] <JEEBsv> #vp8 on freenode
[17:15] <Wolfed> Yeah I am there
[17:15] <JEEBsv> k
[17:16] <JEEBsv> I know that x264 f.ex. by default only installs its binary and no headers or libraries. If you --enable-shared or --enable-static it then builds either a static or shared lib, and installs headers too
[17:16] <JEEBsv> which makes sense
[17:17] <JEEBsv> installing the library while not installing the header is relatively crazy :D
[17:17] <Wolfed> I'll try building libvpx with --enabled-shared then
[17:17] <Wolfed> See if that brings up anything new
[17:17] <JEEBsv> you don't want shared
[17:17] <JEEBsv> you want --disable-shared and --enable-static
[17:17] <Wolfed> Ah
[17:17] <Wolfed> lol
[17:18] <Wolfed> Alright, giving it a try
[17:20] <Wolfed> Nope, didnt work :P
[17:20] <JEEBsv> not surprising
[17:20] <JEEBsv> anyways, you're better off poking #vp8
[17:23] <Wolfed> Yeah, alredy stated the issue, lets see if anyone can help me out
[17:36] <Wolfed> Now thats weird, I just got a fresh copy off git and it compiled the headers
[17:39] <Wolfed> Wow.. a new error. How mean
[17:49] <Wolfed> FOR FRIGGING SAKE OF THE OLD LORDS
[17:50] <Wolfed> Alright. Installed webm with headers, reconfigured and recompiled ffmpeg. It still tries 0.9.7...
[17:51] <Wolfed> Is there anyway to force it to use one specific version?
[17:54] <JEEBsv> Wolfed: did you set the folders added to the search paths correctly
[17:54] <Wolfed> Yes
[17:55] <JEEBsv> also did you compile static or shared
[17:55] <JEEBsv> aka do you have a shared lib there?
[17:55] <Wolfed> http://pastebin.com/pz1PEPYw
[17:55] <Wolfed> libvpx was compiled with a simple ./configure
[17:55] <JEEBsv> yeah, then it has that
[17:55] <Wolfed> So go disabled-shared enablñe-sgtatic?
[17:55] <JEEBsv> sudo ldconfig
[17:56] <JEEBsv> and then retry with ffmpeg
[17:56] <Wolfed> Oh dude.. I forgot about that
[17:56] <Wolfed> lol
[17:56] <JEEBsv> which is why static libs rock for custom builds :P
[17:56] <Wolfed> And rebuild ffmpeg
[17:56] <Wolfed> right?
[17:56] <Wolfed> recompile*
[18:00] <Wolfed> Well, nope. keeps going with 0.9.7
[18:02] <JEEBsv> dunno then
[18:02] <JEEBsv> there is a cflag that will make gcc ignore all usual paths
[18:02] <JEEBsv> but you'll have tons of fun entering the standard ones manually in that case :D
[18:03] <Wolfed> LOL
[18:03] <Wolfed> After like 8 hours... what harm can a few extra hours do ¬¬'
[18:03] <Wolfed> :P
[18:03] <JEEBsv> you should find it from gcc's documentation with some gooling
[18:03] <JEEBsv> *googling
[18:16] <DelphiWorld> Hello
[18:16] <DelphiWorld> how do i read dvd's using ffmpeg ?
[18:26] <DelphiWorld> guys how to rippe dvd using ff?
[18:26] <microchip_> you can't atm, ffmpeg can't read DVDs directly
[18:26] <microchip_> i think someone is working on that, tho
[18:33] <DelphiWorld> microchip_: so how to rippe ?
[18:35] <Wolfed> You'd have to turn the vob files into something else
[18:37] <DelphiWorld> Wolfed: how ?
[18:38] <Wolfed> ffmpeg -i whatever.vob whatever.whatever?
[18:48] <DelphiWorld> Wolfed: do you know any good commandline ripper?
[18:51] <Wolfed> I guess thats up to you...
[18:51] <Wolfed> Which format do you want it to, resolution, quality, etc...
[18:51] <DelphiWorld> Wolfed: default mkv
[19:05] <DelphiWorld> please someone tel me how to mkv my dvd :)
[19:05] <klaxa> take a look at HandBrake it's designed for exactly that purpose
[19:07] <DelphiWorld> klaxa: yup. it's giving me error and segfault. can you try to see what's going on ?
[19:08] <klaxa> HandBrake is segfaulting? did you download the latest version or even better, built from source?
[19:08] <klaxa> *build
[19:08] <DelphiWorld> klaxa: do they use git? if yes: please, give me the git url
[19:09] <microchip_> DelphiWorld: if you don't mind using MEncoder -> http://h264enc.sourceforge.net/
[19:09] <DelphiWorld> michaelni: handbrake better
[19:10] <klaxa> https://github.com/HandBrake/HandBrakeMirror <-- that's a mirror from the svn repo i think?
[19:10] <DelphiWorld> klaxa: you have the svn repo?
[19:11] <klaxa> do you have internet? i'm actually just using google here
[19:11] <DelphiWorld> klaxa: no, i dont have internet :)
[19:11] <DelphiWorld> LOL
[19:13] <Wolfed> FFS
[19:13] <Wolfed> I finally got it
[19:13] <Wolfed> No idea how
[19:32] <DelphiWorld> LOL diegoviola ?
[19:39] <DelphiWorld> klaxa:  i'm building hb
[19:45] <Wolfed> Resources-wise, if I plan to be constantly converting ffmpeg, do I need a good hardware set?
[19:45] <Wolfed> with ffmpeg*
[19:48] <Mavrik> depends on how long you want to wait
[19:48] <Wolfed> lol
[19:48] <Wolfed> Meaning?
[19:50] <DelphiWorld> Mavrik: LOL
[20:15] <amaurea> Hi! ffmpeg wouldn't happen to support emulator audio files such as SPC and similar, would it? If not, what steps are needed for adding support for it?
[20:49] <Devon> how do i fix video where video and audio are not in sync
[20:51] <derric> Is converting from fixed frame rate to variable frame rate possible with ffmpeg and libx264?
[21:21] <derric> I also noticed that what ever value I set for crf, ffmpeg seems to add five to that number, so if I set -crf 23, then the q-28.0 is the result when encoding.  What is that about?
[21:22] <JEEBsv> the q is IIRC average qp
[21:22] <JEEBsv> or so
[21:23] <JEEBsv> it's not straight correlation, different things
[21:29] <derric> Hey, I went back and read though all you posted the other day.  I missed a lot of stuff.
[21:30] <derric> I understand things much better now.
[21:30] <JEEBsv> congrats then
[21:42] <derric> Any idea how to convert a constant frame rate video to a variable frame rate one?
[21:47] <burek> derric use a variable bitrate parameters :)
[21:47] <burek> framerate*
[22:29] <hurfadurf> hey, anybody in here familiar with libavdevice's jackaudio code?
[22:33] <derric> burek: So set up abr as usual then add a '-framerate *'?  I see in some encoded videos there is a max and min frame rate set, usually min=12 and max=24.
[22:49] <dbarrett> Hi, I am looking to 'chroma key' out the background of an FLV and make it transparent - does anyone know if this can be achived with ffmpeg?
[22:55] <derric> dbarrett:  Cinellerra will do it.
[22:56] <derric> s/cinellerra/cinelerra/
[22:59] <dbarrett> Thanks for that, sorry I should have said I need to be able to do this programatically as I have 1TB of videos to get through :(
[23:01] <llogan> dbarrett: maybe one of the frei0r filters would be useful
[23:05] <dbarrett> Thank you! I think they will be, I will have a go and let you know.
[23:35] <llogan> two XLR cables from monoprice: one has the ground extended from pin 1 to the "shell" and the other doesn't.
[23:37] <llogan> the one that goes from pin 1 to the shell or chassis or whatever it's called doesn't work.
[23:40] <llogan> "enclosure". yeah, that's the word.
[23:46] <derric> dbarrett: I found chromakey stuff with my homemade greenscreen and improvised lighting too significant tweaking to get to work acceptably well.  If you have video with professional lighting and background then 'maybe' an unattened program could do it.
[23:46] <derric> s/too/took/
[23:48] <knoopx> hi
[23:51] <knoopx> I've been trying to use this method and it seems no longer works, can anybody confirm it?
[23:51] <knoopx> http://ffmpeg.org/trac/ffmpeg/wiki/Create%20a%20thumbnail%20image%20every%20X%20seconds%20of%20the%20video
[23:52] <knoopx> rate arg seems to not accept fractions anymore
[23:55] <dbarrett> derric: There videos of karaoke, so CGI, I just need to remove the black background so I can overlay on some other videos
[00:00] --- Mon Nov 19 2012


More information about the Ffmpeg-devel-irc mailing list