[Ffmpeg-devel-irc] ffmpeg.log.20111115

burek burek021 at gmail.com
Wed Nov 16 02:05:01 CET 2011


[00:02] <necromancer> okay so my problem of ffmpeg failing to convert FLV->MP3 is still happening, even though i compiled from source and configured it properly
[00:02] <necromancer> here's the log: https://gist.github.com/71a9cf8c67ee74bd7185
[00:04] <iive> necromancer: please get a newer source and compile it.
[00:05] <iive> i mean, get current git ffmpeg and compile it.
[00:06] <iive> also, are you sure the clip is working properly? no partial downloads and broken files?
[00:08] <necromancer> iive: i can definitely hear it through my flash app
[00:08] <necromancer> iive: i actually compiled everything from source, and ffmpeg is from git.videolan.org/ffmpeg
[00:09] <necromancer> iive: actually i may have figured it out...do i have to do --enable-speex to get speex support in FLVs?
[00:09] <necromancer> my FLVs are definitely using Speex
[00:11] <iive> necromancer: let me take a look
[00:12] <iive> it's -enable-libspeex and yes, you have to install that lib first then use the enable.
[00:15] <necromancer> thanks iive
[00:15] <iive> i see that the second pastebin is with newest ffmpeg and it does recognize speex
[00:15] <necromancer> ubuntu's apt-get sucks
[00:15] <necromancer> wait what?
[00:16] <necromancer> well it's recognizing that it is using Speex but not able to process it
[00:16] <necromancer> because --enable-libspeex isn't in config
[00:16] <iive> yes, I replied after looking at your first pastebin text... 
[00:17] <iive> i'm saying that because recognizing it is a good thing, it makes it much more likely that the decoding would work.
[00:19] <iive> about apt-get, you probably need the package and the -dev version.
[00:19] <iive> it must be listed in the optional packages too.
[00:23] <shysoft> With a prior version the libav* tools (6.3.something) , a packet retrieved with av_read_frame was correctly associated with the stream with which it came, but when I upgraded to the most recent version, this is no longer the case. Does anyone know why?
[00:23] <necromancer> iive: interesting. thanks
[00:26] <necromancer> thanks iive it at least ./configure'd properly. now on to make && make install
[00:38] <necromancer> iive: okay so it configured but when i tried to make install, i got the following error(s): https://gist.github.com/1365595
[00:38] <necromancer> seems this has to do with speex
[00:52] <Breakable> Does it need more RAM or more CPU to process many video converts ?
[00:52] <Breakable> at the same time
[00:55] <Breakable> Help!
[00:59] <iive> Breakable: cpu.
[01:01] <Breakable> iive: does high RAM increases the speed as well or not much important ?
[01:02] <iive> no
[01:02] <Breakable> K great, web servers are cheaper with high CPU
[01:02] <iive> create swap partition and enable it
[01:03] <iive> if it starts using the swap then you need more ram.
[01:04] <Breakable> iive:  is 32x1700Gz &  15GB RAM good for like 20-30 concurrent converts?
[01:04] <Breakable> 32x1700Gz  = 32 CPU of 1700Gz
[01:04] <iive> depends on the encoder and the used settings.
[01:04] <Breakable> Ah I c
[01:05] <iive> i would recommend you to do some tests on your own.
[01:05] <Breakable> yes I think this is the best way to know .. 
[01:05] <iive> but having one cpu per convert seems quite resonable.
[01:05] <Breakable> K that's great.. 
[01:06] <Breakable> iive: thank you for your help! 
[01:06] <iive> yes
[01:06] <iive> i just want to be quite clear. having more ram than needed won't make it any faster, but having less would make if painfully slow.
[01:07] <iive> the extra ram is just used as harddisk cache.
[01:07] <Breakable> Ah , I got it .. 
[01:08] <Breakable> I'll go for the higher CPU , with moderate amount or RAMs
[02:02] <JuanCri> hey? anybody knows how to use the "dvb" subtitles codec in ffmpeg?
[02:02] <JuanCri> I see it's on the list of supported codecs? but I'm not sure how to use it? any help would be appreciated :)
[02:17] <pasteeater> JuanCri: what are you trying to do?
[02:18] <JuanCri> my goal is to inject dvb pixmap subtitles into a live mpeg-ts? for that I'm evaluating libavcodec? so I wanted to try first with ffmpeg
[02:18] <JuanCri> as it seems to support the codec
[02:20] <JuanCri> but I didn't find any example :(
[03:06] <Stemby> hi, i'm testing ffmpeg
[03:07] <Stemby> i have some images (00000.jpg ... 00019.jpg)
[03:07] <Stemby> i tried:
[03:08] <Stemby> $ ffmpeg -r 6 -i %05d.jpg test.ogg 
[03:08] <Stemby> but i get a video with only the first frame
[03:08] <Stemby> where is the mistake?
[03:14] <teratorn> anyone know how to copy an AVFrame including the ->data buffers
[03:28] <Stemby> nobody can help me? i tried also:
[03:29] <Stemby> $ ffmpeg -f image2 -i %05d.jpg test.mpg 
[03:29] <Stemby> with the same result
[03:29] <alyawn> Stemby, do you see any errors in the output of ffmpeg?
[03:29] <Stemby> no
[03:30] <Stemby> Press ctrl-c to stop encoding frame=   20 fps=  0 q=2.0 Lsize=      62kB time=0.76 bitrate= 668.3kbits/s     video:61kB audio:0kB global headers:0kB muxing overhead 1.015115% 
[03:30] <Stemby> it seems ok, right?
[03:31] <alyawn> why press ctrl-c?
[03:32] <Stemby> to stop the encoding, i think
[03:32] <alyawn> oh.. so you have a lot of images, then
[03:32] <alyawn> more than 20?
[03:32] <Stemby> no
[03:32] <Stemby> exactly 20
[03:33] <Stemby> in fact i didn't stop ffmpeg
[03:33] <alyawn> oh ok, so it exited on its own then
[03:33] <Stemby> yes
[03:33] <Stemby> this is only the output
[03:33] <Stemby> (the last 3 lines)
[03:34] <alyawn> can you pastebin the full output of the ffmpeg output?
[03:35] <Stemby> ok, 2 minutes
[03:35] <alyawn> my guess is that you need to set the rat to 6 on the output as well
[03:35] <alyawn> rate*
[03:38] <Stemby> alyawn: http://paste.debian.net/145475/
[03:39] <Stemby> alyawn: http://paste.debian.net/145476/
[03:41] <Stemby> alyawn: http://paste.debian.net/145478/
[03:41] <alyawn> the first attmept to create test.mpg seems like it's giving a warning: VBV buffer size not set, muxing may fail
[03:46] <alyawn> I don't see anything obvious in the other two.. the durations seem to be correct, etc
[03:46] <Stemby> i opened the files with vlc
[03:47] <alyawn> does `ffmpeg -i test.ogg` report correct properties, duration, etc 
[03:48] <Stemby> alyawn: http://paste.debian.net/145479/
[03:48] <Stemby> it seems ok
[03:50] <alyawn> well, I can't see anything obviously wrong based on outputs... the only comment I have is that the sample is incredibly small and could be causing issues
[03:51] <Stemby> mmm, can i send you the images?
[03:55] <Stemby> ah, 00000.jpg is corrupted!
[03:55] <Stemby> 2 minutes
[03:59] <alyawn> is it possible to use a rawvideo input for the movie src filter? I'm trying to use a raw yuv as input and i dont know where to set the size, framerate, etc
[04:00] <relaxed> alyawn: you may need to stick it in an avi or mkv container
[04:03] <alyawn> relaxed, yeah, that's what I ended up doing, looks like I'll need to write a simple program to handle this, since my input formats are static. thanks for the info
[04:04] <Stemby> sorry, i had to reboot
[04:04] <Stemby> alyawn: i found the problem:
[04:05] <Stemby> my photos were identical
[04:05] <alyawn> Stemby, lol... that happens to me all the time l)
[04:05] <Stemby> lol
[04:06] <Stemby> using the command line, i didn't note the problem
[04:06] <alyawn> yep
[04:30] <Stemby> o/
[07:47] <monstaRtruck> hi
[07:47] <monstaRtruck> i recorded my desktop w ffmpeg 
[07:47] <monstaRtruck> <monstaRtruck> but its too loud
[07:47] <monstaRtruck> <monstaRtruck> i cant turn it down
[07:47] <monstaRtruck> <monstaRtruck> its distorted
[07:48] <monstaRtruck> how
[09:47] <giany> hello
[09:47] <giany> i'm running : 
[09:47] <giany> /usr/bin/ffmpeg   -v 9 -loglevel 99 -y -i 'sites/default/files/videos/original/kb1.mp4' -acodec libfaac -ab 256k -pass 2 -vcodec libx264 -b 1000k -threads 0  'sites/default/files/videos/converted/kb1.mp4'
[09:47] <giany> and i get this msg
[09:47] <giany> [libx264 @ 0x8553660] requested bitrate is too low. estimated minimum is 9887 kbps
[09:47] <giany> Error while opening encoder for output stream #0.0 - maybe incorrect parameters such as bit_rate, rate, width or height
[09:48] <giany> any idea what is wrong?
[09:49] <teratorn> is there a generic way to detect if a given AVFrame contains an alpha channel, and a generic way to extract those alpha channel values in to their own plane?
[09:51] <Tjoppen> a) check codec->pix_fmt   b) data[3] is planar, else you have to repack it
[09:51] <Tjoppen> *if planar
[10:01] <teratorn> Tjoppen: so I just have to have a big list of pix_fmt's that I support, and custom code for each that extracts the alpha values (in the case where data[3] is not already provided) ?
[10:02] <Tjoppen> see pix_fmt.h - there's a struct there that can help you
[10:02] <Tjoppen> erm, pixdesc.h
[10:03] <Tjoppen> AVPixFmtDescriptor should allow you to detect alpha in a bit more generic way
[10:03] <Tjoppen> but an explicit list is probably a simple enough solution
[10:06] <teratorn> hmmmm
[10:07] <teratorn> Tjoppen: not really sure I can see how :/
[10:26] <teratorn> Tjoppen: do you know if PIX_FMT_BGRA is always packed using machine-specific endianness?
[10:30] <Tjoppen> uhm, it's not
[10:30] <Tjoppen> PIX_FMT_BGRA,      ///< packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
[10:30] <Tjoppen> #define PIX_FMT_RGB32   PIX_FMT_NE(ARGB, BGRA)
[10:31] <Tjoppen> PIX_FMT_RGB32 is the endian-specific one
[10:45] <teratorn> I see
[10:46] <teratorn> Tjoppen: well, I solved my problem by discovering the YUVA420P format, so instead of hacking my own alpha channel in the unused data[3] buffer of a YUV420P frame, I just use that :)
[10:46] <jluttine> i'm wondering why ffmpeg doesn't find some shared libraries.. i'm compiling+installing ffmpeg locally to ~/.local/ and the installation works. but when i try to run ffmpeg it doesn't find some of the shared libraries (those that are in ~/.local/lib). among other flags, i give these to configure: --prefix=$HOME/.local --extra-cflags=-I$HOME/.local/include --extra-ldflags=-L$HOME/.local/lib . if i run ffmpeg as ...
[10:46] <jluttine> ... "LD_LIBRARY_PATH=$HOME/.local/lib ldd ffmpeg" but how could i hard-code this path to the ffmpeg executable..??
[10:48] <jluttine> hmm, i meant to say that ffmpeg works if i set LD_LIBRARY_PATH but not otherwise
[10:48] <teratorn> that's fairly common for unix software
[10:49] <teratorn> jluttine: just set LD_LIBRARY_PATH in your shell's environment startup file
[10:51] <teratorn> I'm not sure there is any way around it for dynamically-linked binaries
[10:51] <jluttine> teratorn: the "problem" is that i want to use the local-libraries only for my local ffmpeg and system-libraries for all other software
[10:51] <jluttine> i thought that -L flag would hardcode the path to the executable, not only use it when compiling
[10:52] <teratorn> well, ld.so runs before any code in the binary, so you're pretty much screwed
[10:53] <teratorn> if you just care about the 'ffmpeg' command-line tool, you can write a trivial shell-script wrapper
[10:53] <teratorn> LD_LIBRARY_PATH=$HOME/.local/lib ffmpeg "$@"
[10:54] <jluttine> teratorn: yep. i'm not sure whether ffmpeg then works with, for instance, blender..
[10:54] <jluttine> well, that's maybe a different issue then
[10:55] <jluttine> anyway, to me it's confusing that different library paths are used when compiling and executing
[10:56] <jluttine> but maybe that shell-script is the only solution :/
[10:58] <jluttine> teratorn: yes, thanks for your help, i'll do this shell-script
[10:59] <jluttine> or maybe i should link those libraries statically to ffmpeg
[10:59] <teratorn> jluttine: yeah, the difference between compile-time and run-time sux
[10:59] <teratorn> jluttine: that would also work
[12:49] <HektoR> Hello guys. i just discovered that ffmpeg can stream pre-recorded files to RTMP server. i'm interested is this feature stable ? can ffmpeg change Adobe Flash Media Encoder ?
[12:56] <zamba> i want to encode two flv files into one, how do i do that?
[12:56] <zamba> i tried -i <file1> -i <file2>, but that didn't work
[12:57] <sacarasc> zamba: Look up concat in the manual.
[13:01] <zamba> ffmpeg -i concat:229.flv\|229-02.flv - -vcodec copy -acodec copy -f flv output.flv
[13:01] <zamba> i tried that
[13:01] <zamba> didn't work
[13:02] <zamba> i also tried cat 229.flv 229-02.flv | ffmpeg -i - -vcodec copy -acodec copy -f flv output.flv
[13:02] <zamba> but no. i only get the first file
[13:33] <ensoe> good morning. Does anyone here use ffmpeg to stream to justin.tv or twitch.tv? I don't seem to be able to pick any of the lossless -vpre presets to stream with. Was wondering if anyone was able to do HQ or pretty much raw live streaming (I followed this guide: http://fuckingsound.com/2011/06/27/justin-tv-streaming-in-linux/ )
[13:37] <zamba> sacarasc: got any other suggestions?
[13:44] <cofol1986> Hi
[13:45] <zamba> seriously.. is it impossible to join two flv files together using ffmpeg?
[13:45] <zamba> this is stupid
[13:45] <cofol1986> hello everyone, I compile ffmpeg in Ubuntu 11.10.after compiling,I start up ffmpeg,and get an error "ffmpeg: relocation error: /usr/local/lib/libswresample.so.0: symbol av_get_default_channel_layout, version LIBAVUTIL_51 not defined in file libavutil.so.51 with link time reference"
[13:45] <zamba> http://www.ffmpeg.org/faq.html#How-can-I-join-video-files_003f
[13:45] <cofol1986> can anyone help me ?
[13:45] <zamba> i see that, but that's like 10 lines to do something really simple
[13:48] <cofol1986> has anyone met this problem---"ffmpeg: relocation error: /usr/local/lib/libswresample.so.0: symbol av_get_default_channel_layout, version LIBAVUTIL_51 not defined in file libavutil.so.51 with link time reference"?
[15:16] <Milos|Netbook> Yay, a channel!
[15:16] <Hyperi> "Ok"
[15:16] <Hyperi> .D
[15:17] <Milos|Netbook> Alright, ffmpeg is freaking awesome right, but how can I use
[15:17] <Milos|Netbook> ffmpeg -v 0 -f video4linux2 -i /dev/video0 -vcodec png -r 6/60 -y image.png
[15:17] <Milos|Netbook> Notice how I have -y which means overwrite OK
[15:17] <Milos|Netbook> and notice how I don't have a %d
[15:17] <Hyperi> :D
[15:17] <Hyperi> webcam?
[15:17] <Milos|Netbook> I want to overwrite the same file, every 10 seconds.
[15:17] <Milos|Netbook> Yes webcam :D
[15:17] <Milos|Netbook> Please tell me this is possible.
[15:17] <Milos|Netbook> I'd hate to fork ffmpeg every 10 seconds.
[15:17] <Hyperi> Why not?
[15:17] <Hyperi> I bet it's more efficient that way
[15:17] <Milos|Netbook> You think?
[15:17] <Milos|Netbook> It seems like a lot of overhead.
[15:18] <Hyperi> Well running something every 10seconds vs. Running something all the time
[15:18] <Hyperi> :P
[15:18] <ubitux> why do you need to override the image?
[15:18] <Milos|Netbook> In terms of CPU usage
[15:18] <Milos|Netbook> Press [q] to stop, [?] for help
[15:18] <Milos|Netbook> [image2 @ 0x159bf90] Could not get frame filename number 2 from pattern 'image.png'
[15:18] <Milos|Netbook> av_interleaved_write_frame(): Invalid data found when processing input
[15:18] <ubitux> if you want to stream a "regular screenshot", mjpeg is appropriate
[15:18] <Milos|Netbook> If I don't specify %d I get the above
[15:18] <Milos|Netbook> even with -y
[15:19] <Milos|Netbook> I want it to take an image snapshot.
[15:19] <ubitux> i'm curious about the use case behind :)
[15:19] <Milos|Netbook> I have my own streaming script
[15:19] <Milos|Netbook> it loads an image filename
[15:19] <Milos|Netbook> so I would prefer a constant image filename
[15:20] <Milos|Netbook> is it at all possible?
[15:20] <ubitux> i'm not sure you can do that
[15:20] Action: Milos|Netbook edits ffmpeg source and recompiles, then
[15:20] <ubitux> are you sure you can't workaround this with mjpeg?
[15:21] <ubitux> mjpeg is simply a stream in which you output jpeg files one by one
[15:21] <Milos|Netbook> well forgive my ignorance, I haven't looked into mjpeg much yet
[15:21] <ubitux> also, browser are able to render it directly with <img src="...
[15:21] <ubitux> +s
[15:21] <Hyperi> ubitux 1 - Milos|Netbook 0
[15:21] <Hyperi> :D
[15:21] <ubitux> it's just an idea, i don't know if it's really appropriate to his use case
[15:21] <Milos|Netbook> can I specify a frame rate then...?
[15:22] <ubitux> yes afaik
[15:22] <Milos|Netbook> because I don't want people to be able to view any moment in time
[15:22] <Milos|Netbook> if you get what I mean
[15:22] <Milos|Netbook> it's meant to be every 10 secs
[15:22] <Milos|Netbook> so 9 secs need to be cached
[15:22] <Hyperi> cached or devnulled?
[15:22] <Milos|Netbook> devnull sounds like a waste of resources
[15:22] <ubitux> it should be possible, but if not, you can still use -vf select to drop the frame range you don't want
[15:22] <Milos|Netbook> I see
[15:23] Action: Milos|Netbook grabs a bag of chips and comes to read more on mjpeg
[15:23] <ubitux> your webcam might even support mjpeg output natively btw
[15:24] <ubitux> most of the recent ones do at least
[15:24] <Hyperi> Milos|Netbook: devnulled means it isn't used at all, hecne the nickname from /dev/null
[15:24] <Hyperi> Figured everyone knew that nickname of behaviour .<
[15:26] <Milos|Netbook_> wtf
[15:26] <Milos|Netbook_> disconnection
[15:26] <Milos|Netbook_> anyway
[15:26] <Milos|Netbook_> but /dev/null implies something is being sent there
[15:26] <Milos|Netbook_> which means something is being received,
[15:26] <Milos|Netbook_> which is not used
[15:26] <Milos|Netbook_> so it's a waste of resources getting something that isn't used
[15:26] <Milos|Netbook_> so it shouldn't be got in the first place
[15:26] <Hyperi> 16:24:09 <    Hyperi> Milos|Netbook: devnulled means it isn't used at all, hecne the nickname from /dev/null
[15:26] <Hyperi> 16:24:28 <    Hyperi> Figured everyone knew that nickname of behaviour .<
[15:26] <Milos|Netbook_> o.O
[15:27] <Hyperi> O.o
[15:29] <Milos|Netbook_> mjpeg still wants %d?
[15:36] <ubitux> no, it's a single stream/file
[15:36] <Milos|Netbook_> [image2 @ 0x16d3f90] Could not get frame filename number 2 from pattern 'image.jpg'
[15:38] <fris> anyone ever have any issues with ffmpeg just doing nothing, like ffmpeg from console, just hangs, doesnt print out anything like it suppose too, i had it working before
[16:26] <Milos|Netbook_> Can I have ffmpeg use stdout as the output file?
[16:27] <sacarasc> Yes.
[16:27] <Milos|Netbook_> Great, what's the syntax?
[16:28] <sacarasc> -f blah - or -f blah /dev/stdout
[16:28] <Milos|Netbook_> Thanks.
[16:28] <sacarasc> blah depends on what you want the output format to be.
[16:28] <Milos|Netbook_> Yup.
[16:30] <Milos|Netbook_> Perfect thank you.
[17:20] <Milos|Netbook_> What happens when img->img_number exceeds 4 bytes?
[18:53] <Zeonixa> Can someone tell me how to ensure that the width and height are divisible by 2 when grabbing a window?
[18:54] <Zeonixa> This is the script I found, http://pastebin.com/1UH0vZBA, but after I changed it to x264, it very much didn't like the odd width values, how would I scale up on that?
[19:00] <matthiaskrgr> hi
[19:00] <matthiaskrgr> can anyone reproduce this bug ? https://bugs.archlinux.org/task/26905 
[19:01] <matthiaskrgr> or is  " For x264 encoding use -preset and/or -tune "  from the topic referring to that bug? :)
[19:18] <monstaRtruck> hi
[19:19] <monstaRtruck> my sound and video dont sync in desktop recording
[19:20] <fris> what would cuase this ffmpeg-0.7.7,1 depends on shared library: x264.116 - found then it gives me error ERROR: libx264 not found
[19:28] <monstaRtruck> fris
[19:28] <monstaRtruck> thats easy 
[19:28] <monstaRtruck> you need to download a lib
[19:28] <monstaRtruck> google libx264 
[19:28] <monstaRtruck> and ull find the ubuntu package that contains it
[19:29] <monstaRtruck> i dl it last nite myself
[19:29] <monstaRtruck> but i forgot what the package was
[19:35] <relaxed> matthiaskrgr: try with the latest from git
[19:36] <fris> monstaRtruck i have x264 installed
[19:37] <monstaRtruck> not sure then
[19:37] <fris> in /usr/local/lib/libx264.so
[19:37] <relaxed> fris: you're trying to compile ffmpeg?
[19:37] <fris> relaxed yes
[19:38] <relaxed> any reason you're going with an older release?
[19:39] <fris> im on freebsd, ffmpeg-devel is broken it says (port)
[19:39] <fris> ffmpeg-2011.10.09 is marked as broken: does not build.
[19:42] <relaxed> which freebsd version?
[19:43] <relaxed> It's not marked broken on http://www.freshports.org/multimedia/ffmpeg-devel/ Are your ports up to date?
[19:44] <fris> relaxed i updated them yesterday
[19:45] <fris> i will redo the ports tree
[19:48] <monstaRtruck> my sound and video dont sync in desktop recording
[19:49] <relaxed> monstaRtruck: pastebin your command
[19:49] <monstaRtruck> i dont have it w me
[19:50] <relaxed> then how is anyone supposed to help you?
[19:51] <pasteeater> also, is it out of sync with ffplay as well?
[20:01] <monstaRtruck> ill get the comand this evening
[20:01] <monstaRtruck> thx
[20:47] <Zeonixa> How do I make sure that ffmpeg makes the video input even if it's odd?
[20:48] <sacarasc> Scale or crop it to something even.
[20:55] <pasteeater> Zeonixa: scale="trunc(oh*a*2)/2:720"
[20:55] <pasteeater> or scale="trunc(oh*a*2)/2:720"
[20:55] <pasteeater> if i understand your question correctly
[20:56] <pasteeater> last paste should have been scale="1280:trunc(ow/a/2)*2"
[21:07] <Zeonixa> Is it possible to record from the output to the speakers and the input from the mic at the same time?
[21:08] <Zeonixa> Also, thanks, although it didn't like scale and I just ended up doing it manually.
[21:22] <Gateway69> anyone know why on the latest ffmpeg im trying to encode x264 with a main profile but its barfing on this..x264 [error]: main profile doesn't support 4:4:4
[21:22] <Gateway69> [libx264 @ 0000000001D1E360] Error setting profile main
[21:24] <relaxed> pastebin your command and all output
[21:25] <necromancer> i'm installing ffmpeg on debian
[21:25] <necromancer> using APT
[21:26] <necromancer> do i have to specify --enable-speex? and if so, where?
[21:26] <pasteeater> Gateway69: your input is probably rgb. you probably want to add -pix_fmt yuv420p
[21:26] <sacarasc> necromancer: No, that's only if you want to compile yourself.
[21:26] <necromancer> oh great
[21:26] <sacarasc> If you don't have speex enabled it in, you;d have to find an alternate build or compile yourself.
[21:26] <necromancer> ahhh ok
[21:26] <necromancer> thanks
[21:26] <pasteeater> or maybe pipe to a speex encoder
[21:27] <Gateway69> pasteeater: thanks! that worked
[21:27] <necromancer> so if i need to compile ffmpeg with speex (which i believe i need to do), i want to compile my own ffmpeg from source then
[21:28] <pasteeater> necromancer: you can adapt this: http://ubuntuforums.org/showthread.php?t=786095
[21:29] <pasteeater> install libspeex-dev and then add --enable-libspeex to your ffmpeg configure
[21:29] <necromancer> libspeex-dev isn't available on debian
[21:29] <necromancer> just ubuntu's APT
[21:30] <necromancer> pasteeater: i already got libspeex compiled though that's not a problem
[21:30] <necromancer> i'm just trying to avoid compiling from source as much as possible
[21:30] <necromancer> in this case
[21:30] <necromancer> but i am in fact following that guide, pasteeater, and adapting it
[21:33] <pasteeater> ok. then you'll just need to use --enable-libspeex in the ffmpeg configure if you already compiled and installed speex properly
[21:33] <pasteeater> i thought debian would have libspeex-dev, but i guess not
[21:35] <relaxed> it does
[21:36] <relaxed> you mean ffmpeg with libspeex support?
[21:42] <pasteeater> necromancer mentioned that the libspeex-dev packages is not available in debian repo while i was assuming it was available, but i didn't look.
[21:43] <Gateway69> anyone help me with an error im setting,, parser not found for codec rawvideo, packets or times may be invalid? my line is
[21:44] <Gateway69> ./ffmpeg -i ../test.avi -vcodec libx264 -vprofile main -pix_fmt yuv420p test.h264
[21:44] <relaxed> necromancer: Are you running debian sid?
[21:45] <necromancer> might have libspeex
[21:45] <necromancer> but not -dev
[21:45] <necromancer> and i shouldn't need "--enable-libxvid" to transcode FLV into MP3, right?
[21:45] <necromancer> relaxed: no idea. how do i find that out (it's on Dreamhost VPS)
[21:46] <necromancer> sorry relaxed freenode is giving me a lot of alg
[21:47] <alyawn> Gateway69... pastebin the entire ffmpeg output...
[21:50] <pasteeater> necromancer: no, you don't need "--enable-libxvid" to transcode FLV into MP3, but you'll need --enable-libmp3lame
[21:50] <necromancer>  yeah i was thinking my ./configure could be trimmed down a bit
[21:50] <pasteeater> unless the flv already contains mp3 and you use -acodec copy in your comand
[21:50] <necromancer> from that Ubuntu example
[21:51] <necromancer> adding --enable-libmp3lame --enable-libspeex of coruse
[21:51] <necromancer> course8
[21:51] <necromancer> course*******
[21:52] <pasteeater> libxvid should have been removed from that guide, unless you're looking at some old thing
[21:52] <pasteeater> ah, it's still there for hardy
[21:54] <pasteeater> gone now
[22:01] <skepster> Hi, does there exist a way to change the packetization mode of the x264 encoder to 0?
[22:03] <Gateway69> alyawn, sorry i had to grab some lunch here is pastbin http://pastebin.com/x6YjvXVh
[22:07] <monstaRtruck> is it possible to record desktop video and sound with out it lagging
[22:08] <monstaRtruck> and in sync
[22:15] <alyawn> Gateway69, that warning is common with uncompressed AVIs I get it all the time and has caused any problems
[22:15] <alyawn> hasn't*
[22:16] <Gateway69> ok thanks
[23:28] <Gateway69> anyone know off the top of their heads in the new ffmpeg how to disable deblock filter.. my old commands do not seem to work
[23:33] <pasteeater> Gateway69: why do you want to do that? what's your old command?
[23:34] <Gateway69> -deblock 
[23:35] <Gateway69> we are doing some testing here and our engeinner asked for the deblock off
[23:35] <pasteeater> your complete command
[23:40] <Gateway69> err cant find it, i have so many docs with profiles in them.. one of the main reasons is with deblockign disabled it decreases cpu load with h264 decoding
[23:44] <pasteeater> if decreasing load on decoding is important then also disable CABAC
[23:44] <Gateway69> yea we are trying one thing at a time to get this to run on our tiny device
[23:44] <Gateway69> i think im looking for skiploopfilter
[23:45] <pasteeater> -coder 0
[23:46] <pasteeater> for cabac disable
[23:46] <Gateway69> yea. got that in our line
[23:47] <JEEB> there's a general tune in libx264 to make it "as quickly decodable as possible"
[23:47] <JEEB> -tune fastdecode
[23:48] <JEEB> that's good for usage with stuff like the original xbox :3
[23:49] <Gateway69> this is what i have now.. http://pastie.org/private/sa28jmxn7jh2pp3fgi5ua
[23:49] <Gateway69> yea this cpu is a bit slower than that xbox :)
[23:49] <JEEB> well, I was just generally saying :)
[23:50] <Gateway69> no deff thanks.. any thing we can try to help, im just having hard time finding the deblocking
[23:50] <pasteeater> I'd go with JEEB's suggestion.
[23:50] <JEEB> - fastdecode: --no-cabac --no-deblock --no-weightb --weightp 0
[23:50] <JEEB> from x264's --fullhelp
[23:50] <JEEB> so you don't have to set all that stuff by yourself if you use the tune
[23:51] <Gateway69> yea no-deblock didnt work on cmd line for ffmpeg :(
[23:51] Action: JEEB points at the tune
[23:51] <JEEB> -tune fastdecode
[23:51] <Gateway69> :)
[23:51] <JEEB> should be usable from ffmpeg/avconv now as well
[23:51] <Gateway69> looking at the docs
[23:51] <JEEB> -tune should now just feed the tuning value to libx264
[23:51] <JEEB> thus
[23:51] <JEEB> -tune fastdecode = --tune fastdecode if you were using x264's command line encoder
[23:52] <JEEB> as far as I can see
[23:52] <JEEB> and unless you were overriding it yourself
[23:52] <Gateway69> cabac disables our b frames which we need to add
[23:52] <JEEB> what
[23:52] <JEEB> CAVLC/CABAC have nothing to do with bframes
[23:53] <JEEB> do you see --tune fastdecode touching the amount of bframes?
[23:53] <Gateway69> one sec.. reading some specs :)
[23:53] <JEEB> this is where I'm starting to take out my facepalm.jpg
[23:54] <Gateway69> thank you sir may i have another :)
[23:54] <JEEB> you don't even have to read the specs to know what libx264 does
[23:54] <JEEB> http://x264.fushizen.eu/files/x264_settings.txt <- --fullhelp of a build of mine
[23:54] <JEEB> scroll down to the tunes
[23:55] <JEEB> and see what fastdecode does
[23:56] <pasteeater> fastdecode will do what you want. otherwise i don't think it's possible to set '-flags -loop' or 'x264opts no-deblock' with ffmpeg right now
[23:56] <Gateway69> trying now :)
[23:56] <JEEB> it only switches to CAVLC instead of CABAC (because it's computationally faster), disables deblock (-||-), disables weighted b-frame prediction (-||-), disables weighted p-frame prediction (-||-)
[23:56] <Gateway69> now to see what it looks like on our tiny device :( thanks JEEB
[23:57] <JEEB> also, I would generally not set as many options
[00:00] --- Wed Nov 16 2011


More information about the Ffmpeg-devel-irc mailing list