[Ffmpeg-devel-irc] ffmpeg.log.20140626

[burek]

[00:08] <ogrgkyle> Hello
[00:08] <ogrgkyle> I've set up FFmpeg + CodeBlocks
[00:08] <ogrgkyle> I can successfully compile the sample FFmpeg scripts
[00:08] <ogrgkyle> Now...
[00:09] <ogrgkyle> How can I loop through the frames and display them on screen?
[00:10] <ogrgkyle> Basically, I need to loop through the frames of the input vid and "give" them to OpenCV, which will detect certain things.
[00:11] <ogrgkyle> And then I'd like to display the video frame, with the detected boxes drawn on top
[00:41] <elisa87> is this wrong for converting mp3 to ogg? ffmpeg -i Most_Precious_Possession_Master.mp3 -acodec libmp3lame Most_Precious_Possession_Master.ogg
[00:42] <c_14> yes
[00:42] <c_14> you want -c:a vorbis
[00:42] <elisa87> c_14: were you talking to me?
[00:42] <c_14> yep
[00:46] <Hello71> that would work great if you wanted to convert from mp3 to mp3
[00:47] <c_14> Nah, pretty sure ogg won't eat mp3.
[00:47] <c_14> That'll fail.
[00:47] <c_14> Rather noisily too, iirc.
[00:47] <Hello71> maybe, i dunno if you can put mp3 in ogg
[00:47] <Hello71> nope
[00:47] <Hello71> [ogg @ 0x1f636f0] Unsupported codec id in stream 0
[00:48] <Hello71> Could not write header for output file #0 (incorrect codec parameters ?): Operation not permitted
[00:48] <Hello71> that's aac to mp3 in ogg but I'd assume mp3 -> mp3 is same
[00:48] <elisa87> what about this? Unknown encoder 'libmp3lame'
[00:48] <c_14> Your ffmpeg isn't compiled with the libmp3lame library.
[00:49] <Hello71> hm, mp3 is its own container format
[00:49] Action: Hello71 chuckles
[00:49] <c_14> elisa87: if you want ogg, you'll need to use libvorbis as the audio codec.
[00:49] <Hello71> so apparently you can put mp3 in wav
[00:49] <c_14> or vorbis, depending
[00:50] <c_14> Wasn't wav supposed to be for lossless formats?
[00:50] <Hello71> no, that's just the most common codec
[00:50] <Hello71> wav is a generic container
[00:50] <Hello71> it's just that usually people put pcm in there
[00:50] <elisa87> c_14: http://paste.ubuntu.com/7703060/
[00:51] <elisa87> any idea?
[00:51] <c_14> First of all, you're using libav not ffmpeg, either see #libav or download/compile FFmpeg.
[00:51] <c_14> ie
[00:51] <c_14> should suit you nicely
[00:51] <c_14> And it might be vorbis instead of libvorbis.
[00:51] <Hello71> vorbis is the ffmpeg one
[00:51] <Hello71> which usually outputs crap
[00:52] <c_14> The gusari static build is built with libvorbis.
[00:52] <c_14> Might use that.
[00:52] <c_14> The johnvansickle one appears to be built with it as well, so I guess it doesn't matter.
[00:54] <elisa87> this gave me error too aaron at aaron-desktop:~/Music$ ffmpeg -i Most_Precious_Possession_Master.mp3 vorbis  Most_Precious_Possession_Master.ogg
[00:54] <elisa87> can you please write the correct command?
[00:55] <c_14> ffmpeg -i file.mp3 -c:a libvorbis out.ogg
[00:55] <c_14> if you got one of the aforementioned static bulids
[00:55] <c_14> *builds
[00:57] <elisa87> c_14: http://paste.ubuntu.com/7703085/
[00:58] <c_14> You're still using libav, not FFmpeg.
[00:58] <c_14> Download one of the static builds and just execute that.
[01:02] <David-A> elisa87: if the option -c:a is not recognized, you can try -acodec, but you should also specify a quality level (or bitrate) if the content is precious
[01:12] <elisa87> how can I reduce the size of an mp3 file to its half or its 3/4?
[01:12] <c_14> chop the bitrate in half or 3/4, you will, however, lose half to 3/4 of the quality even if you may not hear it.
[01:19] <David-A> elisa87: (the words "precious" and "master" in your filename suggests that you should be careful about the quality of the content)
[01:24] <David-A> elisa87: if a 50% or 75% bitrate reduction is sensible or not depends on what bitrate you start from. cuting in half from 300 to 150 kbit/s you may not hear any difference (depending on your ears and the kind of music), but going from 128 to 64 kbit can be ruining
[01:26] <David-A> elisa87: are you transcoding from mp3 to ogg because you think ogg will be more compact?
[01:48] <elisa87> naaah because of browser compatibility in my website for HTML5 audio
[01:49] <elisa87> naaah because of browser compatibility in my website for HTML5 audio David-A
[01:59] <David-A> elisa87: okay, as long as you keep the originals, nothing can go wrong :)
[02:08] <hai> hello. I'm trying to capture my desktop, precisely a full screen video - think "YouTube video on full screen situation". everything's great - got video and audio of the 'full screen video', but the output video is somewhat wobbly/laggy.. it's not right, what I'm trying to say. I think it might have to do with FPS?
[02:08] <hai> here's the command I used: ffmpeg -f dshow -i audio="Stereo Mix (Realtek High Defini":video="UScreenCapture" -r 24000/1001 -q 1 out.avi
[02:09] <hai> and I'm on Windows, if it's important. but I guess not, since the output on Linux would be the same wobbly/laggy video
[02:56] <David-A> hai: maybe it depends on the graphics driver, the graphics chip, the cpu speed, or the chosen encoding. when I grab 1600x1200 @ 25fps on xubuntu I see no lag, but lots of encoding artifacts. (probably because my cpu cannot encode in real time with high enough bitrate)
[02:57] <hai> David-A: but have you also tried recording a playing Flash video where there's movement?
[02:58] <David-A> hai: yes, but the flash player is only 640x360 pixels on the screen, so I moved around a window too.
[03:03] <hai> David-A: does it matter what FPS I'm setting when I'm starting to record? I have to specify the Flash video's FPS?
[03:06] <David-A> hai: I dont know. I suppose if capture fps is higher than display fps or flash fps, you would get duplicated frames, and if capture fps is lower you would miss video frames.
[03:07] <David-A> hai: I also suppose the processes are not synced, so it might be that if the fps's are exactly the same, there may be occational missed/duplicated frames anyways
[03:09] <Hello71> that is impossible
[03:09] <Hello71> assuming the timing is constant
[05:10] <ffmpeg953> Hey everybody
[05:10] <ffmpeg953> Got a question that I think is super simple
[05:10] <ffmpeg953> but before I destroyed my confidence I wanted to double check with you guys
[05:11] <ffmpeg953> drawtext=fontfile=/home/fonts/arial.ttf:text='.This is my text One Example'fontsize=20:fontcolor=white:x=40:y=300:borderw=4:bordercolor=black
[05:11] <ffmpeg953> Can I add another drawtext command to the same ffmpeg command?
[05:31] <relaxed> ffmpeg953: should be able to
[06:14] <ffmpeg953> relaxed what would I use to separate the two?
[06:14] <ffmpeg953> relaxed: just add another command?
[06:25] <relaxed> yes
[06:38] <ffmpeg953> sorry relaxed I keep missing you
[06:38] <ffmpeg953> relaxed do I separate with comma? I am a newb
[06:38] <ffmpeg953> relaxed: do it like this?  drawtext=fontfile=/home/fonts/arial.ttf:text='.This is my text One Example'fontsize=20:fontcolor=white:x=40:y=300:borderw=4:bordercolor=black, drawtext=fontfile=/home/fonts/arial.ttf:text='.This is my text Two Example'fontsize=20:fontcolor=white:x=40:y=300:borderw=4:bordercolor=black
[06:38] <relaxed> https://trac.ffmpeg.org/wiki/FilteringGuide
[06:40] <ffmpeg953> it doesn't show multiple filters does it? I just read through it
[06:43] <relaxed> reread Filtergraph,Chain,Filter relationship
[06:43] <relaxed> but yes, use a comma
[06:45] <ffmpeg953> I'm using 2.2.3
[06:45] <ffmpeg953> and I don't ever use the [in] command
[06:46] <ffmpeg953> http://pastie.org/9325127
[06:46] <ffmpeg953> this is how my command looks right now
[06:47] <ffmpeg953> Can you show me proper format?
[06:48] <ffmpeg953> the "chains" you referenced use [in]
[06:48] <relaxed> -filter:v filter=blah,filter=bleh
[06:52] <ffmpeg953> running command now
[06:52] <ffmpeg953> it hasn't kicked me out yet, lol
[06:57] <ffmpeg953> nope, it failed :(
[13:54] <vtorri> hey
[13:54] <vtorri> in http://www.ffmpeg.org/general.html, i can see :
[13:54] <vtorri> HEVC	X (for encoding
[13:55] <vtorri> and in the remarks :
[13:55] <vtorri> encoding supported through the external library libx265
[13:55] <vtorri> shouldn't it be E instead of X for encoding ?
[13:55] <vtorri> (if the remarks is still valid)
[13:56] <sacarasc> Should what be E instead of X?
[13:57] <vtorri> X :)
[13:57] <vtorri> HEVC E
[13:57] <vtorri> instead of
[13:57] <vtorri> HEVC X
[13:57] <vtorri> look at the link...
[13:58] <sacarasc> I can't see "HEVC X" on that page at all.
[13:59] <vtorri> search just HEVC then...
[13:59] <vtorri> the copy past is wrong
[13:59] <sacarasc> Oh, that X is just showing that it can be done.
[13:59] <sacarasc> It's a check mark.
[13:59] <vtorri> because of the html table in the link, i guess
[13:59] <vtorri> no
[14:00] <vtorri> it means it can be done with libavcodec
[14:00] <vtorri> but the remarks says it's through libx265
[14:00] <sacarasc> So it does.
[14:00] <vtorri> if an external lib is used, then it should be E
[14:00] <vtorri> then the remark is wrong
[14:22] <relaxed> ffmpeg -codecs shows DEV.L. hevc
[15:40] <ffmpeg953> If you don't know the image size of an input, what is the command to make an image 640x480
[16:42] <Sashmo> has anyone been able to get VDPAU to work with FFMPEG and get a noticeable increase in encoding?
[16:42] <klaxa|work> vpdau is not for encoding
[16:42] <Sashmo> oh..... really?
[16:43] <Sashmo> what is it for then?
[16:43] <klaxa|work> decoding
[16:43] <Sashmo> ah....
[16:43] <Sashmo> can ffmpeg support GPU encoding?
[16:43] <Sashmo> getting mixed answers in google
[16:44] <klaxa|work> there is some support for opencl i think
[16:44] <Sashmo> yeah thats what I saw, but is there a noticeable increase?
[16:46] <klaxa|work> depends on your definition of noticeable
[16:46] <klaxa|work> google for benchmarks, here is on with x264 http://forum.doom9.org/showthread.php?t=164960&page=11
[16:47] <klaxa|work> although i'm not sure if this can be applied to ffmpeg
[16:47] <klaxa|work> here is a mailing list thread http://ffmpeg-users.933282.n4.nabble.com/OpenCL-What-Effect-Does-It-Have-td4660469.html
[16:56] <dirty_d> Hey, is it possible to just change the H.264 level with ffmpeg without re-encoding? I have some low-bitrate videos foolishly set to level 5.2. It seems Apple devices just look at the level and give up, because manually hex editing the level to 4.1 makes the video play fine.
[16:56] <dirty_d> not sure if there is a more sane way to fix them
[16:57] <sacarasc> No, you can't.
[17:11] <dirty_d> hmm, if i just set a similar bitrate and the level it re-encodes at like 400fps
[17:11] <dirty_d> how could that be?
[17:11] <dirty_d> encoding with libx264 is usually much much slower for me
[17:12] <Sashmo> klaxa|work: thanks for the links.
[17:21] <brontosaurusrex> dirty_d, whats the resolution of the originals?
[17:22] <dirty_d> 426x240
[17:22] <brontosaurusrex> well, thats like lowres nowadays, so 400 fps looks good to me
[17:22] <brontosaurusrex> and you are not doing any scaling i suppose
[17:22] <brontosaurusrex> *re-scaling
[17:23] <dirty_d> hmm
[17:24] <dirty_d> well what option would I use to keep all the codec parameters exactly the same as the original video except change the level?
[17:49] <xlinkz0> hey, i'm having trouble with a complex filtergraph that should mix two videos side by side
[17:50] <xlinkz0> here's my report : http://sprunge.us/NGNC
[17:50] <xlinkz0> and how i build the filtergraph ( for clarity ) : http://sprunge.us/cAVQ
[18:00] <xlinkz0> smh.. filter_complex not "filter complex"
[18:00] <xlinkz0> my bad
[18:08] <xlinkz0> can I concatenate and cut videos using filters?
[18:51] <xlinkz0> say I have 4 videos, ( a, b, c, d). I want to concat a+b and c+d and play them side by side
[18:51] <xlinkz0> so a+b , c+d should be played side by side
[18:51] <xlinkz0> i've tried this : http://codepad.org/EcVn4nYn
[18:52] <xlinkz0> however the output is "a, a"
[18:52] <xlinkz0> that is, the first video is played side-by-side with itself
[18:54] <xlinkz0> JEEB: please.. help a poor guy out :D
[18:54] <JEEB> > implying I have any idea about the filter_complex stuff
[18:54] <xlinkz0> :(
[18:55] <xlinkz0> well i guess it's not called _compled for nothin
[18:55] <xlinkz0> *complex
[18:55] <JEEB> the syntax is something I never felt like teaching myself since everything I wanted to do was possible by just chaining stuff in -vf
[18:55] <xlinkz0> i .. completely understand :p
[19:00] <sfan5> xlinkz0: You could try concatting a+b & c+d seperately and then doing it how it is written in https://lists.ffmpeg.org/pipermail/ffmpeg-user/2013-June/015662.html
[19:02] <xlinkz0> sfan5: i can merge two files with overlay.. the problem arises when I try to add the concat filter
[19:02] <xlinkz0> sfan5: by separately you mean different ffmpeg calls?
[19:02] <sfan5> yes
[19:03] <xlinkz0> i'd rather not do that since it would require transcoding twice
[19:03] <xlinkz0> and.. i kind of need to do it in bulkk
[19:03] <xlinkz0> i do need accurate seeking
[19:07] <xlinkz0> ubitux: have you got some spare time? .. or could you please send me to someone else I can pester for a few minutes about this complex filtergraph? :D
[19:08] <jleclanche> using ffmpeg 2.2.4, the following command completely hangs and is unkillable even with kill -9. ffmpeg -f v4l2  -i /dev/video0 output.mkv   --- this is an strace output, it hangs right at the end there. http://dpaste.com/2JCN4H0
[19:08] <jleclanche> any idea?
[19:09] <sfan5> kernel bug?
[19:09] <jleclanche> likely, since its unkillable
[19:09] <sfan5> or maybe I/O wait
[19:09] <jleclanche> "mpv tv:///" works fine, however when I execute that ffmpeg command, mpv tv:/// will also hang
[19:09] <sfan5> you can't kill that either
[19:09] <jleclanche> yes, if i reboot it'll say "/home is busy"
[19:10] <jleclanche> mpv tv:/// when it actually works says [tv] v4l2: your device driver does not support VIDIOC_G_STD ioctl, VIDIOC_G_PARM was used instead.
[19:10] <jleclanche> i dont know if thats related
[19:11] <jleclanche> This all used to work a few months ago
[19:11] <jleclanche> maybe new gcc, new linux or new ffmpeg has the issue, who knows
[19:11] <sfan5> I don't think it's gccs fault
[19:12] <jleclanche> this is on arch linux fwiw, dunno if anyone else has the issue
[19:18] <checkit> Hey guys, I haven't had to use ffmpeg for sometime now but a project came up recently that I think might make good use of it. I need to capture a visitors audio submission as part of a song but I'm not sure how to go about it. Any guidance here?
[19:46] <Sergadra> hello
[20:36] <woof-woof> Can anyone explain to me why this happens with ffmpeg? > http://woof-woof.site11.com/files/sample.png
[20:39] <xlinkz0> woof-woof: different encoder options
[20:40] <woof-woof> ?
[20:40] <xlinkz0> what don't you understand from that statement?
[20:40] <woof-woof> What options?
[20:41] <xlinkz0> https://trac.ffmpeg.org/wiki/Encode/H.264
[20:41] <woof-woof> That's not h264
[20:41] <ChocolateArmpits> Has anyone compiled ismindex here before?
[20:42] <woof-woof> I just downscaled the image...
[20:42] <xlinkz0> !testcase
[20:42] <xlinkz0> well.. post your full command
[20:43] <woof-woof> http://paste.ubuntu.com/7707305/
[20:46] <woof-woof> No.!
[20:51] <woof-woof> That happens when videos are downscaled too.
[20:56] <klaxa> woof-woof: so i'm not exactly sure what i'm supposed to see, but i think you are talking about downscaling of the images makes it appear blockier in ffmpeg than in gimp?
[20:58] <woof-woof> klaxa, Those details are smoother in GIMP. I just want to know if ffmpeg can do it similar. Because it would be visually better.
[20:58] <klaxa> if so, it entirely depends on the video filter used. the scale video filter has different options, look at those, same goes for gimp
[20:59] <klaxa> see: https://www.ffmpeg.org/ffmpeg-filters.html#scale-1
[21:00] <klaxa> huh? wait... i thought it had more options...
[21:08] <woof-woof> klaxa, I found it!
[21:09] <woof-woof> The option to use is "full_chroma_int" (full chroma interpolation)
[21:09] <woof-woof> Now it looks the way I wanted!
[21:09] <woof-woof> Thanks!
[21:34] <ffmpeg953> omg I can't find a good watermark command for 2.2.3 at all
[21:34] <ffmpeg953> any one have one?
[21:58] <t4nk590> hi, i'd like to hardsub an mkv file that has multiple ssa subtitle streams. is there a way to pick which stream to use? i'm currently using -vf subtitles=$FILE
[22:00] <t4nk590> the :si=3 option does not seem too work as i get an "Error initializing filter 'subtitles' with args" error
[00:00] --- Fri Jun 27 2014