[Ffmpeg-devel-irc] ffmpeg.log.20141209

[burek]

[00:10] <llogan> x264 requires width and height to be divisible by 2 when outputting to yuv420p, so that is one case where -2 is useful
[04:24] <pinkette> what is best way to convert VOB to mp4  but keeping the same mpeg2 and ac3
[04:48] <tyr> ChanServ: red5 ffserver whick better
[06:29] <visiot> can anyone help me on pcr repetition error
[06:29] <visiot> my network analyzer is continously saying pcr repition error
[07:13] <pentanol> visiot hi, could you show the dump?
[07:14] <visiot> right now i dont have the setup. But i can give you more information about the scenario
[07:14] <visiot> that i'm using
[07:14] <visiot> pantanol:
[07:15] <pentanol> visiot then what you did?
[07:16] <visiot> pentanol: i'm streaming udp multicast an SPTS and piping the output of it to opencaster to insert null packets to achjieve CBR
[07:16] <visiot> as i haven't seen anything related to null packet insertyion in ffmpeg
[07:17] <visiot> but when i'm montiring the transport stream , with stream analyzer i'm getting pcr continuity error
[07:17] <visiot> pcr repititon error
[07:17] <visiot> and video isn't playing even , it got stucked at one frame
[07:19] <visiot> pentanol : this is all i've
[07:52] <pentanol> visiot ffmpeg not saying anything?
[07:54] <visiot> pentanol :  can we add null packets to a transport stream through ffmpeg
[08:42] <pentanol> visiot no, not sure
[09:35] <bencc> if I'm targeting desktop browsers can I use -preset veryslow
[09:40] <spaam> en /win 94
[10:02] <m_m> Hi. I need to cut my movie into parts. By parts i mean coorination parts. Ex. from left corner to the middle. Is it possible ?
[10:14] <m_m> Any Ideas. Please.
[11:09] <abijinx> Hi, I am trying to update ffmpeg libs in my android app from 0.8.6 to 2.3.4 with 64 bit support. I am getting lot of undefined reference errors with ***_neon. even after adding this in my makefile "aarch64/mpegaudiodsp_init.c \" May I know how to resolve the errors or disable neon if it is not required
[11:12] <BtbN> 0.8 is ancient and in no way api/abi compatible.
[11:13] <BtbN> You have to rebuild and propably fix the code of everything using that old stuff
[11:16] <abijinx> I have configured my ./configure_android.sh file and all the make files to latest NDK. I have used the same flags in configure_android.sh file
[11:16] <BtbN> So?
[11:17] <BtbN> You still won't be able to use code which is written against 0.8 with 2.3
[11:17] <Mavrik> BtbN, um
[11:17] <Mavrik> he's trying to compile ffmpeg for Android
[11:18] <Mavrik> what are you talking about? O.o
[11:18] <BtbN> He has integrated it into his App, aparently entirely bypassing the ffmpeg build system.
[11:19] <Mavrik> abijinx, it's going to be way easier if you use a standalone toolchain to compile ffmpeg "properly"
[11:19] <Mavrik> and then link against static/shared libraries only in your NDK build scripts
[11:19] <abijinx> Yes. I am compiling ffmpeg for android. I am getting those errors at that stage itself.
[11:19] <BtbN> And you are not using the ffmpeg configure script for that?
[11:23] <abijinx> actually, I was able to build the vanilla ffmpeg code. Now I am trying to build the code with a wrapper which is used in that app. Hence using the NDK and other configuration files similar to what was used in 0.8.6 code.
[11:24] <BtbN> So you are trying to build ffmpeg without using its own configure script. That's now going to work.
[11:24] <BtbN> Also, your wrapper most likely also won't build anymore.
[11:31] <abijinx> I am just enabling and disabling the flags in the configure script, Other than that all files are intact
[11:36] <BtbN> So what do you mean with 'even after adding this in my makefile "aarch64/mpegaudiodsp_init.c \"' then?
[11:43] <plode> hello
[11:44] <plode> Ima having a problem
[11:44] <plode> I was trying to upgrade ffmpeg
[11:44] <plode> and it wouldn't update to a newer version
[11:44] <plode> so I went and removed it with apt
[11:45] <plode> and then added jon sevverinson's ppa
[11:45] <plode> and then installed to install from his software source
[11:45] <plode> but nothing
[11:45] <plode> now my terminal says I have no software installation candidate for ffmpeg
[11:46] <abijinx> There is a file to mention all the source files that are to be included in the build. These neon functions are defined in the "aarch64/mpegaudiodsp_init.c \" file. So even after including that file, I am getting the undefined ref error
[11:46] <plode> halp!
[11:47] <plode> @abijinx sounds more like #ffmpeg-devel than user
[11:48] <plode> or were you trying to build from source
[11:52] <abijinx> I have built the vanilla ffmpeg. Then included that built src with libs in my jni folder with wrapper. Then building with NDK.. Am I doing something wrong? This was the procedure followed while using 0.8.6 version.
[12:13] <tyr> langxm at langxm-ThinkPad-L440:~/flv$ sudo ffmpeg -f rtp -ac 1 -i hw:CARD=PCH  -f video4linux2 -i /dev/video0 ~/flv/out.flv
[12:13] <tyr> ffmpeg version 2.5 Copyright (c) 2000-2014 the FFmpeg developers
[12:13] <tyr>   built on Dec 10 2014 17:48:49 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[12:13] <tyr>   configuration:
[12:13] <tyr>   libavutil      54. 15.100 / 54. 15.100
[12:13] <tyr>   libavcodec     56. 13.100 / 56. 13.100
[12:13] <tyr>   libavformat    56. 15.102 / 56. 15.102
[12:13] <tyr>   libavdevice    56.  3.100 / 56.  3.100
[12:13] <tyr>   libavfilter     5.  2.103 /  5.  2.103
[12:13] <tyr>   libswscale      3.  1.101 /  3.  1.101
[12:13] <tyr>   libswresample   1.  1.100 /  1.  1.100
[12:13] <tyr> hw:CARD=PCH: Protocol not found
[12:13] <tyr> protocol not found
[12:13] <tyr> why?
[12:36] <spaam> tyr: try replace -f rtp with alsa
[12:37] <spaam> -f alsa..
[12:37] <tyr> sudo ffmpeg -f rtp -ac 1 -i hw:CARD=PCH  -f video4linux2 -i /dev/video0 ~/flv/out.flv
[12:38] <tyr> then i got the same result
[12:38] <tyr> hw:CARD=PCH: Protocol not found
[12:38] <tyr> i have some problemm with alsa
[12:38] <tyr> langxm at langxm-ThinkPad-L440:~$ sudo ffmpeg -f alsa -ac 1 -i hw:CARD=PCH  -f video4linux2 -i /dev/video0 ~/flv/out.flv
[12:38] <tyr> ffmpeg version 2.5 Copyright (c) 2000-2014 the FFmpeg developers
[12:38] <tyr>   built on Dec 10 2014 17:48:49 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[12:38] <tyr>   configuration:
[12:38] <tyr>   libavutil      54. 15.100 / 54. 15.100
[12:38] <tyr>   libavcodec     56. 13.100 / 56. 13.100
[12:38] <tyr>   libavformat    56. 15.102 / 56. 15.102
[12:38] <tyr>   libavdevice    56.  3.100 / 56.  3.100
[12:38] <tyr>   libavfilter     5.  2.103 /  5.  2.103
[12:38] <tyr>   libswscale      3.  1.101 /  3.  1.101
[12:38] <tyr>   libswresample   1.  1.100 /  1.  1.100
[12:38] <tyr> Unknown input format: 'alsa'
[12:39] <tyr> seems my computer does not suport alsa format
[13:19] <wallbroken> what's the difference between CounterClockwise and Clockwise ?
[13:21] <azk> left->right vs right->left
[13:22] <c_14> https://en.wikipedia.org/wiki/File:Clockwise_arrow.svg
[13:22] <c_14> https://en.wikipedia.org/wiki/File:Counterclockwise_arrow.svg
[14:20] <tyr> i changed ffserver.config ,but it does not work
[14:21] <tyr>  17 <Stream live.flv>
[14:21] <tyr> i change live.flv to anything ,but only access video via :8090/test.flv
[14:29] <Popara> Hello, any way to change font size etc in the -vf subtitles ?
[16:31] <shevy> guys I get an error like "Encoder (codec mp3) not found for output stream #0:0"
[16:31] <shevy> if I reconfigure ffmpeg with --enable-libmp3lame, do you think this error will go away?
[17:00] <Elirips> Hello. The doc says about '-strict': 'strict integer (decoding/encoding,audio,video)      Specify how strictly to follow the standards. ' But what is the int-value? Does it start at 0 and 0 is 'very' ?
[17:01] <shevy> reconfiguring ffmpeg fixed it. bye
[17:03] <Elirips> Btw, this is amazing how easy hlc works with ffmpeg :D
[17:04] <Elirips> err.. mean hls
[17:30] <pierre_> where can i find ffmpeg-2.4.3 for windows with shared libraries, ?  i am looking at google, but nowhere to find, where can download archive binaries for windows ?
[17:50] <b_jonas> pierre_: look in http://ffmpeg.zeranoe.com/builds/win64/shared/
[19:38] <canbal> Hi guys, Im trying to enable multithreading for H264 decoding. There have been some discussions in the mailing list, but I couldnt find any resources if its already made available. Can anyone direct me to the right resources?
[19:55] <llogan> canbal: ffmpeg -h decoder=h264
[19:55] <llogan> Threading capabilities: frame and slice
[19:56] <canbal> @llogan, Thanks for the info. Do you know how to enable the threads in the C++ library?
[19:57] <canbal> And where can I find the details of the difference btw frame and slice mt
[19:57] <llogan> FFmpeg is written in C.
[19:57] <canbal> Sorry, yes C
[19:57] <canbal> Im using it in a C++ code using extern C
[19:57] <canbal> My bad
[19:58] <llogan> i don't know much about the libraries. you could always look in doc/examples, and refer to the doxygen
[20:08] <canbal> @llogan thanks!
[20:08] <llogan> and you could ask on the libav-user mailing list. the FFmpeg libraries ML
[20:25] <justinX> would https://launchpad.net/~mc3man/+archive/ubuntu/trusty-media be the right repository to add to get ffmpeg in linuxmint17?
[20:28] <justinX> would there be any reason to go with an older version of ffmpeg than 2.5, that was released last week ?
[20:31] <llogan> justinX: is mint 17 derived from Trusty?
[20:31] <justinX> yes, mint17 (actually mint17.1) is based on ubuntu14.04
[20:32] <llogan> then yes, it should be fine. it will just provide ffmpeg, ffplay, and ffserver, as far as i know
[20:32] <justinX> "just" as in there are other parts?
[20:33] <llogan> it doesn't provide the libraries
[20:33] <llogan> i think
[20:34] <llogan> or you could compile http://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[20:35] <justinX> what I want to accomplish is going through lots of vacation film here that takes to much space. Converting them from 50fps 1280×720 to half that resolution in 25fps
[20:35] <llogan> or download: http://johnvansickle.com/ffmpeg/
[20:36] <justinX> I don't think I would be able to compile ffmpeg and stuff, I might be able to compile a small tetris that doesn't depend on any libraries :-D
[20:36] <llogan> if you can copy and paste you can compile
[20:36] <llogan> but the PPA will be fine for you
[20:36] <justinX> somehow I always fail with some strange error when I try compile things
[20:37] <justinX> yeah I think I go with PPA for now, maybe try compiling things again later :-)
[20:39] <justinX> the differnce with johnvansickle is that would be installed in my user home directory instead or? so like a "ffmpeg portable" so to speak?
[20:41] <llogan> you could place it anywhere you like. you could execute it via the full path to this binary, or place it somewhere in your PATH
[20:45] <justinX> the mp4-files I want to convert is H264 - MPEG-4 AVC (part 10) (avc1) video, with MPEG AAC Audio (mp4a) sound, says vlc. anything special I should do when converting to same format only half resolution (to 640x360) and 25 frames/s instead of 50 ?
[20:45] <justinX> with the same quality ideally if that is possible....
[20:47] <justinX> well... perhaps I should pack down the sound to save some space too... to mono maybe? I don't know if there is much to save there...
[20:50] <llogan> justinX: use a pastbin link to show the complete console output of: ffmpeg -i input
[20:50] <justinX> ok one moment
[21:11] <justinX> llogan: I suppose I should now upgrade all the packages that is marked as upgradeable too, right? like vlc to version "2.2.0+ppa2.5"
[21:14] <llogan> justinX: i don't know what the PPA's offering of VLC does and differently than repo version. newer version? I'm not sure. I'm not an Ubuntu user.
[21:14] <justinX> llogan: here is the consol text for ffmpeg -i HDV_0071.MP4 http://pastebin.com/Ka1umUqT
[21:15] <llogan> you could always ask questions specific to that PPA on the ubuntuforums. the author (mc4man) is active and responsive.
[21:18] <justinX> llogan: ok, and maybe updating vlc isn't so importan just now, but there are lots of library packages that is now marked as upgradeble if I look at packas from LP-PPA-mc3man-trusty-media/trusty and maybe those are more important like libaacs0
[21:18] <justinX> oh
[21:22] <justinX> was the console output of ffmpeg showing what you wanted to see?
[21:23] <llogan> yes, i mostly wanted to see how many channels of audio there was. just stream copy the audio: -c:a copy
[21:23] <pzich> shouldn't that be in ffprobe?
[21:25] <justinX> ok. no idea to save some MB's on the audio part then
[21:27] <llogan> it's not worth degrading it further to save little. do you want to host these on your own web site, or are you just storing them on your hard drive?
[21:28] <justinX> (well not on exact that file ofcourse but on HDV_0449.MP4  1,6GB could be something)
[21:29] <justinX> currently they are occuping lots of space on my harddisk, but I want to burn them on on a DVD or two to free the disk space
[21:30] <justinX> and maybe keep some of them on the harddisk.
[21:30] <justinX> (not burning as a dvd film I mean, just mp4 files)
[21:30] <wallbroken> llogan, in scale is better to use -1 or -2 ?
[21:31] <llogan> wallbroken: http://ffmpeg.org/ffmpeg-filters.html#scale explains the difference
[21:33] <llogan> justinX: then "ffmpeg -i input -vf fps=25,scale=iw/2:-2 -crf 23 -c:a copy output.mp4" may do what you want. might look shitty with half frame rate. adjust crf until you get an acceptable quality.
[21:33] <llogan> see https://trac.ffmpeg.org/wiki/Encode/H.264
[21:33] <llogan> once you get an acceptable output you can encode all files in a directory with a bash for loop
[21:35] <pzich> llogan: does -vf fps=25 differ from -r 25?
[21:36] <justinX> hmm it seems man file is not installed :-/
[21:36] <llogan> i guess the PPA doesn't install it. 2.5 is close enough to git master that you can refer to the online docs
[21:36] <wallbroken> llogan, i don't find it
[21:37] <justinX> ok I try with one of the smaller films first and see how it looks.
[21:37] <llogan> pzich: yes, but i'm not sure what the differences are.
[21:37] <llogan> trawl ffmpeg-user mailing list
[21:37] <llogan> or ask there
[21:38] <pzich> cool, thanks
[21:38] <llogan> wallbroken: it's under width, w height, h
[21:38] <llogan> If one of the values is -n with n > 1 ...
[21:38] <justinX> is the ,scale=iw/2:-2  same thing as would happen if I did a -s 640x360 ?
[21:38] <llogan> in this case, yes
[21:39] <llogan> but it's easier because it will calculate the values for you
[21:39] <pzich> it sets the scale to half of the width, with aspect ratio matching the height, where the dimensions are a divisible by 2
[21:40] <justinX> ok
[21:41] <wallbroken> llogan, there is not -2
[21:41] <llogan> it is implied with, "If one of the values is -n with n > 1"
[21:42] <wallbroken> oh yes
[21:42] <wallbroken> so why a size should be divisible by n ?
[21:42] <wallbroken> what's the advantage?
[21:42] <pzich> some codecs require it
[21:43] <llogan> try it: ffmpeg -f lavfi -i testsrc=s=501x501:d=3,format=yuv420p -c:v libx264 -f null -
[21:43] <llogan> it will complain
[21:43] <llogan> i mean libx264 will complain.
[21:44] <wallbroken> x264 is not the default encoder on .mov files?
[21:44] <iive> for example if you use yuv420 colorspace, the chroma (color) resolution half width and half height of the luma.
[21:44] <iive> how do you round an odd luma resolution?
[21:45] <iive> also, the encoder usually support something called edge mirroring or edge extension..
[21:46] <iive> it does matter how you handle edges
[21:47] <llogan> wallbroken: i'm guessing libx264 should be default encoder for MOV container if your build has been configured with --enable-gpl --enable-libx264
[21:47] <wallbroken> the build is on windows
[21:47] <wallbroken> it is?
[21:47] <llogan> how am i supposed to know?
[21:47] <wallbroken> ok, you don't know
[21:48] <wallbroken> is there a way to know?
[21:48] <llogan> run "ffmpeg", then look at the output. look for a line that starts with "configuration"
[21:49] <llogan> or just try encoding to mov and look at console output: ffmpeg -f lavfi -i testsrc=d=5,format=yuv420p output.mov
[22:13] <justinX> I tried with crf 23 and it didn't look too bad actually and went 1/15 the filesize. I should try with some other values too first and some other video that don't have lots of snow :-)
[22:37] <justinX> in an completly uncompressed world a file going half x, half y, half framrate would be 1/8 the size so I guess that could perhaps be a ratio to try go for
[23:28] <bakers> I have an old .avi that looks like this http://www.fpaste.org/158149/64115141/
[23:28] <bakers> I want to convert that so it's playable on my chromecast
[23:29] <bakers> Is there a lossless way to do that?
[23:40] <justinX> bakers: https://developers.google.com/cast/docs/media says supported chromecast formats is MP4/H.264 or WebM/VP8 so it would have to be converted to that then?
[23:41] <bakers> justinX: I've been converting the XVid to h.264 and leaving the MP3 audio
[23:41] <bakers> that works on the Chromecast
[23:41] <justinX> nice
[23:41] <bakers> but it's a lossy conversion
[23:41] <justinX> ah...
[23:41] <bakers> Just making sure I'm not missing something obvious
[23:43] <justinX> yeah you would want something that transform the xvid to h264 format, my guess is h264 can do everything xvid can so it would be possible to do it lossless, but I really don't know these things...
[23:45] <BtbN> Lossless transcoding?
[23:45] <BtbN> That simply doesn't exist.
[23:45] <bakers> BtbN: I just meant does the CC support Xvid in a different container
[23:46] <bakers> Isn't XVid a subset of h.264?
[23:46] <BtbN> no
[23:46] <BtbN> xvid/divx is mpeg4 stuff
[23:47] <bakers> BtbN: Ok good
[23:48] <klaxa> for the record, with -crf 0 or preferably -qp 0 x264 encodes lossless
[23:48] <klaxa> i don't remember why -qp is preferred over -crf
[23:49] <BtbN> crf is the encoding mode. qp is the quality value
[23:49] <llogan> just matters if you're using a 10-bit build of x264 then the -crf would differ (or used to at least; not sure of current situation)
[23:50] <BtbN> a lossless encoding would propably be way larger than the original file
[23:50] <BtbN> just aim for the same filesize as the input file, and you won't have too much of a quality loss
[23:50] <BtbN> h264 is way more efficient than mpeg4
[23:50] <llogan> ignore that
[23:50] <bakers> BtbN: Ok sounds like a plan.
[23:50] <bakers> Thanks for the feedback
[23:51] <llogan> don't try to copy the bitrat. that doesn't make sense
[23:51] <BtbN> Trying to go all out on compression will just make it look horrible
[23:52] <BtbN> If size is not an issue, why bother?
[23:52] <BtbN> lossless is just a huge waste though
[23:52] <llogan> klaxa: so if you're using a regular 8-bit build of x264 use either -qp 0 or -crf 0. should give the exact same output
[23:53] <bakers> It's an old 90s TV show. THe original quality isn't that good
[23:53] <bakers> I'm doing a -crf 25 conversion and it looks "good enough" for me
[23:53] <BtbN> qp 0 is lossless, and will give you a huge output file...
[23:53] <BtbN> everything below 18 is already considdered visualy lossless
[23:53] <bakers> I just didn't want to do a lossy conversion if I didn't have to. If I were missing something about the Chromecast
[23:54] <BtbN> The Chromecast is extremely limited in the formats it can play, for no good reason
[23:54] <BtbN> the cpu in that thing is more than fast enough to deal with mpeg4
[00:00] --- Wed Dec 10 2014