[Ffmpeg-devel-irc] ffmpeg.log.20151021

burek burek021 at gmail.com
Thu Oct 22 02:05:01 CEST 2015


[00:02:39 CEST] <sor_> anYc, it would appear to be a problem with ffplay
[04:54:06 CEST] <sor_> anYc, I figured the problem out sorry for all the trouble, it work snow with ffmpeg and your cool program that i am going to add to my /bin... thanks again
[05:44:33 CEST] <Renari> I'm trying to convert a series of images to a video
[05:44:35 CEST] <Renari> ffmpeg -framerate 24 -i AM_TOZ_OP_001%04d.png -c:v libvpx-vp9 -pix_fmt yuv420p -b:v 6000k out.webm
[05:44:53 CEST] <Renari> Can someone tell me why when using this command the output bitrate is 3k and not 6k?
[05:49:45 CEST] <relaxed> Renari: did you see https://trac.ffmpeg.org/wiki/Encode/VP9 ?
[05:50:10 CEST] <relaxed> maybe it can't use that many bits
[05:51:12 CEST] <Renari> Hm, I'll try that then. Thanks.
[06:35:54 CEST] <ibrahim_> hi, good morning i followed step found in hear `https://trac.ffmpeg.org/wiki/CompilationGuide/Centos` but i have an error in this picture `http://imgur.com/08VFwoy`
[06:37:55 CEST] <ibrahim_> in breif the compile of ffmpeg  without `--enable-libvpx` succeeded and with '--enable-libvpx' fails
[06:38:13 CEST] <ibrahim_> so i googled alot but no lick
[07:38:34 CEST] <ibrahim_> hey anybody here
[07:42:21 CEST] <relaxed> ibrahim_: there were some changes in libvpx git that broke support. You can download 1.4.0 or try git again in a few days.
[07:43:53 CEST] <ibrahim_> @relaxed: i tried libvpx-1.4.0.tar.bz2 but give me the same error
[07:44:28 CEST] <ibrahim_> from here `http://www.linuxfromscratch.org/blfs/view/svn/multimedia/libvpx.html`
[07:44:58 CEST] <relaxed> ibrahim_: run make clean; make distclean in ffmpeg and try again
[07:45:40 CEST] <ibrahim_> ok i will try again i let you know the result
[07:45:43 CEST] <ibrahim_> thanks
[07:46:54 CEST] <relaxed> ibrahim_: also make sure you removed the files installed from libvpx git
[07:48:27 CEST] <ibrahim_> @relaxed: how can i know the files installed form libvpx git
[07:48:52 CEST] <relaxed> what prefix did you use to install?
[07:49:38 CEST] <relaxed> my static builds have libvpx support, http://johnvansickle.com/ffmpeg/
[07:49:52 CEST] <ibrahim_> --prefix="$HOME/ffmpeg_build"
[07:51:11 CEST] <relaxed> if you don't mind to start over, rm -rf $HOME/ffmpeg_build/*
[07:51:39 CEST] <ibrahim_> ok i will
[07:51:50 CEST] <ibrahim_> thanks @relaxed:
[09:04:46 CEST] <anYc> sor_: thanks for trying! I'll check if I can reproduce it with ffplay
[09:57:02 CEST] <Mysoft> so, i'm trying to decode a cineform video... (on windows) any directshow based video player... can play it... since i have the propietary encoder installed... but how can i make ffmpeg to use that codec to decode the video? or any other tool that i can use from command line to decode it and pass to ffmpeg...
[10:23:19 CEST] <kahrl> hi, when converting a stream to mp4, I can't play back the video until ffmpeg is done converting
[10:23:35 CEST] <kahrl> while ffmpeg is converting, mpv says: [ffmpeg/demuxer] mov,mp4,m4a,3gp,3g2,mj2: moov atom not found
[10:24:08 CEST] <kahrl> is there a switch so ffmpeg periodically writes a moov atom so I can watch the video while it is still converting?
[10:25:17 CEST] <kahrl> (or writes it once at the beginning, assuming that would be sufficient)
[10:26:08 CEST] <Anoia> kahrl: it will have a moov atom, but I expect it'll be truncated (not finished writing yet) so the viewer ignores it
[10:26:25 CEST] <Anoia> IIRC, the moov contains each frame
[10:26:30 CEST] <kahrl> I see
[10:26:39 CEST] <Anoia> or is taht AVIs...
[10:47:13 CEST] <c_14> kahrl: you can use moov_fragment (or whatever the fflag is called) to create a fragmented moov atom (many players won't support that though)
[10:47:59 CEST] <c_14> ie -movflags frag_keyframe
[10:49:58 CEST] <kahrl> ah nice. I also thought about trying -f ismv
[10:50:21 CEST] <kahrl> now my problem is that I don't actually know how to get youtube-dl to pass these options on to ffmpeg, but that's offtopic for here
[10:51:29 CEST] <Mavrik> kahrl, why not use mpegts?
[10:51:44 CEST] <Mavrik> It's built for streaming and will play immediately. MP4 MOOV just cannot be generated before the file is done.
[10:53:24 CEST] <kahrl> Mavrik: thanks, I'll try that as well
[10:56:43 CEST] <Mysoft> ok fixed my problem using avysynth as input... altough directshow decoding the 4k is using 700mb and encoding a 4k video with too many reference frames makes it blown my 2gb limit :D
[10:56:48 CEST] <Mysoft> thanks anyway
[11:00:47 CEST] <hero_biz> guys
[11:02:17 CEST] <hero_biz> i want to just deinterlace and scale a 1080i .ts file (1440x1080). is following command correct:
[11:02:19 CEST] <hero_biz> ffmpeg -i file.ts -vf "yadif=1,scale=1280x720" -sws_flags lanczos out.mkv
[11:04:28 CEST] <Mavrik> hero_biz, might wanna copy the audio track so you won't reencodfe
[11:04:38 CEST] <Mavrik> hero_biz, and set video codec and CRF explicitly for quality.
[11:06:44 CEST] <hero_biz> marviv, I did this on purpose so I could encode video later with x264. but I will add audio.
[11:06:56 CEST] <hero_biz> but my main problem is that:
[11:07:25 CEST] <hero_biz> 1- scaling to 1280x720 will be correct?or it will be considered as upscale?
[11:07:53 CEST] <hero_biz> 2- is my yadif option correct? what is best approach for deinterlaceing?
[11:08:28 CEST] <hero_biz> i have seen people use "yadif=3:1,mcdeint=2:1" for deinterlacing too.
[11:10:13 CEST] <Mavrik> Yadif is fine.
[11:10:24 CEST] <Mavrik> And I don't understand your "I'll encode later" comment.
[11:10:30 CEST] <Mavrik> If you mess up the quality here, no encoding later will fix it.
[11:11:02 CEST] <hero_biz> I need to encode video here?
[11:11:10 CEST] <hero_biz> while deinteracing?
[11:11:12 CEST] <c_14> Unless you output rawvideo, yes.
[11:11:32 CEST] <c_14> ffmpeg "always" encodes output
[11:11:55 CEST] <hero_biz> unless i use -c:v copy?
[11:13:03 CEST] <hero_biz> i hav never deinterlaced a video before, then I don't know its process exactly.
[11:13:04 CEST] <c_14> You can't use that when filtering the video
[11:13:14 CEST] <hero_biz> oh
[11:13:21 CEST] <c_14> In order to apply a video filter to a video, ffmpeg has to decode the video.
[11:13:32 CEST] <c_14> Unless you then output the decoded rawvideo, you need to reencode it to something.
[11:14:12 CEST] <hero_biz> then how you encode to something that can be encodd with x264 later?
[11:15:07 CEST] <Mavrik> hero_biz, of course you have to encode the video
[11:15:20 CEST] <Mavrik> you can't deinterlace encoded video, frames change afterall.
[11:16:36 CEST] <hero_biz> what should i use for encoding too?
[11:16:52 CEST] <hero_biz> i saw a commaand like this for video deinterlacing too: ffmpeg -i file.ts -vf "yadif=0:0:0,scale=1280x720" -sws_flags lanczos -f yuv4mpegpipe -pix_fmt yuv420p - | x264_64.exe --stdin y4m --level 3.1 --output output.mkv -
[11:17:28 CEST] <hero_biz> so i thought maybe i can save ffmpeg result to a file and then use x264 on resulting file.
[11:17:50 CEST] <Mavrik> Is there a reason you don't wanna do that in a single step?
[11:18:09 CEST] <Mavrik> Since it'll cause you significantly less problem, storing raw video takes up a huge amount of space and brings a ton of problems.
[11:18:15 CEST] <hero_biz> my ffmpeg is not compiled with libx264
[11:18:40 CEST] <hero_biz> so i thought to use x264 for video later.
[11:20:01 CEST] <hero_biz> any good option to preserve quality in ffmpeg so I could encode it later again?
[11:21:49 CEST] <relaxed> hero_biz: http://johnvansickle.com/ffmpeg/
[11:26:48 CEST] <hero_biz> ty for link relaxed, but sadly my kernel is too old for it :D
[11:35:00 CEST] <hero_biz> encoding with this command is not good idea? ffmpeg -i file.ts -vf "yadif=0:0:0,scale=1280x720" -sws_flags lanczos -f yuv4mpegpipe -pix_fmt yuv420p - | x264_64.exe --stdin y4m --level 3.1 --output output.mkv -
[11:38:19 CEST] <Mavrik> It still doesn't set quality.
[11:38:29 CEST] <Mavrik> And you're piping stuff through for no apparent reason.
[11:38:40 CEST] <Mavrik> Just burning CPU.
[11:40:20 CEST] Action: hero_biz had copy/paste it from what has found in internet.
[11:40:41 CEST] <hero_biz> marvik, what will you use as command if you don't have libx264?
[11:41:06 CEST] <hero_biz> i mean ffmpeg is not compiled with libx264?
[11:46:41 CEST] <Mavrik> You compile ffmpeg with libx264.
[11:46:43 CEST] <Mavrik> You got a link.
[11:47:20 CEST] <hero_biz> hm.....k,k :)
[11:47:27 CEST] <hero_biz> I willl recompilel again
[11:48:01 CEST] <Mavrik> Trust me, way easier than alternatives.
[11:48:07 CEST] <Mavrik> Get fdk_aac while you're at it.
[11:48:19 CEST] <n_tish> Hie all, I have one issue. When I try to load native libavcodec.so libavformat.so libavutil.so in Android I get the Relocation error. The android OS is M and the target is 23.
[11:48:56 CEST] <hero_biz> @marvik, I compiled it with fdk_aac
[11:49:21 CEST] <hero_biz> my compilation is this: configuration: --enable-shared --enable-openssl --disable-ffplay --enable-libfdk-aac --enable-libmp3lame --enable-libopus
[11:49:33 CEST] <n_tish> the error is "java.lang.UnsatisfiedLinkError: dlopen failed: /data/app/com.my.application/lib/arm/libav.so: has text relocations "
[11:52:45 CEST] <benbro> anYc: now avcut run without errors
[11:53:13 CEST] <benbro> but there is a gap between the first and middle parts (maybe end as well, didn't check)
[11:53:21 CEST] <benbro> maybe you are skipping a keyframe?
[12:00:06 CEST] <benbro> anYc: ok. the API let you drop parts. I thought that you choose what part to keep
[12:00:12 CEST] <anYc> hm, at least at the end there shouldn't be missing frames as I compared the last frames of the input and output video yesterday
[12:00:40 CEST] <benbro> anYc: what if I have 60 seconds and I want to keep only 40 seconds in the middle?
[12:00:55 CEST] <benbro> do I need to drop the last part with exact frame accuracy?
[12:01:13 CEST] <benbro> or can I just leave the last part and avcut we'll fill it for me?
[12:01:29 CEST] <benbro> avcut in.mp4 out.mp4 0 10 50 60
[12:01:30 CEST] <anYc> ah ok. you use "0 10 50 60" to drop the first and last 10 seconds. But I'm thinking about adding a switch to accept a "whitelist"
[12:01:45 CEST] <benbro> anYc: a whitelist will be nice
[12:01:49 CEST] <benbro> maybe:
[12:01:57 CEST] <benbro> avcut in.mp4 out.mp4 0 10 50
[12:02:12 CEST] <benbro> will automatically expand the last part to the exact video length?
[12:02:20 CEST] <anYc> in your case it would be: out.mp4 10 50
[12:02:38 CEST] <benbro> in whitelist but how do I do it with blacklist?
[12:02:56 CEST] <benbro> the last part has to match the exactu duration or will you expand it automatically?
[12:03:07 CEST] <benbro> "avcut in.mp4 out.mp4 0 10 50" last part is missing. avcut could guess it
[12:03:18 CEST] <benbro> maybe "avcut in.mp4 out.mp4 0 10 50 -"
[12:03:29 CEST] <benbro> or "avcut in.mp4 out.mp4 0 10 50 end"
[12:03:38 CEST] <anYc> right now you have to provide a sufficiently large last value if you don't know the exact length, e.g., you can write "50 99999" to drop all frames after 50 seconds
[12:03:49 CEST] <benbro> ok
[12:03:56 CEST] <anYc> yeah, I also thought about using "-" or something similar
[12:04:01 CEST] <benbro> whitelist will be useful
[12:04:14 CEST] <benbro> I'll test it with my videos.
[12:04:44 CEST] <benbro> my python script has an issue where I combine the parts so yours probably better :)
[12:04:53 CEST] <benbro> cutting takes a bit of time
[12:05:04 CEST] <benbro> is it because you extract all the frames?
[12:05:19 CEST] <benbro> maybe you can just extract the frame info around the borders?
[12:05:52 CEST] <benbro> "avcut in.mp4 out.mp4 20 40" -> you only need the info near 20 and near 40
[12:05:52 CEST] <anYc> ok, thank you! Yes, I also think decoding takes the biggest part. I was thinking about some optimizations but I tried to get it to work first
[12:06:01 CEST] <benbro> cool
[12:06:17 CEST] <benbro> do you expect it to be stable now?
[12:08:28 CEST] <anYc> I only tested it shortly with three videos yesterday as I had to go to bed but with those it worked flawlessly
[12:08:49 CEST] <anYc> but I wouldn't delete the original videos yet ;)
[12:08:53 CEST] <benbro> ok
[12:11:02 CEST] <anYc> I will also test the result with more players. As avcut sometimes has to change the format to mix frames from different encoding sessions there might be players that don't like this
[12:14:06 CEST] <benbro> different encoding in the same file?
[12:19:17 CEST] <anYc> different encoding sessions. I don't know much about the actual h264 internals but some parameters of the encoder cannot be read from the input file and I had to set them manually (like quality). Hence, the extradata ("header" data) might differ between the copied and the new encoded frames
[12:21:24 CEST] <anYc> it seems like at least mplayer, mpv and vlc don't have an issue with it
[12:21:57 CEST] <anYc> for me it's also important that they work with kodi (xbmc)
[12:42:59 CEST] <pgunnars> yo
[12:51:48 CEST] <yongyung> Does someone here have a prebuilt 64 bit Windows built of ffmpeg with all the "usual" libs (like http://ffmpeg.zeranoe.com/builds/), but with libfdk_aac?
[12:51:58 CEST] <yongyung> -t +d
[12:52:50 CEST] <pgunnars> I'm slicing a segment out of a video stream I'm playing with ffserver, how can I get the segment to include a duration?
[12:54:53 CEST] <c_14> yongyung: FFmpeg built against libfdk_aac is non-redistributable
[12:56:01 CEST] <yongyung> c_14: That's why I'm asking here, I guess ;) Building it myself is going to be a pita...
[13:05:00 CEST] <pgunnars> can any1 help me with my question?
[13:05:55 CEST] <pgunnars> ffserver outputs a stream (so no set duration), but I'm slicing out a specific segment with -t duration. What I need is for that video slice to have a duration field, doesn't come automatically
[13:06:44 CEST] <bove> Can a concat list hold multiple streams (video and audio) on each line?
[13:08:50 CEST] <pgunnars> is it possible to just set the video length to duration?
[13:11:33 CEST] <pgunnars> fprobe -i birds.mp4 -show_format -v quiet | sed -n 's/duration=//p' just gives me N/A
[13:36:40 CEST] <benbro> anYc: maybe we can pass these paramters that you can't read manually in the command line?
[13:54:23 CEST] <anYc> good idea, yes
[14:57:05 CEST] <__jack__> ffmpeg -i ... -hls_ts_options fix_teletext_pts=0 output.m3u8 : what's wrong with that ?
[14:57:35 CEST] <__jack__> (http://pastebin.com/qWLnwvrf)
[15:32:26 CEST] <pgunnars> how can a video file not know its duration
[15:32:30 CEST] <pgunnars> it knows the framerate
[15:32:35 CEST] <pgunnars> it must know the amount of frames rightÞ
[15:32:36 CEST] <pgunnars> ?
[15:32:39 CEST] <BtbN> No.
[15:33:08 CEST] <pgunnars> how can I let it know the amount of frames
[15:33:16 CEST] <BtbN> Parse the entire file and count.
[15:33:54 CEST] <pgunnars> can I then tell the file how many frames it has?
[15:34:16 CEST] <BtbN> Not neccesarily.
[15:34:32 CEST] <BtbN> Depends entirely on the container which kind of metadata it supports.
[15:34:37 CEST] <pgunnars> webm
[15:36:20 CEST] <pgunnars> can i do it then?
[15:57:32 CEST] <pgunnars> any1 know how I can simulate ffserver output stream, so I can create tests for ffmpeg execution with ?date on the stream
[16:33:49 CEST] <phibonacci> Hello. My old MP3 player, an Archos that supports mp3, ogg, flac, can't read one of my FLAC songs so I converted it in ogg with FFmpeg but it reads it as a 15mn long song even so it's 5mn long. Do you have any idea what could create this issue and how to fix it? Everything works fine with both the FLAC and ogg on VLC.
[17:26:01 CEST] <sor_> anYc, I figured the problem out sorry for all the trouble, it work snow with ffmpeg and your cool program that i am going to add to my /bin... thanks again
[17:40:34 CEST] <brbblnch> hello
[18:03:13 CEST] <dot> hello, why doesnt video size work in this case: ffmpeg -f avfoundation -i "<screen device index>:<audio device index>" out.mov
[18:03:33 CEST] <dot> even tough I set this: -video_size 1280x720
[18:30:01 CEST] <worst[je]> I am trying to adapt a script I found online which should be ready to interact with the exact kind of files I'm using it on to adapt it to my needs. However.. it does not seem to work to begin with. Can someone point out where the problem lies? http://pastebin.com/zS8rGVnw
[18:42:58 CEST] <sor_> worst[je], it's not that you don't have the codec it's that it is using "incorrect codec parameters" -- i would say this is a clue "deprecated pixel format used, make sure you did set range correctly" -- my guess would be you are converting compressed to raw uncompressed and they are incompatible formats
[18:45:14 CEST] <c_14> Nah, you can ignore that warning.
[18:48:02 CEST] <worst[je]> I wouldn't know how to improve the parameters though. They are indeed gigantic raw files in a sense.. I know Windows needs a special codec in order to have media players and editing software interact with them, but that's about as far as it goes. All I know for sure is that people have used the commandline I used in that script successfully in the past, so I am pretty certain it ought to be
[18:48:03 CEST] <worst[je]> possible somehow.. my best guess right now is that ffmpeg may have seen some development or default settings change which screws it over.
[18:51:20 CEST] <c_14> Get rid of the -c:v rawvideo
[18:51:59 CEST] <c_14> (Or set it to wrapped_avframe)
[18:54:03 CEST] <worst[je]> A quick command line tests brings a lot of garbledeegook, so looks good! Will throw it into the script and see how the results come out, but I assume it ought to be fine as this involves the reading and not the output part of the command. :) Thanks!
[19:10:02 CEST] <brontosaurusrex> Can I use alac in mp4 with AVC video?
[19:12:18 CEST] <c_14> I think you have to use the mov muxer
[19:13:55 CEST] <brontosaurusrex> c_14: thanks
[20:15:48 CEST] <worst[je]> Suppose I have a set of AVI files. I want to concatenate and x264 re-encode these to eventually put them into a mp4 container. Should I re-encode all my files separately and then find a way to concatenate them whole, or should I first concatenate them all and then encode them in a single go? Since I have the first half of the former figured out, I'd prefer that, but then I am not sure which
[20:15:48 CEST] <worst[je]> concat option to use.. demuxer is my guess? So much to learn.
[20:18:48 CEST] <durandal_1707> there is concat filter
[20:21:58 CEST] <worst[je]> Right, but that says 'if you want to re-encode' when I check the FAQ. Since in that case (encode separately first, concatenate second) I am already done with all the encoding, I obviously wouldn't want to make ffmpeg re-encode it and shit all over the hard work x264 did, right?
[20:25:45 CEST] <durandal_1707> you can concat and after that encode
[20:26:10 CEST] <durandal_1707> In one command
[20:27:02 CEST] <aargh> hi all
[20:27:47 CEST] <aargh> i have a question (hopefully a simple one) about ffmpeg command line options
[20:28:48 CEST] <aargh> i have a stereo mp2 source, but when converting to mp3 the output becomes joint stereo
[20:29:12 CEST] <aargh> how can i force stereo output?
[20:29:51 CEST] <JEEB> it is stereo IIRC, just a method of coding it. the mode is IIRC defined by the encoder library, so you should try raising the bit rate
[20:30:02 CEST] <durandal_1707> look at encoder options
[20:30:09 CEST] <JEEB> and that, if it exists
[20:30:44 CEST] <durandal_1707> ffmpeg -h encoder libmp3lame
[20:30:48 CEST] <aargh> bitrate is 256k source and i specify 256k out
[20:31:28 CEST] <aargh> hmm, i'll try that, durandal_1707.  thanks
[20:31:40 CEST] <JEEB> then it might just be a setting. that said I have no idea why you'd be recoding it like that
[20:31:48 CEST] <durandal_1707> the options are also documented
[20:31:59 CEST] <JEEB> lossy and then some more lossiness without anything gained
[20:32:35 CEST] <aargh> yeah, the options are documented.  but not necessarily intelligible
[20:33:20 CEST] <durandal_1707> they should be simple to understand
[20:33:52 CEST] <aargh> if they were simple, i wouldn't be here.  but i do thanks you for the response
[20:35:08 CEST] <hero_biz> guys. how I can compile ffmpeg with 10 bit support? I need to compile x264 as 10 bit and then use its library to build ffmpeg?
[20:35:18 CEST] <JEEB> yes
[20:35:46 CEST] <JEEB> decoding is in always unless you disable the avc decoder, but x264 has to be built with 10bit encoding support
[20:36:24 CEST] <hero_biz> ty info
[20:36:51 CEST] <furq> libmp3lame encoder AVOptions:
[20:36:51 CEST] <furq>   -joint_stereo      <int>        E...A... use joint stereo (from 0 to 1) (default 1)
[20:36:55 CEST] <furq> that seems pretty intelligible to me
[20:39:41 CEST] <hero_biz> hm...
[20:39:57 CEST] <aargh> thanks, furq.  i was looking on that option as only a way to specify joint, not to turn it off.
[20:40:18 CEST] Action: hero_biz wonders how I can compile x264 and ffmpeg in a way that both support x264 and lavf...
[20:40:20 CEST] Action: aargh pats aargh on top of the head.....  amusing child
[20:40:38 CEST] <JEEB> aargh: btw do you know what joint stereo is?
[20:41:10 CEST] <hero_biz> when I want to compile x264, I need ffmeg for lavf support. for compiling ffmpeg I need x264 for libx264...
[20:41:28 CEST] <hero_biz> how should I compile that both support both features?
[20:41:33 CEST] <aargh> not well enough to explain it, JEEB.  but i have heard some people consider it inferior quality
[20:42:40 CEST] <JEEB> hero_biz: just build x264 with --disable-lavf and link it against FFmpeg
[20:42:40 CEST] <JEEB> done
[20:42:41 CEST] <furq> if anything it should be better quality with mp3
[20:43:32 CEST] <JEEB> hero_biz: if you really want to use x264cli with lavf support, you then build x264cli without libraries with lavf enabled
[20:43:35 CEST] <JEEB> done
[20:43:57 CEST] <hero_biz> ok,ty info :)
[20:43:58 CEST] <JEEB> the lavc/lavf thing is from before FFmpeg was easily correctly usable from the command line
[20:44:08 CEST] <aargh> furq - you are saying that in my scenario you would expect joint to be a better output option?
[20:44:34 CEST] <JEEB> "Some early MP3 encoders didn't make ideal decisions about what mode to use from frame to frame in joint stereo files, or how much bandwidth to allocate to encoding the side channel. This led to a widespread but mistaken belief that an abundance of M/S frames, or the use of joint stereo in general, always negatively impacts channel separation and other measures of audio quality. This is not an issue with mode
[20:44:40 CEST] <JEEB> rn encoders. Modern, optimized encoders will switch between mid-side coding or simple stereo coding as necessary, depending on the correlation between the left and right channels, and will allocate channel bandwidth appropriately to ensure the best mode is used for each frame."
[20:44:44 CEST] <furq> with a cbr encode from mp2 i wouldn't expect there to be any difference at all
[20:44:56 CEST] <hero_biz> btw JEEB, is it possible for x264 to use static libraries of lavf? or it only use dynamic ones?
[20:45:04 CEST] <furq> but joint stereo should provide identical quality at a lower bitrate
[20:45:10 CEST] <furq> s/cbr/256k cbr/
[20:45:17 CEST] <JEEB> hero_biz: of course it is possible to link static lavc/lavf against x264
[20:45:21 CEST] <JEEB> *x264cli
[20:46:01 CEST] <hero_biz> ty :)
[20:47:03 CEST] <aargh> hmm, obviously i was not aware of this.  i assumed conversion to joint would open up possibility of translation error
[20:47:23 CEST] <JEEB> anyways, you're re-encoding lossy mpeg-1 part 2
[20:47:30 CEST] <furq> it does in theory but it's not an issue with a good mp3 encoder
[20:47:32 CEST] <JEEB> which by itself makes no sense
[20:47:35 CEST] <furq> and lame is a very good mp3 encoder
[20:48:11 CEST] <furq> there's a reason why joint stereo is the default
[20:48:52 CEST] <aargh> yeah, i'm doing it for someone who has an old mp3 player that doen't seem to recognize mp2
[20:49:21 CEST] <aargh> i'll go with the joint then.  thanks
[20:49:59 CEST] <aargh> got to run....  thanks again for the help, furq and JEEB
[21:23:05 CEST] <massimodipierro> hello everybody. new here. I have a question. Has anybody used ffmpeg with Mobotix MxPEG files? Is there any documentation about ffmpeg supporting this format. The Mobotix site claims ffmpeg supports it but I did not see any proof on the ffmpeg side.
[21:23:39 CEST] <BtbN> Well, does it work?
[21:24:25 CEST] <massimodipierro> I have not tried it. First I would like to know if there is an official stand on it.
[21:24:40 CEST] <Anoia> when I looked at mxpeg, it was JPEG with a different chunking
[21:25:09 CEST] <massimodipierro> yes but they claim to compress in time too and not in space.
[21:25:33 CEST] <BtbN> So... like every normal video codec?
[21:25:52 CEST] <JEEB> lol time and space. just call it intra and inter coding :V
[21:26:17 CEST] <massimodipierro> Hey. I am a physicist. Always time and space.
[21:26:28 CEST] <JEEB> yes, but you know... video compression :P
[21:26:34 CEST] <massimodipierro> OK
[21:26:50 CEST] <JEEB> intra is within something, inter is between things
[21:27:09 CEST] <massimodipierro> So back to my question. Has anybody tried it? Is there any thing the docs that mentioned this file format?
[21:27:12 CEST] <JEEB> aka intra coding codes whatever is being coded within that thing, and inter coding can use other things to code what it is coding :P
[21:27:24 CEST] <JEEB> massimodipierro: pretty sure people don't give a damn about that proprietary format in general
[21:27:29 CEST] <JEEB> thus you will have to get a sample and test
[21:27:33 CEST] <JEEB> otherwise nothing can be said
[21:27:52 CEST] <massimodipierro> That's fair. that's what I expected.
[21:28:02 CEST] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=commit;h=9d09ebf1ed489b26d2bb09549016e114ef8d54b2
[21:28:07 CEST] <JEEB> there's a demuxer for their container
[21:28:13 CEST] <JEEB> from 2010
[21:28:35 CEST] <JEEB> but nothing is  found with "mobotix" as a search term otherwise from the git history
[21:29:40 CEST] <massimodipierro> That is a good sign. I will give it a try.
[21:29:52 CEST] <massimodipierro> At least there is hope. :-)
[21:29:57 CEST] <massimodipierro> Thank you
[21:39:28 CEST] <Rene_> hey guys
[21:39:58 CEST] <Rene_> rasample audio in ffmpeg in c
[21:40:37 CEST] <Rene_> does anyone know how to do?
[21:41:00 CEST] <JEEB> two libraries are available for that
[21:41:19 CEST] <JEEB> libavresample which is both in libav and ffmpeg, and libswresample which is only in ffmpeg
[21:41:27 CEST] <JEEB> both have very similar APIs
[21:41:40 CEST] <JEEB> I'm pretty sure there's an example in using one of them in the docs :P
[21:42:42 CEST] <furq> http://www.ffmpeg.org/doxygen/trunk/resampling__audio_8c_source.html
[21:42:47 CEST] <hero_biz> I want to compile ffmpeg with these options: --enable-openssl --enable-libfdk-aac --enable-libmp3lame --enable-libopus --enable-avisynth --enable-libx264 --enable-libvpx  --enable-libass  --enable-libfreetype  --enable-libvorbis  --enable-libtheora --enable-libxvid
[21:43:06 CEST] <hero_biz> any other options is recommanded for enabling?
[21:43:42 CEST] <JEEB> I bet you don't even use half of those
[21:43:59 CEST] <JEEB> just build shit with what you're actually using
[21:44:05 CEST] <hero_biz> I added some for their decoders
[21:44:18 CEST] <JEEB> like literally one I see
[21:44:36 CEST] <Rene_> I can send you my code? For already I tried to do and I am not knowing where to draw the function of ffmpeg to do the exchange format
[21:44:45 CEST] <JEEB> since lavc supports aac decoding, mp3 decoding, opus decoding, vp8/9 decoding, vorbis decoding, theora decoding, mpeg-4 part 2 decoding
[21:44:52 CEST] <JEEB> without any extra libs
[21:45:09 CEST] <JEEB> ok, I can't see a single one :P
[21:45:39 CEST] <Rene_> http://pastebin.com/qCHBhZdb
[21:45:46 CEST] <hero_biz> but it is said that vorbis support is primitive.that's why I added.
[21:45:52 CEST] <JEEB> *encoding*
[21:45:52 CEST] <furq> does it actually use any of those libs for decoding if they're available
[21:45:53 CEST] <hero_biz> I can remove easy :P
[21:46:06 CEST] <TD-Linux> yeah if you plan on encoding a lot of stuff then that's fine
[21:46:09 CEST] <JEEB> vorbis *encoder* in libavcodec is fucking atrocious
[21:46:33 CEST] <JEEB> probably somewhere worse than the libavcodec WMA encoder
[21:50:58 CEST] <hero_biz> jeeb: I was thinking if I should add --enable-libtesseract to support ocring subtitles...
[21:51:17 CEST] <JEEB> suit yourself
[21:51:33 CEST] <hero_biz> is it usefull?
[21:53:38 CEST] <Rene_> JEEB you looked?
[21:53:48 CEST] <yongyung> I want to decode a 44.1khz mp3 file to a 48khz wav file. Tried it with -r:a 48000 but the output wav is still 44.1khz. How do I do it? :/
[21:54:13 CEST] <JEEB> Rene_: no. I was going to write that if you have absolutely no idea, you should look into getting paid help for your application
[21:54:55 CEST] <yongyung> okay -ra worked lmao... all the other options were converted to q:a, b:a c:a format, but I guess not r:a
[22:42:28 CEST] <kagami_> Hi. Does anyone know why mpdecimate doesn't actually want to drop frames? I have sequence of 8 frames like this: A A A B B B C C and want to get A B C. I run ffmpeg like this "ffmpeg -i t.y4m -vf mpdecimate -y t2.y4m" and still get 8 frames in output. Though I can see "dup=5 drop=0" in ffmpeg's output.
[23:02:31 CEST] <durandal_1707> Perhaps because framerate is same ?
[23:04:22 CEST] <kagami_> durandal_1707: thanks, you are right, it works with "-r 8". But how can I know target framerate in general case?
[23:05:50 CEST] <pamra> hi, can anyone tell me how to generate HLS variant playlist file with ffmpeg?
[23:33:02 CEST] <onefix> I have a project I'm wanting to do, and I just wanted to make sure that I'm not reinventing the wheel before I jump into it
[23:42:20 CEST] <durandal_1707> onefix: what project?
[23:59:49 CEST] <onefix> My plan is to run FFmpeg on a small, cheap device (like the $9 CHIP) ... write a script to split a file
[00:00:00 CEST] --- Thu Oct 22 2015


More information about the Ffmpeg-devel-irc mailing list