[Ffmpeg-devel-irc] ffmpeg.log.20140131

[burek]

[00:00] <Fusl> can someone tell me if i somehow can set a timeout to the mmsh protocol?
[00:01] Action: Qantourisc is verry confused on the quality settings thing of pp
[00:01] <Fusl> i'm trying to grab a mmsh-stream and save it to a file but sometimes the mmsh-source freezes and then ffmpeg doesn't write anything anymore where ffmpeg should then just exit with a warning somehow like "Source timed out, exiting."
[00:01] <DopeLabs> -shortest
[00:02] <DopeLabs> ?
[00:03] <DopeLabs> basically as soon as the first input stream ends ffmpeg will stop
[00:03] <Qantourisc> "a autoq CPU power dependent enabler" ?
[00:06] <PlaNick> hello all
[00:06] <PlaNick> is there anybody attending the FOSDEM this week end ?
[00:08] <llogan> Fusl: some of the protocols support a -timeout option. not sure about mmsh but you can give it a try. seconds or ms i'm guessing. if not then: patch welcome.
[00:10] <Fusl> llogan: isn't there such option to say "if nothing has been converted since a minute, kill it"?
[00:10] <Fusl> because the protocols documentation http://www.ffmpeg.org/ffmpeg-protocols.html#mmsh doesn't say anything about timeout
[00:13] <Fusl> llogan: Option timeout not found.
[00:13] <DopeLabs> does -shortest output.whatever work?
[00:13] <Fusl> why would that work DopeLabs?
[00:14] <Fusl> i think -shortest just says ffmpeg to finish when the first input source has been finished reading
[00:14] <Fusl> which hasn't, because it just freezed
[00:14] <DopeLabs> 'freezed'
[00:15] <DopeLabs> can mean lots of things
[00:15] <DopeLabs> one would think
[00:15] <DopeLabs> if its not getting any data
[00:15] <DopeLabs> thats the end
[00:15] <DopeLabs> and disconnect
[00:15] <Fusl> with freezed i mean the server didn't send any data anymore
[00:15] <Fusl> but ffmpeg was still connected to the server
[00:15] <DopeLabs> right
[00:15] <DopeLabs> try -shortest
[00:15] <Fusl> k
[00:16] <Fusl> gonna test that
[00:16] <Fusl> not sure if the server again drops my connection
[00:16] <Fusl> could take up to 6 hours ^_^
[00:16] <DopeLabs> are you sure that you arent being sent any data when this 'freeze' happens
[00:17] <Fusl> not sure, nope
[00:17] <Fusl> i just saw that the timestamp didn't change anymore
[00:17] <Fusl> after trying to kill it with ctrl-c it also did not exit
[00:18] <Fusl> (first ctrl-c would be finish encoding and then exit)
[00:18] <Fusl> it only exited after the second ctrl-c
[00:18] <DopeLabs> i kinda has this same issue... using a shoutcast server for the audio source... sometimes during live shows djs dont have the best connections.. any kind of buffering on the shoutcast would cause ffmpeg ti stop... because i had the -shortest command in there... took it out... works fine now
[00:19] <Fusl> so taking -shortest out of the command worked for you? O.o
[00:19] <DopeLabs> yes because it WAS stopping
[00:19] <Fusl> ah
[00:19] <Fusl> and you didn't want it to stop?
[00:19] <DopeLabs> correct
[00:19] <Fusl> ah, okay
[00:19] <DopeLabs> im not writing to a file
[00:19] <DopeLabs> i mstreaming back out
[00:20] <DopeLabs> basically this
[00:20] <DopeLabs> http://www.dopelabs.com/shoutcast-to-ustream-tv-justin-tv-twitch-tv-youtube-wowza-flash-with-ffmpeg-on-ubuntu/
[00:25] <Fusl> DopeLabs: tbh, i'm also not sure the container where the ffmpeg process is running on didn't just run out of disk space because if had only 2GB free now
[00:25] <Fusl> s/if/i/
[00:25] <DopeLabs> well i guess it all depends on what your trying to write then i guess
[00:26] <DopeLabs> but generally.. 2gb isnt much
[00:26] <Fusl> yep, i know
[00:26] <Fusl> i raised the limit up now so it has a few tb's free again
[02:39] <sineb> best dvd authoring software (free) ?
[02:43] <llogan> sineb: dvdauthor is one
[02:45] <sineb> thanks
[08:23] <Keestu> where i can find a sample code to stream rtsp video from ffmpeg into SDL2 ?. it is very much hard to figure out. KIndly someone help me
[09:03] <brontosaurusrex> valid git is still : git clone http://git.videolan.org/git/ffmpeg.git < ?
[09:05] <ubitux> http://ffmpeg.org/download.html
[09:05] <ubitux> > git clone git://source.ffmpeg.org/ffmpeg.git ffmpeg
[09:06] <ubitux> (it's still pointing on videolan though)
[09:06] <brontosaurusrex> ubitux, ok, but that 2nd url is better?
[09:06] <brontosaurusrex> and source.ffmpeg.org has a http:// version?
[09:07] <ubitux> well, i suppose the official website is a trustable source for its repository
[09:07] <ubitux> i suppose, but git:// is better
[09:07] <brontosaurusrex> yeah, but i'am firewalled
[09:07] <brontosaurusrex> badly
[09:16] <Qantourisc> Any idea why dctdnoiz is missing from my ffmpeg -filters list (and doesn't work if i specify it)
[09:20] <ubitux> Qantourisc: you probably disabled it
[09:20] <ubitux> note that it's *really* slow
[09:21] <ubitux> like several minutes per frame sometimes :p
[09:27] <Qantourisc> ubitux: jikes :p well better then banding ?
[09:28] <Qantourisc> ubitux: how about v1.2.4 that could also cause missing no ?
[09:28] <brontosaurusrex> any recent examples on how to use fdk-aac for variable-bitrate (quality based) encoding?
[09:29] <brontosaurusrex> fdk-aac-0.1.3 is what i got, the tarball
[09:36] <Qantourisc>  http://trac.ffmpeg.org/wiki/AACEncodingGuide last modifed: 3 days ago
[09:37] <brontosaurusrex> Qantourisc, thanks
[09:38] <brontosaurusrex> my google-foo is low today ....
[09:38] <Qantourisc> brontosaurusrex: check history for what has been modified when ;p
[09:39] <ubitux> Qantourisc: 1.2.4 is really extremely old&
[09:40] <ubitux> Qantourisc: to remove banding you have gradfun
[09:40] <Qantourisc> ubitux: figures, for some reason it was in "very unstable don't try to install"
[09:40] <ubitux> and for denoising i suppose you have the old sharpen
[09:40] <Qantourisc> instealled 2.1.3 now though
[09:41] <ubitux> i recommend to play with the overlap to make the filter usable
[09:41] <Qantourisc> "the overlap" ?
[09:41] <ubitux> overlap option, in dctdnoiz
[09:41] <ubitux> i plan to make the filter a lot faster, but ENOTIME for now
[09:42] <ubitux> (patch welcome obviously)
[09:42] <Qantourisc> :)
[09:42] <Qantourisc> well was mainly wondering why a lot of filters where missing
[09:42] <Qantourisc> found that now :p
[09:43] <Qantourisc> suppose I coud use hqdn3d and gradfun
[09:44] <ubitux> yeah
[09:44] <ubitux> owdenoise as well
[09:44] <Qantourisc> that's also slow no ?
[09:44] <Qantourisc> ow depth
[09:45] <ubitux> nothing is as slow as dctdnoiz :)
[09:46] <Qantourisc> imo just chuck in fftw ?
[09:46] <Qantourisc> should be quite optimzed
[09:47] <ubitux> overkill
[09:47] <ubitux> and we have some asm all around, we just need to write some little asm for that filter
[09:51] <Qantourisc> This filter is designed for playback only.  Do not use it prior to lossy compression. because compression tends to lose the dither and bring back the bands.
[09:51] <Qantourisc> hmmmm
[10:03] <anshul> I have some memory leakage in avformat_find_stream_info I have already close the codec and fmt context, but enable to get what I have to do clean this memory
[10:07] <Qantourisc> "could not write header for output file #0 (icorrect paramters ?): Invalid Argument"
[10:10] <anshul> are you using ffmpeg command
[10:11] <Qantourisc> ffmpeg -i somefile.avi
[10:11] <Qantourisc> ow
[10:11] <Qantourisc> put the -tune film in the wrong pace :p
[10:43] <pvagner> guys I have issues figuring out a proper way on how to transcode the audio using libavcodec. I would like to be able to transcode mpeg2 audio to VORBIS and AAC. I have self-compiled ffmpeg 1.1 containing libavcodec 54. I am only setting up global_quality, channels, sample_fmt, sample_rate, channels and bitrate in the output codec context. The error I am receiving says "sample_fmt invalid or not supported by the codec". I am using the same samp
[10:43] <pvagner> SAMPLE_FMT_S16 is it predictable what I might be missing?
[10:44] <anshul> Why not use git, if you are using for presonel
[10:45] <JEEB> pvagner, you should check what sample formats the coder you're using supports
[10:45] <JEEB> and if needed, convert with one of the two libraries for that
[10:45] <anshul> you can start with looking trancode_aac example at doc/example/ but it is with git not in ffmpeg 1.1
[10:45] <JEEB> (libav/swresample)
[10:46] <anshul>  I have some memory leakage in avformat_find_stream_info I have already close the codec and fmt context, but enable to get what I have to do clean this memory
[10:47] <anshul> i have checked the same format with ffmpeg,so I think there is some step misssing that would release the memory
[10:47] <pvagner> JEEB: Can I see manually which sample_fmts a particular codec supports?
[10:48] <anshul> you can look inside libavocodec folder of ffmpeg vorbis and aac encoder and decoder file to see what they support
[10:48] <JEEB> pvagner, look at the identification part in libavcodec/***.c
[10:48] <JEEB> that has a list of sample_fmts
[10:49] <JEEB> of course it's possible to negotiate, too
[10:49] <JEEB> or well, get the list
[10:49] <JEEB> and then you can roll over that
[10:55] <pvagner> JEEB: AVCodec class has a SAMPLE_FMTS member. I think this is the array of supported sample formats. Most likelly I will try to choose most suitable sample format at runtime then. Can I assume when I am getting this error that it is really a sample format issue not an error with encoder configuration? e.g. trying to specify bitrate 128000 and quality 4 while encoding to vorbis?
[10:57] <JEEB> yes, generally that error comes from that. If you set multiple rate control modes I have no idea what happens, but you probably wouldn't get an error about sample formats :P
[10:57] <JEEB> either only one of the rate control modes get used, or you get another error
[11:03] <pvagner> JEEB: thank you verry much for the guidance. I will try to implement what you have suggested tonight and mostlikelly I will return back with additional questions then.
[12:24] <anshul> I am having memory leakage in read_frame_internal -> parse_packet -> av_dup_packet -> copy_packet_data
[12:43] <anshul> packet passed to decoder are to be freed after encoding has been done, otherwise memory leak is observed
[13:13] <FR^2> I want to "copy" all my dvds to my harddisk for easy access via xbmc. I'm sure there's some good how-to on this - any hints are welcome!
[13:14] <FR^2> Of course I could just use dd, but de-interlacing and encoding it to a more efficient format seems to make sense :)
[13:15] <FR^2> okay, okay, postpone that - the wiki seems quite good :D
[13:16] <theholyduck> FR^2, basic dvdripping with minimal concern for internet distrebution is pretty eays, in addition to the wiki, there should be plenty of other good sources aswell
[13:17] <FR^2> When simply searching on the internet I ran across a lot of out-dated or simply wrong howtos :)
[13:17] <theholyduck> well that will also happen
[13:18] <theholyduck> FR^2, if you want a little bit more simplified process, handbrake is basicly just a decent wrapper for ffmpeg/x264 ++
[13:18] <theholyduck> and i THINK it supports decrypting and sourcing from dvds without any fuss
[13:36] <FR^2> Hmmhmm. handbrake. I will have a look, thank you!
[13:38] <DeadSix27> FR^2: yep, handbrake is what i always used to rip BD's and DVD's
[13:39] <FR^2> DeadSix27: :)
[13:40] <DeadSix27> later due to other reasons i switched to meGUI and or python based batch scripts and the raw encoding binarys though
[13:40] <DeadSix27> had too much trouble with attachments/subtitle languages/script handling/timing
[13:41] <DeadSix27> but without ffmpeg all that would be taking annoyingly long
[13:42] <DeadSix27> FR^2: Also, i'm not sure, but when i last used Handbrake it had trouble handling PGS Subtitles, so be sure that works in newer versions as intended.
[13:43] <DeadSix27> (Been a year or so, i doubt it still gives trouble)
[13:43] <FR^2> subtitles are uninteresting for me.
[13:43] <FR^2> When trying out encoding DVDs I always ended up with a/v out of sync
[13:43] <DeadSix27> well then, handbrake is a easy to use GUI for little exprienced users
[13:43] <DeadSix27> it mainly gives good descriptions about its options.
[13:45] <FR^2> thanks
[14:44] <DeadSix27> Can someone tell me what this means, and how i'm able to fix this issue? "[abuffer @ 042989e0] Value inf for parameter 'time_base' out of range [0 - 2.14748e+009][abuffer @ 042989e0] Unable to parse option value "(null)" as sample format"
[14:44] <DeadSix27> > http://pastebin.com/JT6833Gg
[15:30] <anshul> how to seek file to start when file is finished in liavcodec libavfilter
[15:32] <anshul> DeadSix27, isnt that errror message self explanatory, are you using ffmpeg cmd or library
[15:32] <DeadSix27> ffmpeg cmd
[15:33] <anshul> if you
[15:33] <anshul> unspecified sample format
[15:33] <anshul> Consider increasing the value for the 'analyzeduration' and 'probesize' options
[15:34] <anshul> I think your stream is somekind of special one
[15:35] <anshul> cany you try some cmd like ffmpeg -i title.stream some.mp2
[15:35] <anshul> mp2 is most of the time stable then other codecs
[15:38] <DeadSix27> anshul: http://pastie.org/private/haqypaezalo2ug1ipg8cyg
[15:44] <anshul> if you have noticed Stream #0:1[0x80]: Audio: ac3, 0 channels
[15:44] <anshul> is that title.stream is playable
[15:44] <anshul> ffplay title.stream
[15:44] <DeadSix27> yep, atleast using ext. lav filters in mpc, and vlc
[15:45] <DeadSix27> but ffplay only plays the video
[15:45] <DeadSix27> and gives the same error
[15:45] <DeadSix27> as ffmpeg
[15:45] <DeadSix27> (well the yellow warning)
[15:45] <DeadSix27> i mean -> [mpeg @ 03d534c0]
[15:45] <DeadSix27> (as pastie has no colors)
[15:48] <anshul> There might be some similar bugs in trak ffmpeg, you can help the developer working on similar issue,
[15:49] <anshul> may be trying some option of ffprobe can help you analyse your stream
[15:49] <anshul> bye
[16:19] <Axel_> Hi all. Could someone tell me how can I record a video using ffmpeg?
[20:10] <hayuto> [A
[20:10] <hayuto> anyone have idea what ffmpeg options are best for hls streaming?
[20:10] <hayuto> hi
[20:25] <fragamus> hi does anyone know the Y411 format well enough to tell me about the macroblock? Is it long and flat? is it square? What is the geometric location within the macroblock of each of the six samples?
[21:12] <llogan> hayuto: http://ffmpeg.org/ffmpeg-formats.html#hls
[21:43] <odie33> does anyone have experience with cross compiling libvpx under linux for win32? i'm running into linking/compilation errors while building my ffmpeg libraries and couldn't find help anywhere else
[22:06] <SenseiV183> Do I need to ffmpeg-10bit for transcoding AVCHD camcorder and GoPro Hero 2 formats.  Does ffmpeg-10bit support all the same command line options as does ffmpeg?  I couldn't find any detais from google.
[22:09] <SenseiV183> ffmpeg version 2.1.3-   http://johnvansickle.com/ffmpeg/    Copyright (c) 2000-2013 the FFmpeg developers
[22:09] <SenseiV183>   built on Jan 22 2014 19:50:14 with gcc 4.8 (Debian 4.8.2-14)
[22:11] <SenseiV183> ffmpeg-10bit -h shows less options than ffmpeg -h
[22:13] <hayuto> llogan: i was thinking about encoding options like for example -tune zerodelay
[22:13] <hayuto> llogan: i know how to make m3u8 :P
[22:13] <hayuto> llogan: i dont know what codecs and options to use so it would be best to play on android
[22:15] <relaxed> SenseiV183: do you need 10 bit h264 video?
[22:17] <SenseiV183> relaxed, I want to make an intermediate format and proxy format.  Intermediate format shall be of highest quality to save in case need that footage for Cinema or Broadcast.
[22:17] <SenseiV183> (Pre video editing work)
[22:18] <SenseiV183> Mostly I will be working with this type of format:  Stream #0:0(eng): Video: h264 (Main) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 30049 kb/s, 59.94 fps, 59.94 tbr, 90k tbn, 119.88 tbc (default)
[22:19] <relaxed> SenseiV183: That is my build. ffmpeg-10bit is exactly the same, it's just compiled against libx264 with 10-bit output support.
[22:19] <SenseiV183> Ok
[22:19] <SenseiV183> I get an error.  Hold on.
[22:20] <relaxed> pastebin.com
[22:24] <SenseiV183> relaxed, http://pastebin.com/Xpmk6zZq
[22:25] <SenseiV183> My concern is the incompatible pixel format yuv420p and h264
[22:25] <pvagner> guys I am trying to transcode mpeg2 audio stream into aac using libavcodec. the source stream is 48000Hz s16p stereo. libfaac encoder does not handle s16p sample format. it only supports s16. Do I need to resample or how can I make it work?
[22:29] <relaxed> SenseiV183: Your source is yuv420p, which is 8-bit. 10-bit is needed if your source is 10-bit, or if you want better compression. However, most software/hardware expects 8-bit yuv420p.
[22:30] <relaxed> http://x264.nl/x264/10bit_02-ateme-why_does_10bit_save_bandwidth.pdf
[22:31] <relaxed> AVCHD is 8-bit h.264.
[22:32] <relaxed> SenseiV183: h.264/yuv420p is pretty much the standard.
[22:33] <SenseiV183> relaxed, Thank you, I tried without -10bit and I didn't get the yuv420p error :)
[22:34] <relaxed> SenseiV183: right, it only supports yuv420p10le yuv422p10le yuv444p10le nv20le
[22:36] <SenseiV183> relaxed, Where Can I read about which formats are best for intermediate and proxy files?  I want to use Blender and possibly lightworks and cinelerra to do editing (If I can ever get cinelerra to compile on saucy salamander)
[22:37] <relaxed> libx264 provides a lossless mode, but first see if you can import the video directly. Google: ffmpeg libx264 (other software project)
[22:38] <relaxed> ^^ should give you an idea
[22:38] <SenseiV183> I've had unexpected results with backwards video in blender and suspect maybe I need all I-frames video?
[22:39] <relaxed> maybe someone in #blender can give you an answer.
[22:41] <SenseiV183> True.  I am here to learn more about FFmpeg as well.
[22:45] <relaxed> I'm not telling you to leave. Only that other project specific requirements would be better off asked in their channel :)
[22:46] <SenseiV183> I think for making intermediate files I want to use mpeg2 I frame format.  I'm up for all recommendations on where else to find answers for my questions.  Thank you relaxed.
[22:48] <relaxed> Each time you encode you're losing quality. mpeg2 is old and is not the answer.
[22:48] <relaxed> Unless you use a lossless codec.
[22:49] <SenseiV183> I'm thinking a format that has all the original quality like .pngs wrapped up in video.  Where there is no decompression to playback
[22:50] <SenseiV183> like what gopro studio does but without the cineform codec.  Is there a way to use cineform with ffmpeg on linux?
[22:51] <SenseiV183> sorry asking to many questions at once.
[22:52] <relaxed> ask the blender people which container and lossless codecs you can use for both, audio and video, for input.
[22:54] <relaxed> Then I can recommend something. Until then were just stabbing randomly in the dark.
[22:54] <relaxed> we're*
[22:54] <SenseiV183> I really want to ask a codec expert which codecs are best for each application I have and do blender experiments myself.
[22:54] <SenseiV183> I can do the same experiments on lightworks and cinelerra
[22:55] <SenseiV183> to help choose another video editor to use along side blender.
[22:55] <SenseiV183> on a project
[22:55] <SenseiV183> still picking my tools and developing my workflow  and learning.
[22:56] <SenseiV183> No one talked on #cinelerra the other day for hours straight there where like 8 people in the channel.
[22:57] <SenseiV183> But it looks like a very capable editor handling huge video formats like I think 2.5k or 4k maybe unlimitied.
[23:00] <SenseiV183> I wish I have found the magic video codec bible book with application specific examples geared towards the video editor that needs access to all platforms, cinema, broadcast, bluray, dvd HTML5
[23:11] <SenseiV183> Any recommendations on a linux distro that supports the latest FFmpeg out of the box compiled with all codec support?  I'm using ubuntu studio and ffmpeg is deprecated and saucy salamander uses avconv instead of ffmpeg.
[23:19] <ubitux> SenseiV183: any distro not debian based
[23:23] <SenseiV183> ubitux, Really?  wow.  I only picked debian because it is famous for stability and runs webservers that stay up.  But for my needs maybe it's a terrible pick.
[23:23] <ubitux> you still have that for ubuntu i suppose
[23:23] <ubitux> deb-multimedia repo for debian
[23:24] <ubitux> ...and you can also complain to your distro
[23:24] <SenseiV183> :)
[23:25] <SenseiV183> I'm using the johnvansickle static :)  Do I need to set environment varibles for blender to know to use the new one?  I don't know how to know what it's using...
[23:26] <ubitux> no idea
[23:26] <ubitux> blender is probably patched to use the fork in your distro
[23:26] <SenseiV183> Or maybe I have more vested intrist in using a different distro all together.  One that compiles software well.
[23:26] <SenseiV183> And low latency
[23:27] <SenseiV183> I run blender pre-built - latest version
[23:27] <ubitux> i don't know if blender is using the tool and/or the libraries
[23:28] <SenseiV183> I think I'm missing all the advanages of compiling all programs myself because of the learning curve.
[23:28] <ubitux> gtg, gl..
[23:29] <SenseiV183> ?
[23:31] <SenseiV183> ubitux, [gtg, gl..]?
[23:33] <ss_haze> http://serverfault.com/questions/571410/cant-load-localhost-when-connected-to-internet-using-apache-ubuntu-12-04
[23:39] <SenseiV183> Ah.. much later I think I figured out ubitux meant, got to go, good luck ....
[00:00] --- Sat Feb  1 2014