[Ffmpeg-devel-irc] ffmpeg.log.20140805

[burek]

[00:13] <TD-Linux> GPUs don't magically make your code faster.
[00:14] <TD-Linux> many parts of video encoding are highly serial which works poorly on a GPU.
[01:31] <voip1__> TD-Linux, i tink for parallel transcoding (for example 50 streams) it will be fine.
[01:32] <TD-Linux> I highly doubt it. More CPU cores would be far more cost effective.
[01:36] <voip1__> You realy tinking 1 intel xeon with 8 cores able transcode more streams than 1 Radeon card with 2816 Shading Units?
[01:36] <TD-Linux> yes.
[01:37] <voip1__> interesting
[01:38] <TD-Linux> if you want to fit 2816 "cores" on a chip the same size as an 8 core xeon, obviously those cores are going to have compromises
[01:38] <TD-Linux> one is that usually groups of cores need to execute in lockstep
[01:39] <TD-Linux> in amd case it's 32 threads bundled
[01:39] <voip1__> as you know for example adobe premiere can use GPU during encoding, its works fine
[01:40] <TD-Linux> yes. but it only partially uses the GPU and produces encodes of worse quality than x264.
[01:41] <llogan> and not any faster
[01:42] <llogan> i prefer to use a frameserver instead of AME https://trac.ffmpeg.org/wiki/Encode/PremierePro
[01:42] <TD-Linux> for video encodes I use amazon c3.8xlarge instances, with 16 physical xeon cores (32 threads). It is by far the most cost effective solution.
[01:43] <TD-Linux> well, maybe not "by far". but useful for my use case.
[01:45] <voip1__> i dont know
[01:45] <TD-Linux> voip1__, write some code for a GPU and then come back with your own opinion :)
[01:46] <voip1__> guys, im not developer :)
[04:45] <MarcelvanLeeuwen> Morning!
[04:46] <MarcelvanLeeuwen> Very early morning&
[04:50] <MarcelvanLeeuwen> I have a little script which is encoding movies placed in a directory. Now would it be nice if movies with two audio track also be correct encoded..
[04:51] <MarcelvanLeeuwen> the script can only encode movies with one audio track
[04:51] <MarcelvanLeeuwen> is it possible to adjust this?
[06:25] <rjp421> when playing a file and streaming to flv over rtmp, the play time clock goes fast and skips ahead, while the video slowly speeds up until it crashes
[06:37] <rjp421> floods with then ends at *** dropping frame 22208 from stream 0 at ts 22206
[06:37] <rjp421> No more output streams to write to, finishing.
[08:51] <termos> is there some way to set an exact gop in ffmpeg when doing x264 encoding?
[08:56] <termos> or at least set it so that it's not varying by a lot
[08:57] <termos> the problem is that I will sometimes get very small hls segments when doing hls muxing
[10:24] <zenderz> anyone familiar with the following error:
[10:24] <zenderz> [ismv @ 0x9ba7d60] Malformed AAC bitstream detected: use audio bitstream filter 'aac_adtstoasc' to fix it ('-bsf:a aac_adtstoasc' option with ffmpeg) av_interleaved_write_frame(): Operation not permitted
[10:24] <ubitux> maybe you need to do what the error message suggests?
[10:25] <zenderz> i did and it works. however i want to know whats wrong with the input file and how i can fix it before hand
[10:25] <zenderz> http://pastebin.com/BhFMTm5h
[10:26] <zenderz> is there a way I can apply the bitstream filter to the original file
[10:27] <ubitux> i don't think it's problematic to have ADTS in mpeg
[10:28] <zenderz> the thing is that i want the input file to have exactly the correct codecs so that no transcoding/filtering is needed. this will make the smoothstreaming conversion process easier
[10:29] <zenderz> so can i do something like: ffmpeg -i testinput codec:a copy -bsf:a aac_adtstoasc -codec:v copy inputfixed.mpg
[10:33] <roee> Hi Guys, I'm trying to capture raw data from my webcam and analyze it using node.js so i created the following code just for debuging and understanding the content of the data generated by ffmpeg
[10:33] <roee> http://pastebin.com/mUVgBBPf
[10:34] <roee> I than opened the test file i created and saw that ffmpeg creates some kind of header and only after that the raw data
[10:34] <roee> how can i find out what
[10:34] <roee> what's the size of that header
[10:35] <roee> *
[10:35] <Mavrik> hmm
[10:35] <Mavrik> MPEG-TS doesn't have a header per-se
[10:35] <Mavrik> but it does inject PAT/PMT packets which describe content of streams
[10:35] <Mavrik> which is what you're probably seeing now
[10:35] <roee> i took mpeg-ts from a sample
[10:35] <roee> could you suggest something better?
[10:35] <Mavrik> well... depends on what do you want to achieve really
[10:35] <roee> i want to detect the dominate color in each frame
[10:36] <roee> a simple image analysis
[10:36] <Mavrik> hmm
[10:36] <Mavrik> thing about video is that you really need at least some kind of container around it
[10:36] <Mavrik> lemme see if ffmpeg can do some kind of delimited raw output
[10:36] <roee> thanks
[10:38] <zenderz> so if i try fix the input file i get the following:
[10:39] <zenderz> http://pastebin.com/EHD9rBqF
[10:39] <zenderz> [mpegts @ 0x98ee640] AAC bitstream not in ADTS format and extradata missing
[10:39] <zenderz> which is strange, because when using the same filter for smoothstreaming format, it works fine
[10:39] <Mavrik> roee, -f rawvideo should give you raw video output
[10:40] <roee> ok i'll try that and let you know
[10:40] <roee> [NULL @ 0x9ebd5e0] Requested output format 'rawvideo' is not a suitable output format
[10:41] <roee> I replace the mpegts with rawvideo
[10:41] <roee> replaced*
[10:42] <roee> do i need to rebuild ffmpeg with this capability?
[10:42] <Mavrik> hmm, it seems that rawvideo muxer still hasn't bene implemented.
[10:42] <zenderz> roee: have a look at http://zulko.github.io/blog/2013/09/27/read-and-write-video-frames-in-python-using-ffmpeg/
[10:42] <zenderz> its for python but its similar to what you need
[10:42] <Mavrik> mhm
[10:43] <Mavrik> it seems that you have to leave out -f and explicitly say codec:v rawvideo
[10:43] <Mavrik> never did that tho
[10:43] <Mavrik> zenderz, well you ARE explicitly converting ADTS audio to ASC audio :)
[10:43] <roee> ok
[10:43] <roee> i'll try
[10:45] <zenderz> Mavrik, are they not just different containers for aac?
[10:46] <Mavrik> yes
[10:46] <Mavrik> they're a way of coding AAC
[10:46] <Mavrik> and MPEG2-TS standard DEMANDS that AAC is encapsulated in ADTS packets inside
[10:46] <Mavrik> hence muxer aborts encoding of invalid format
[10:47] <zenderz> ah okay I see. so there is no way to trans-mux from mpegts to smooth streaming format without any codec changes then?
[10:57] <termos> I am now setting AVFrame->pict_type = AV_PICTURE_TYPE_NONE before calling avcodec_encode_video2 and I get perfect stable GOP size. Is there a less hacky way to do this?
[10:59] <roee> Unable to find a suitable output format for 'pipe:' pipe:: Invalid argument
[11:00] <roee> cp.spawn("ffmpeg", [             "-f", "video4linux2",             "-s", "176x144",             "-r", "15",             "-input_format","yuyv422",             "-i", "/dev/video0",             "-vsync", "drop",             "-vcodec", "rawvideo",             "-movflags", "faststart+frag_keyframe",             "-"
[11:00] <roee> am i missing something?
[12:24] <zenderz> look at https://www.ffmpeg.org/ffmpeg-protocols.html#pipe
[12:25] <zenderz> roee
[12:50] <rjp421> when playing a file and streaming to flv over rtmp, the play time clock goes fast and skips ahead, while the video slowly speeds up until it crashes
[13:08] <Nickwiz> Is it possible to use fifo for audio input?
[13:10] <Nickwiz> ffmpeg ... -i /tmp/fifo_audio ...
[13:12] <Nickwiz> Reason is that my soundcard does not support capture, but I can use pacat. Then, if possible, I could pipe audio to fifo file from pacat and use ffmpeg -i /fifofile
[13:13] <Nickwiz> The idea is to record screen with audio.
[13:13] <Nickwiz> in one go.
[14:09] <Hello71> !tias
[14:17] <DopeLabs> =]
[14:20] <klaxa> Nickwiz: you are using pulseaudio then, right?
[14:21] <klaxa> have you tried ffmpeg -f alsa -i pulse ?
[14:21] <Nickwiz> klaxa: I have various "modes" I operate in, depending on what I do. Pure ALSA, Pulse or JACK.
[14:21] <Nickwiz> yes. I have tried that one. Only silence.
[14:22] <klaxa> have you tried using pavucontrol to change the stream that is being recorded?
[14:22] <klaxa> if you want to record your internal audio change the recording stream to something like "Monitor of Analog Stereo" or something
[14:24] <Nickwiz> The recording does not have any entries. I thought that was because my soundcard does not have any audio record support.
[14:24] <Nickwiz> It si typically only mic and line in.
[14:25] <klaxa> did you enable showing all streams?
[14:25] <Nickwiz> Yes. No diff.
[14:25] <klaxa> there is a drop-down switcher at the bottom where you can set it to "All input streams"
[14:25] <klaxa> *devices even
[14:26] <klaxa> that's pretty weird, how can pacat do it then
[14:28] <Nickwiz> Not sure. Thought is hooked directly into the sink. But, yes I find it weird it is not in pavucontrol. Only recall I messed with this earlier. Then I found that the soundcard did nto have any "record" option. Do not recall waht exactly it is called. Ended up with using pacet.
[14:28] <Nickwiz> *pacat
[14:28] <Nickwiz> That is; I can record mic and line of course, but not e.g. streaming audio.
[14:29] <Nickwiz> Some limitations on soundcard HW.
[14:29] <klaxa> most stuff in pulse is software though
[14:29] <klaxa> i don't think using pacat should be necessary
[14:30] <klaxa> you can use named pipes though if you really need to
[14:30] <klaxa> you will have to set the input format as an ffmpeg option though
[14:32] <klaxa> like: ffmpeg -c:a pcm16le -ar 44100 <whatever the command for two channels is> -i <named pipe> ...
[14:37] <Nickwiz> klaxa: Thanks, that's great. I'll look at it. I ended up recording audio and video separately in two terminals and it is now encoding (pos processing). I have to do a compile job after that, but I'll look at it later. Once again. Thank you.
[14:37] <Nickwiz> s/pos/post
[14:38] <Nickwiz> I also need to upgrade my HW, :/
[14:53] <jhon> Hello i'm trying to convert video to x264, i'm getting this error Unkown encoder 'libfaac'. Where I can find libfaac encoder i'm on windows
[15:02] <JEEBsv_> you compile a binary yourself with fdk-aac instead; both faac and fdk-aac are "nonfree" (incompatible with GPL, binary distribution not possible), and fdk-aac is much better
[15:02] <sacarasc> jhon: To get libfaac, you'd have to compile ffmpeg yourself as it is a non-free codec. There are a few other AAC encoders you can use, though. The one just called 'aac' is the better quality one allowed for a distributable ffmpeg.
[15:02] <sacarasc> To use aac, though, you have to also use -strict experimental in your command line.
[15:03] <sacarasc> I wonder how much of what I said got through there. :(
[15:05] <jhon> sacarasc: any guide for compiling it. I'm new to this
[17:33] <dreamcat4> hello! is this patch needed for 55.52.102 ???
[17:33] <dreamcat4> https://github.com/dreamcat4/tvheadend/commit/d18bc8dc65eea6f7a2e094379a6e4a193b6da258
[17:33] <c_14> That's a patch for libav. Ask #libav for information.
[17:34] <JEEB> seems general lavc, and that's just an update for a newer API
[17:34] <dreamcat4> c_14 even the libav provided by ffmpeg ?
[17:34] <JEEB> and no idea about the random library version :P
[17:35] <JEEB> if that version number is from FFmpeg then it's different than Libav, and it should be rather obvious which defines you have
[17:35] <JEEB> herp derp
[17:35] <dreamcat4> JEEB i guess i should dig the code
[17:36] <dreamcat4> if says 'ffmpeg 2.2.4'
[17:36] <JEEB> dreamcat4, then it has a relatively new API
[17:36] <sacarasc> dreamcat4: There's libav then there's libav*, so there can be some confusion. :D
[17:36] <dreamcat4> looks like it's been downloaded from http://ffmpeg.org/releases/
[17:37] <dreamcat4> (the FreeBSD port Makefile)
[17:37] <dreamcat4> i can look for those defines in the tarball
[17:37] <JEEB> I'm pretty sure it uses those newer namings there
[17:38] <JEEB> and if that project is made with libav only in mind, then it just doesn't have a libav vs ffmpeg configure check
[17:38] <JEEB> the version numbers are not the same in libraries, and so forth
[17:39] <t4nk001> Does AVCHD format work with ffmpeg ?
[17:39] <dreamcat4> JEEB: yep. in 2.2.3 tarball it use the newer AV_ defines. So that's needs the ".56" patch in URL above then
[17:40] <dreamcat4> i guess the configure check should exclude versions =< 55.0.0 too, (since they seem to break each other)
[17:41] <JEEB> dreamcat4, no
[17:41] <JEEB> libav and ffmpeg have different points
[17:41] <JEEB> the version numbers are not the same
[17:42] <JEEB> if you just lower the number it will enable one to build with an older libav
[17:42] <JEEB> which doesn't have it
[17:42] <JEEB> thus you have to add a check for libav or ffmpeg
[17:42] <JEEB> and then have separate version checks for both :P
[17:47] <dreamcat4> JEEBus, that's potentially version confusing!
[17:48] <dreamcat4> grateful to be made aware of that
[17:48] <JEEB> for an example of how to check you can check ffms2's configuration
[17:49] <JEEB> https://github.com/FFMS/ffms2/blob/master/configure.ac#L113 and forward
[17:51] <dreamcat4> thanks JEEB. if only the project used autotools properly and actually had a configure.ac file.
[17:52] <t4nk001> Does AVCHD format work with ffmpeg ?
[17:52] <JEEB> t4nk001, in general yes
[17:52] <dreamcat4> i think maybe pkg_config can output if ffmpeg is installed (*maybe*)
[17:52] <JEEB> hahaha
[17:53] <JEEB> just look at the code I linked :P
[17:53] <JEEB> it does use pkg-config first that the libraries are there
[17:53] <JEEB> then it checks a few things
[17:54] <dreamcat4> ah ok JEEB - thanks i guess at least some of it can be usef
[17:56] <t4nk001> Thanks JEEB , It is not working with FFmpeg version 0.6.5, Do not know the problem with specific version
[17:56] <JEEB> welcome to ancient FFmpeg versions
[17:56] <JEEB> in other word, update
[17:57] <t4nk001> :) Can we update it with yum or manually ?
[17:58] <JEEB> your distro most probably doesn't give you newer versions. for quick testing you can try grabbing a static build from the places linked under FFmpeg's downloads section
[17:59] <JEEB> and if those just don't work and/or you want to update, then I recommend you compile it yourself
[18:00] <dreamcat4> just shitting myself now. trying to figure out how i'm gonna explain all that to the other people on my project
[18:01] <JEEB> "There are two projects, their version numbers are not compatible. We have to check for which we are dealing with first, and then check accordingly"
[18:01] <JEEB> herp derp
[18:07] <dreamcat4> JEEB: can't i just parse the output of 'ffmpeg -version' instead of using such autoconf macros ?
[18:08] <dreamcat4> it would be easier for us (since not using autoconf)
[18:08] <sacarasc> What if they installed the libraries only, and not the executable?
[18:08] <JEEB> uhh, some of those checks are rather simple :P
[18:08] <JEEB> sacarasc, not to mention that it could be a random ffmpeg binary in your PATH
[18:08] <JEEB> not even a system packaged one
[18:10] <dreamcat4> JEEB: thanks for trying to help (its another thing to know which version the source code in the project expects)
[18:11] <JEEB> the "100" version check is the general way of determining libav vs ffmpeg btw :P
[18:11] <JEEB> which was in the ffms2 configure stuff, too
[18:12] <dreamcat4> ah right... that 100 check is a C MACRO not an autoconf macro... i can use that
[18:12] <JEEB> a lot of what I linked you to is usable by you :P
[18:12] <JEEB> a _lot_
[18:13] <JEEB> i you know what it does in general
[18:14] <dreamcat4> JEEB: sure thing. i just know nothing about ffmpeg. finding it quite the stretch for simpletons like myself
[18:22] <dreamcat4> presumably if CODEC_ID_G2M doesn't exist then compilation will fail, meaning it's libav, and not ffmpeg
[18:23] <JEEB> that's a more specific check
[18:25] <dreamcat4> JEEB maybe it's not needed anymore, if everyone is on 55 by now
[18:25] <dreamcat4> doesn't hurt though
[18:28] <dreamcat4> JEEB: https://gist.github.com/dreamcat4/42013f8fe6e637aa856a
[19:48] <sirnfs> Hi guys.  I have done a simple transcoding of an mp4 video using ffmpeg -i input.mp4 output.mp4, and I have noticed that in certain players (VLC, HTML5 player), the last couple of frames are not shown and the audio cuts off.  Playing the video in Quicktime or Windows Media Player produces the correct output.  I am using libx264 with the ffmpeg command line.  Any ideas?
[19:49] <sirnfs> The video can be seen here:  https://www.dropbox.com/s/qkuh2w7f9y25p0l/jesse_outTRANSNOSEEK.mp4
[19:55] <bazurbat> Hello, have anybody tried to use afade filter programmatically?
[19:56] <brontosaurusrex> sirnfs, what version of vlc is that, what html5 player?
[19:57] <sirnfs> brontosaurusrex: 2.1.5 Rincewind.  Not sure about the html5 player
[19:57] <sirnfs> I'm using the latest version of google chrome
[20:00] <sirnfs> brontosaurusrex: here is the original video file:  https://www.dropbox.com/s/yo1v2jxk639hffj/jesse_in.mp4
[20:01] <ramu> hi, trying to stream continous set of images encoded from ffmpeg (command line)via ffserver (command line). THe video is received in chrome
[20:02] <ramu> everything with in syste,m, but video is jerky
[20:02] <brontosaurusrex> sirnfs, can't spot a problem, but here are some files with burned in timecode http://sourceforge.net/projects/ffdrop/files/video/syncTestBunny/24and25.7z/download
[20:02] <brontosaurusrex> if you want to do some easily spotable tests
[20:02] <ramu> using vp8 encoding
[20:03] <sirnfs> brontosaurusrex: at the very end, when he says "next LEVEL"
[20:03] <brontosaurusrex> 24fpsSyncTester.mp4 < is what you want sirnfs
[20:03] <brontosaurusrex> sirnfs, yeah works fine here in vlc and in chromium
[20:03] <sirnfs> You compared it with the original?
[20:04] <brontosaurusrex> sirnfs, nope
[20:08] <brontosaurusrex> sirnfs, they sounds the same to me, out point is edited out a bit too soon
[20:08] <brontosaurusrex> i'd give it 5 frames more at the end
[20:08] <bazurbat_> Hello again, does somebody know how to make filtering_audio example work with afade filter?
[20:08] <sirnfs> brontosaurusrex: yeh, it seems to cut off at the end
[20:08] <sirnfs> brontosaurusrex: would you do that with the -ss command or ?
[20:09] <brontosaurusrex> sirnfs, yeah, but the original is the same
[20:09] <sirnfs> brontosaurusrex: for me, his lips seem to end in a different position
[20:09] <brontosaurusrex> sirnfs, i'd do it it in the video editing app
[20:10] <sirnfs> brontosaurusrex: I normally would, but these videos come from users.  Need something more automated
[20:13] <brontosaurusrex> sirnfs, i'd test various x264 switches first on a good video
[20:13] <brontosaurusrex> (the link i gave you)
[20:14] <brontosaurusrex> so that # of frames and playback behaviour is constant
[20:14] <brontosaurusrex> on all platforms
[20:14] <brontosaurusrex> else if that fails, try some other (older) x264 version
[20:15] <brontosaurusrex> until everything works
[20:15] <sirnfs> brontosaurusrex: ok, thanks for the feedback.
[20:15] <brontosaurusrex> another idea might be too add some black at the end of videos (but really you shouldn't)
[20:17] <sirnfs> I have an unrelated question: Is it possible to request certain chunks of a file from a users hard drive through some type of upload script that runs on a server?
[20:17] <sirnfs> Seems that the sandbox will prevent that, but maybe there is a workaround
[20:18] <brontosaurusrex> sirnfs, possibly with something like node.js
[20:19] <sirnfs> Thanks
[20:23] <dreamcat4> how can i find when the #defines became prefixed with AV_* in libavcodec ?
[20:24] <dreamcat4> i know they are AV_ in 2.2.4, but how far back. maybe look in git history on src code git repo ???
[20:25] <dreamcat4> or the change was pre-github era ? i should instead download random old tarballs and grep them ?
[20:26] <c_14> Well, the git history goes back to 2001, so it should be in there somewhere.
[20:27] <c_14> Assuming the change was properly documented.
[20:29] <ghospich> How to pass filename with spaces to subtitles filter? I can do `ffmpeg -i video.avi -vf subtitles=subtitle.srt out.avi`, but how to do the same, except with spaces in filename?
[20:29] <llogan> dreamcat4: APIchanges may mention it, but I didn't look
[20:31] <sacarasc> ghospich: Put "" around the file names.
[20:31] <c_14> dreamcat4: You can also try running a git blame over the lines and seeing what commit added them.
[20:32] <llogan> he's gone, jim.
[20:32] <llogan> i mean dead
[20:32] <ghospich> c_14, not working for me
[20:33] <ghospich> Though if i pass same string to ls, it's confirming that file exists
[20:34] <ghospich> okay
[20:35] <llogan> dreamcat4: what you missed: "APIchanges may mention it, but I didn't look"
[20:35] <llogan> <c_14>  dreamcat4: You can also try running a git blame over the lines and seeing what commit added them.
[20:41] <ghospich> Ohh, i found what's wrong. I passed -map 0:a:1 and -an to ffmpeg.
[20:41] <ghospich> Silly me.
[20:48] <brontosaurusrex> what machine was used with x264 threads=60 ?
[21:06] <dreamcat4> thanks llogan !
[21:14] <Carlos-Riper> hola
[21:14] <Carlos-Riper> me pueden ayudar porfavor :D
[21:17] <Carlos-Riper> I need to mount a streaming with ffserver and ffmpeg, webcam and my web browser, please.
[21:20] <llogan> Carlos-Riper: i'm not familiar with ffserver. this may help: https://trac.ffmpeg.org/wiki/Streaming%20media%20with%20ffserver
[22:24] <FrEaKmAn_> hi all.. I'm trying to understand iframes more
[22:25] <FrEaKmAn_> so to be able to have a better seek (video is played with OSMF), I should put an iframe every 1 second
[22:25] <FrEaKmAn_> but based on https://lists.ffmpeg.org/pipermail/ffmpeg-user/2011-September/002349.html
[22:25] <FrEaKmAn_> I'm not pretty sure if I must define each x frame or each second?
[22:41] <brontosaurusrex> FrEaKmAn_, also decoding complexity
[22:41] <FrEaKmAn_> I followe this tutorial https://support.octoshape.com/entries/25126002-Encoding-best-practices-using-ffmpeg
[22:42] <FrEaKmAn_> just instead of 48 I used 24
[22:42] <FrEaKmAn_> so every second
[22:42] <FrEaKmAn_> seems to work
[22:43] <Mavrik> oh god, why is he trying to make people use static keyframe ordering
[22:44] <brontosaurusrex> to get reliable seeking?
[22:45] <Mavrik> that makes no sense - setting GOP size will get you 1 sec (or less) spacing keyframes always
[22:45] <Mavrik> without compromising quality due to forcing them to EXACTLY 1 sec spacing
[22:46] <Mavrik> on fast / action movies you'll be burning alot of bitrate for pointless full-frame storage
[22:46] <FrEaKmAn_> so what should I do?
[22:47] <brontosaurusrex> Mavrik, right
[22:48] <Mavrik> FrEaKmAn_, just do whatever, keyframe intervals are always passed in frame numbers
[22:48] <brontosaurusrex> althought CBR is even worse idea there
[22:48] <Mavrik> mhm, unless video is prepared for streaming
[22:49] <FrEaKmAn_> it's for streaming
[22:49] <FrEaKmAn_> in my case
[22:49] <brontosaurusrex> and how is CBR good idea for streaming?
[22:49] <Mavrik> brontosaurusrex, overlooked the -mirate
[22:50] <Mavrik> *minrate
[22:50] <Mavrik> that's madness.
[22:50] <brontosaurusrex> yeah
[22:50] <Mavrik> but yes, I have gotten requests from my customers to implement full CBR streaming
[22:50] <Mavrik> Explaination: "When using Envivio the bitrate graph looks prettier."
[22:50] <brontosaurusrex> Mavrik, any "whys" ?
[22:50] <brontosaurusrex> oh
[22:51] <Mavrik> and then they showed me a MPEG2 stream that was stuffed for exact 20Mbit CBR meant for DVB-T broadcasting
[22:51] <Mavrik> which needs exact bitrate :P
[22:51] <brontosaurusrex> well, not such a horrible idea when you have the reserved channel i guess
[22:52] <brontosaurusrex> but even then
[22:52] <Mavrik> Well, they didn't.
[22:52] <Mavrik> Luckly I managed to explain to them to stop allowing management people to look at technical data.
[22:53] <brontosaurusrex> FrEaKmAn_, so you are using osmf and octoshape?
[22:53] <FrEaKmAn_> only osmf
[22:53] <FrEaKmAn_> and wowza for streaming
[22:56] <brontosaurusrex> FrEaKmAn_, examples here http://www.wowza.com/forums/content.php?192-How-to-encode-video-on-demand-content
[22:57] <brontosaurusrex> "-g" = gop size
[22:58] <brontosaurusrex> this are 1pass vbr
[23:05] <FrEaKmAn_> ok
[23:05] <brontosaurusrex> why are you using osmf? isnt that like a dead project?
[23:06] <FrEaKmAn_> thanks,, will check first thing in the morning
[23:06] <FrEaKmAn_> yes, but works great
[23:13] <brontosaurusrex> ok, how about mobile devices?
[00:00] --- Wed Aug  6 2014