[Ffmpeg-devel-irc] ffmpeg.log.20141030

[burek]

[02:22] <edman007> Hi, how do I change the sample rate of a an audio file? I got a 44.1kHz mp3 I'm trying to encode to a 48kHz ac3 file... is there an audio filter I need to run it through?
[02:35] <edman007> oh, I guess -ar works
[03:00] <edibile> it's not possible with a complex filter to avoid transcoding, right?
[03:07] <edibile> i'm combining multiple videos into a matrix and trying to reduce cpu resources required but it doesn't appear there is a way to do this without transcoding
[03:13] <relaxed> edibile: correct
[03:20] <khaosmarmoset> hey. can I do with the ffmpeg binary everything that I can do with ffplay, ffserver, ffprobe?
[03:21] <khaosmarmoset> I probably only will need ffmpeg, although I check stuff with ffprobe once in a while
[08:51] <Baked_Cake> Baked_Cake_ (~Baked_Cak at c-24-20-70-164.hsd1.or.comcast.net) has disconnected: Ping timeout: 256 seconds
[08:51] <Baked_Cake> o:
[10:00] <wh-hw> hi, all
[10:00] <wh-hw> what does this mean : value inf for parameter 'time_base' out of range ?
[11:00] <khaosmarmoset> infinity
[11:54] <khaosmarmoset> -f tee is giving me grief.
[11:55] <khaosmarmoset> when I replace my -f flv out.flv with -f tee -map 0:a -map 0:v "[f=flv]out1.flv|[f=flv]out2.flv" I get a complaint about there being nothing to map.
[11:55] <khaosmarmoset> the example with ffmpeg piping to another ffmpeg that uses -c copy works as advertised.
[11:55] <khaosmarmoset> but why doesn't -f tee work out of the box?
[12:06] <relaxed> khaosmarmoset: pastebin your command and output
[12:16] <khaosmarmoset> http://pastebin.com/sg4JTCpN
[12:18] <khaosmarmoset> ok I changed the -map 0:v to -map 1:v because of course my first input has no video, just like it says
[12:23] Action: khaosmarmoset bestows relaxed the best honorary rubber duck award
[12:24] <khaosmarmoset> multiple problems between chair and keyboard.
[13:14] <relaxed> khaosmarmoset: it happens :)
[13:46] <damnesia> Hi. For a yuv420p | i420 video, is it encoded as YYYYYYY (repeated width*height) UU (repeated w*h / 2) VV (repeated w*h/2) for every frame?
[14:06] <termos> Is there a good and general way to increase the input buffer to FFmpeg, I need a way to make it more robust to laggy or unstable input streams
[14:08] <khaosmarmoset> termos: not having read the docs at all, what I tried with some success was to add gstreamer in front of ffmpeg and use its queue element to buffer things up
[14:09] <khaosmarmoset> termos: I'm sure there's an ffmpeg way to do this too.
[14:09] <khaosmarmoset> termos: the gst way works if you like rube goldberg :)
[14:09] <termos> ah thanks, unfortunately I won't be able to use the gstreamer way of doing it. Hmm
[14:10] <termos> hehe
[14:10] <termos> actually I am using the C api, so I guess I need to make my own kind of buffering system before the encoding step
[14:12] <relaxed> termos: look at probesize and analyzeduration in the man page
[14:16] <termos> I will look into those, I just found something that might work though
[14:17] <termos> setting the rtmp_buffer value in AVDictionary and passing it to avformat_find_stream_info()
[14:21] <termos> No, it did not work...
[14:44] <termos> relaxed: do you know how I can set those using the C api by any chance? I can't figure it out, it's really weird
[15:40] <Diogo> hi.... Command: ffmpeg -y -i "http://dl.dropboxusercontent.com/u/2561075/2014-10-30%2014.21.41.mov" -c:v libx264  -f mp4 result.mp4
[15:40] <Diogo> ---> I having a problem with mov attom and lots of HTTP 206 resquest. What is the solution for this probem. I want to move the attom to the begin of file and the source came from url. Because with other format mp4/avi/webm this command works very well
[15:42] <kepstin-laptop> Diogo: if you're writing to a file, you can use the '-movflags faststart' option to get ffmpeg to move the moov atom to the beginning of the file.
[15:42] <Diogo> but i can't download the file..
[15:42] <Diogo> :(
[15:43] <Diogo> because i need to process the file very fast
[15:43] <kepstin-laptop> oh, the problem's that you can't play the file off the dropbox url?
[15:44] <Diogo> i want that this command works for any video..
[15:44] <Diogo> ffmpeg -y -i "<URL>" ....
[15:45] <Diogo> this command works for any format except mov files.. mov files ffmpeg send lots of HTTP 206 request before start encode the video
[15:45] <Diogo> i want to know why this happend...because with other formats ffmpeg only do 2/3 http 206 requests..
[15:46] <kepstin-laptop> well, unless the moov atom is at the start of the file, ffmpeg has to look through the file to find it before it can start playing
[15:47] <kepstin-laptop> inherent problem in the mov/mp4 format.
[15:47] <Diogo> can we save the moov atom in other storage and after that concatenate and pipe to ffmpeg?
[15:47] <Diogo> ffprobe <URL> --> this command only do 2 http requests
[15:48] <Diogo> the same mov file
[15:48] <kepstin-laptop> well, ideally would be that you remux the original file (using ffmpeg with -c copy or the old qt-faststart tool) before uploading it, if you have control of that...
[15:48] <Diogo> i can't change the original content :(
[15:49] <Diogo> because its a user content if i change any part of file the md5sum change :(
[15:49] <kepstin-laptop> well, then either live with it, or download the entire thing locally before encoding.
[15:50] <Diogo> i need a pratical solution for this... 1º Http 206 to get moov atom 2º Concatenate in the begin of video 3º start encode..
[15:50] <Diogo> right?
[15:51] <kepstin-laptop> the problem is finding the moov atom in the first place
[15:51] <kepstin-laptop> btw, your example video has your gps coordinates in it.
[15:54] <Diogo> sorry :(
[15:58] <Fyr> how do I get mediainfo using ffmpeg?
[16:01] <kepstin-laptop> Diogo: basically, to implement something like dropbox has, you have to have users upload the video directly to you so that you can do the fast-start processing on the whole file, possibly saving to a second file, before or while putting it into storage.
[16:03] <Nadaesque> Hello.
[16:07] <Nadaesque> I was wondering ... I have a source video that is about 1280x720 I would like to put on a DVD for my mom's ancient DVD player.  I know I'll need dvdauthor later, but in terms of ffmpeg, is there anything I need besides "-target ntsc-dvd" in the conversion to get it to work, since the original aspect ratio is 16:9 and NTSC for DVDs is very much not 16:9?
[16:11] <Mavrik> um
[16:11] <Mavrik> NTSC for DVDs can easly be 16:9 :)
[16:11] <kepstin-laptop> Nadaesque: NTSC on DVDs can be either 4:3 or 16:9, and ffmpeg will set up the aspect ratio in the stream correctly.
[16:11] <Mavrik> target parameter should take care of these things
[16:15] <Nadaesque> Really?  Even ancient POS DVD players can handle it?
[16:17] <Nadaesque> It'll just do the padding automatically?  Or will it store it anamorphically?  Will it to this only for 4:3 and 16:9, or can other aspect ratios deal with it?
[16:17] <kepstin-laptop> Nadaesque: both 4:3 and 19:6 video on dvds are anamorphic
[16:18] <kepstin-laptop> usually players have an option to switch between letterboxing and anamorphic video output for widescreen dvds on the analogue output
[16:18] <Mavrik> yeah, 16:9 is something that predates DVDs, so all players must support 16:9 AR
[16:19] <Nadaesque> I know that the TV pixels are tall rectangles rather than square, so the 720x480 turns into 640x480 when it hits a TV screen.
[16:19] <kepstin-laptop> Nadaesque: technically not correct, particularly on crts
[16:20] <Nadaesque> No?  What is it then?
[16:20] <kepstin-laptop> you're right, the pixels are tall rectangles. you're wrong that the resolution changes to 640x480
[16:20] <Mavrik> 16:9 CRTs and other TVs just stretched the NTSC input on 16:9
[16:21] <Nadaesque> I'm just, uh, talking about how you would draw it out with a piece of graph paper on a TV CRT.
[16:21] <Nadaesque> Hrm, but other aspect ratios besides 4:3 and 16:9 will require the expression evaluator to perform padding magic in the video filter?
[16:21] <kepstin-laptop> since computer monitors have square pixels, computer players often rescale dvd video to 640x480, but CRT tvs don't have pixels
[16:22] <kepstin-laptop> remember that each line of video on a CRT is a continuous analogue voltage :)
[16:22] <Nadaesque> Yeah.
[16:22] <Nadaesque> <=== had to make an oscilloscope in college for his degree.  GAH.
[16:23] <kepstin-laptop> so the '720 pixels' in the video stream aren't really pixels, but are actually samples used to reconstruct the analogue signal.
[16:24] <Nadaesque> So, things with weird aspect ratios will require extra work?
[16:24] <kepstin-laptop> yes, things with weird aspect ratios will require extra work :) you'll have to pad or stretch them to either 4:3 or 16:9
[16:24] <Nadaesque> I feared as much.
[16:25] <Nadaesque> We sometimes get requests to put something on a DVD so the uh ... less technically adapt professors can play it.
[16:25] <Nadaesque> Erf, adept.
[16:26] <kepstin-laptop> the real fun happens when you encode a video on dvd as 4:3, and then the viewer has a widescreen tv and sets the tv to stretch the signal (i.e. in 16:9 anamorphic mode)
[16:26] <kepstin-laptop> then everything is just wide :(
[16:27] <Nadaesque> Well, I will use my mom's thing as a trial run.  I have shamed myself by trying to put this Lionel Richie concert she wanted to see on DVD.
[16:27] <kepstin-laptop> (only real solution to that problem is to use an upscaling dvd player with HDTV resolution output, preferrably on HDMI)
[16:28] <Nadaesque> Well, I want to try a 4:3 thing for the profs and let them just deal, because they don't know what HDMI is or anything.
[16:28] <kepstin-laptop> and even then people don't like bars around their video, so they stretch it anyways :/
[16:28] <Nadaesque> I once had this professor, she got an AV cart and plugged the power strip into ... itself.
[16:28] <Nadaesque> She didn't understand why it wasn't working.
[16:28] <kepstin-laptop> and then you're all http://i5.minus.com/i6DRQ32FboEix.jpg
[16:29] <Nadaesque> So, ffmpeg -> dvdauthor -> something to make an .iso -> burnination.
[16:31] <kepstin-laptop> hmm, you can't use "just anything" to make the iso; it has to be UDF with possibly some other constraints
[16:31] <kepstin-laptop> many iso creation tools do have an option to put them into a dvd-specific mode, tho
[16:31] <kepstin-laptop> dvd-video-specific
[16:32] <Nadaesque> Yeah, I'll have to find something for that, too.
[16:32] <Nadaesque> I knew there was some filesystem funkiness for them and Joliet was probably =NOT IT=.
[16:33] <kepstin-laptop> on mkisofs, you just have to use the "-dvd-video" option, I think
[16:33] <Nadaesque> Aaah, that sounds good.
[16:33] <kepstin-laptop> (read the man page for details, there's some funkiness about case sensitivity)
[16:33] <flavioberetti> hello. i need to stream video in (near) realtime between to network hosts with a fast local connection. i would like to stream in raw format, without the time to de/encode before transmission. however if i simply use rawvideo with mpegts for example, i get 'stream is not seekable' because there is no information about frames.
[16:33] <flavioberetti> is there a codec, or any way to embed uncompressed video in a network stream?
[16:33] <Nadaesque> I guess, since DVDs are MPEG-2 TS, there's no access to (or reason to do) two pass encoding?
[16:34] <kepstin-laptop> Nadaesque: container format is irrelevant to 2-pass encoding
[16:34] <Nadaesque> So I could do 2-pass for a higher quality?
[16:34] <kepstin-laptop> 2-pass encoding is a way to get the maximum quality possible overall for a given desired video bitrate
[16:34] <Nadaesque> I thought it was only for H.264 stuff.
[16:35] <kepstin-laptop> the concept is unrelated to any particular codec, it's just a method of improving bitrate allocation.
[16:35] <Nadaesque> Hunh.  I guess I have only seen references to it in conjunction with H.264, then.
[16:36] <Nadaesque> Is it just ... overkill for DVD-Video?
[16:36] <kepstin-laptop> it only really picked up when people started doing mpeg4-asp (e.g. divx,xvid) video
[16:37] <kepstin-laptop> Nadaesque: I wouldn't say so; if you select the biggest bitrate that's dvd-spec complient and allows the video to fit on the disc, 2-pass encoding can let the encoder move bits to places where it'll help quality
[16:37] <kepstin-laptop> I *think* the ffmpeg mpeg2 encoder can do 2pass, but I haven't actually checked...
[16:38] <kepstin-laptop> yes, looks like it does
[16:39] <Nadaesque> I sort of want to develop this into a more serious workflow so we can do this in a more automated fashion with high-quality results, but I can sometimes be Mr. Overkill about things.
[16:44] <kepstin-laptop> note that you'll want to select between the "film-dvd" and "ntsc-dvd" targets depending on whether your input is a movie/animation/cinematic-style tv show (24fps) or standard tv (30fps)
[16:45] <kepstin-laptop> although I think the only difference between the two is the selection of output framerate
[16:46] <Nadaesque> Hrm.  I should get a variety of samples and work up a list of potential questions.
[16:47] <Nadaesque> Sometimes we get H.264s clearly ganked off of a PAL DVD via Handbrake.
[16:47] <Nadaesque> And we get a lot of weeeeeird items in the mix.
[16:47] <kepstin-laptop> oh, that'll be fun to deal with. and by fun I mean the complete opposite
[16:48] Action: Nadaesque shuffles to the counter and looks up through foggy lenses.  "I need this Laserdisc and this Betamax put on DVD for a class?"
[16:48] <kepstin-laptop> to encode pal video to ntsc, you need to encode interlaced with a 2:2:2:2:3 pulldown applied, I think :/
[16:48] <kepstin-laptop> (i.e. every 10 fields is turned into 11 fields, to turn the 50 fields/s video into 60 fields/s)
[16:49] <Nadaesque> That makes sense.  Of course, who knows what kind of mangling the file has had before it reached our hands?
[16:50] <kepstin-laptop> if it was a pal encode of a US movie, it was probably sped up from 24 to 25fps first (this often causes noticable audio pitch changes on older stuff)
[16:50] <Nadaesque> Oh, yeah, I hate that.
[16:50] <Nadaesque> I had a really sweet multi-region DVD player once upon a time with a pitch-shifter in it.
[16:50] <Nadaesque> Because that pitch shift was like getting tased in the eardrum.
[16:52] Action: kepstin-laptop just tried encoding a random test video in 2-pass mpeg2 video at 8Mbit/s, and was politely told by ffmpeg that "Using all of requested bitrate is not necessary for this video with these parameters."
[16:53] <Nadaesque> Heh!
[16:53] <kepstin-laptop> (the final output ended up being around 2.5Mbps)
[16:53] <relaxed> kepstin-laptop: ffmpeg knows best
[16:56] <relaxed> kepstin-laptop: did you try forcing it with -minrate ?
[16:57] <Nadaesque> Well, thank you very much for this advice.  I will probably return for more refinements as I run across ever-more-unlikely scenarios.
[16:57] <kepstin-laptop> why? It's not like I actually need the bitrate to be 8mbit if the quality is fine :)
[17:00] <Nadaesque> Must use all available bits.  Yussss.
[17:05] <eristisk> My H.264 video streams from the Canon 5D Mark III play perfectly fine with ffplay, but VLC and Mplayer give lots of distorted/pixelated frames on Debian
[17:06] <eristisk> Mplayer on the terminal complains with many errors "Error while decoding frame!"
[17:07] <kepstin-laptop> which debian version? Note that for a while, they've been compiling stuff to use libav rather than ffmpeg.
[17:07] <c_14> Tried with the VLC nightly or a decent video player like MPV?
[17:07] <eristisk> I am using the Intel GPU on my Ivy Bridge CPU, but there is also an Nvidia card on the system which I normally do not run.
[17:08] <eristisk> c_14: VLC nightly I have not tried.  I thought it was a VLC error until I noticed the same problem in Mplayer.
[17:08] <eristisk> kepstin-laptop: Debian sid... there is weird movements to switch back to ffmpeg, and I think I somehow have both libraries installed, but I am not sure which VLC is using
[17:08] <eristisk> I am installing MPV right now
[17:09] <eristisk> MPV gives the same pixelated/distorted frames and Error while decoding frame!  errors on the terminal
[17:09] <c_14> Is the ffplay you're using system or manually compiled?
[17:10] <kepstin-laptop> eristisk: just out of curiosity, check `ldd` output on your ffplay and mpv and see which libraries they're using
[17:10] <c_14> And what version is it?
[17:10] <eristisk> Installed with APT, not compiled.
[17:10] <eristisk> ffplay version 2.4.2-1 Copyright (c) 2003-2014 the FFmpeg developers
[17:10] <eristisk>   built on Oct  6 2014 19:13:39 with gcc 4.9.1 (Debian 4.9.1-14)
[17:11] <kepstin-laptop> looks like your ffplay is ffmpeg, but I bet the other players are using libav.
[17:12] <eristisk> mpv is using libavcodec.so.56 but ffplay is using libavformat-ffmpeg.so.56
[17:12] <eristisk> kepstin-laptop: Right, seems that way.
[17:12] <eristisk> That is evil, how do I kill libav?  Can I make VLC use ffmpeg?
[17:12] <kepstin-laptop> you have to recompile it.
[17:13] <eristisk> Recompile VLC?  Damn...
[17:13] <kepstin-laptop> well, you might not, but it would be highly recommended to avoid possible incompatibilities :)
[17:16] <kepstin-laptop> oh, hey, ffmpeg's made it into debian unstable now, it's not in experimental any more.
[17:16] <rcombs> wow, that was fast
[17:17] <eristisk> Yeah I know ffmpeg is in unstable, I have it installed.  I just want a VLC compiled against ffmpeg libraries now  :/
[17:17] <eristisk> This experience makes me pissed off at libav
[17:18] <kepstin-laptop> eristisk: you could file a bug with debian saying that vlc fails to play some files, noting that it might be fixed by recompiling with ffmpeg; on the other hand, that might also start up a new flamewar thread ;)
[17:19] <relaxed> ffmpeg won't be in Jessie so I doubt that will happen.
[17:19] <edibile> it's not possible with a complex filter to avoid transcoding, right?
[17:19] <edibile> i'm combining multiple videos into a matrix as a single video stream and trying to reduce cpu resources required but it doesn't appear there is a way to do this without transcoding
[17:20] <c_14> edibile: you are correct
[17:20] <relaxed> correct
[17:20] <edibile> are there any options i should look at to reduce cpu load?  this is for streams to iphones, so resolution/audio quality aren't of great concern.
[17:21] <edibile> (finding a way to play udp streams on an iphone is another matter...)
[17:21] <c_14> edibile: it would help if we had your current commandline
[17:21] <edibile> sure, just a second
[17:22] <kepstin-laptop> edibile: if you're doing h264 encoding with x264, just change the 'preset' value to a faster setting.
[17:24] <edibile> http://pastebin.com/E5kGykCe
[17:24] <edibile> ok, will look at the preset option
[17:25] <c_14> yeah, preset will be your best bet, a higher crf might also help
[17:27] <kepstin-laptop> edibile: preset defaults to 'medium', try changing it to 'fast' or 'veryfast'
[17:27] <relaxed> it's already baseline
[17:28] <relaxed> and can an iPhone decode mp2 audio?
[17:29] <eristisk> When I run apt build-dep vlc then it does not try to install library development headers for ffmpeg libraries.  Only -dev packages for libav stuff...
[17:30] <relaxed> eristisk: I highly recommend https://github.com/mpv-player/mpv-build
[17:32] <eristisk> relaxed: MPV is by default compiled to libav libraries as well... but these sources will compile against ffmpeg on Debian?
[17:32] <kepstin-laptop> eristisk: you'll have to edit the packaging files for vlc to install the ffmpeg dev stuff, and you may need to do additional work to get it to use ffmpeg rather than libav (not entirely sure what that entails)
[17:32] <eristisk> I am sure MPV is great, but I also like the GUI of VLC  (sorry if that is somehow controversial...)
[17:32] <relaxed> eristisk: correct
[17:33] <edibile> i'm not sure on the iphone decoding mp2 audio.  i still haven't found a reliable, free app to stream the udp video
[17:33] <relaxed> edibile: you most likely want aac for audio
[17:33] <Mavrik> iPhone almost certanly doesn't support MP2 audio out of the box
[17:33] <edibile> would reducing frame rate help as well?
[17:33] <Mavrik> also streaming baseline video to iPhones is just a huge waste of quality
[17:34] <kepstin-laptop> apparently it's required to support some really old models :/
[17:34] <edibile> i'm not sure what "baseline video" means
[17:34] <edibile> and what the alternative is
[17:35] <relaxed> edibile: http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC#Profiles
[17:35] <Mavrik> kepstin-laptop, pretty much anything still used out there supports Main
[17:35] <Mavrik> and the quality difference is very noticable
[17:35] <relaxed> 320x480 is tiny
[17:36] <edibile> i might bump up the resolution a bit, but was using that for testing
[17:45] <edibile> great, changing preset to veryfast and profile to main cut cpu usage in half
[17:49] <relaxed> edibile: did you see https://trac.ffmpeg.org/wiki/Encode/H.264 ?
[17:50] <edibile> yeah, i found that after you guys told me what i was looking for
[17:55] <edibile> jus tnoticed the ios compatibility section at the bottom, though
[18:13] <edibile> thanks guys, cpu down to 25% of what it was before
[18:18] <Diogo> hi, this is possible to revert the movflags faststart option?? ffmpeg  -y -i <FILE> -c copy -movflags faststart FILE.mov
[18:18] <Diogo> i want revert to original file
[18:19] <relaxed> you can omit it but all you'll get is the original streams.
[18:32] <edibile> compand is the right filter to use to cut out audio below a specified threshold (squelch, to remove fan noise in this case)
[18:32] <edibile> that was supposed to be a question...
[18:33] <erikblankinship> following instructions on creating a thumbnail for every second of video in an mp4, I am finding my thumbnails slightly off than those I get with the _same_ video source using javascript/html5 or quicktime export.  Suggestions?
[18:33] <erikblankinship> following these instructions: https://trac.ffmpeg.org/wiki/Create%20a%20thumbnail%20image%20every%20X%20seconds%20of%20the%20video
[18:40] <voip__> please help http://pastebin.com/PdmRAcBn
[18:46] <kippi_> hey
[18:48] <kippi_> Is there away to output if the audio level changes? I would like to plot this, I have seen volume detect however this only outputs at the end, I am looking for somthing that is realtime
[18:54] <kaotiko> hi
[19:01] <khaosmarmoset> hey.
[19:01] <khaosmarmoset> someone mentioned syncing audio and video before as something that needed to be done
[19:02] <khaosmarmoset> so at the time I was mostly not actually listening to the audio when I was live streaming and only listening to it when I was doing local tests
[19:02] <khaosmarmoset> also I was doing it with a different computer. now I'm trying different computers, and it seems like the audio sync is horrible.
[19:03] <voip__> relaxed, can you help me with [mpegts @ 0x316cae0] PES packet size mismatchme=00:31:33.14 bitrate=1005.2kbits/s
[19:03] <voip__>     Last message repeated 1 times
[19:03] <voip__> [aac @ 0x317c7e0] Input buffer exhausted before END element found
[19:03] <voip__> Error while decoding stream #0:0: Invalid data found when processing input
[19:03] <voip__> [h264 @ 0x38b40e0] negative number of zero coeffs at 29 18
[19:03] <voip__> [h264 @ 0x38b40e0] error while decoding MB 29 18
[19:03] <voip__> http://pastebin.com/PdmRAcBn
[19:03] <khaosmarmoset> I used to get only a few warnings like about Non-monotonous DTS in output stream 0:0; etc.etc.
[19:04] <khaosmarmoset> but now I get a lot more and the sync seems to change
[19:15] <khaosmarmoset> -async 1 or -vsync 1 are doing nothing for me.
[19:50] <khaosmarmoset> my problem was that the audio was before the video
[19:50] <khaosmarmoset> erh. after. it was lagging a lot
[19:50] <khaosmarmoset> i'm not really sure what I'm doing, but before I added a negative -itsoffset to the audio it seemed like -vsync and -async were not doing very much
[20:02] <khaosmarmoset> seems to work now. I think the negative offset might be helping the other syncs to do their thing
[20:04] <khaosmarmoset> ok. next problem is that if I use -tee to spew to both youtube and the filesystem at the same time everything goes wrong
[20:04] <llogan> the tee muxer?
[20:04] <khaosmarmoset> right
[20:05] <llogan> what do you mean by "everything goes wrong"?
[20:05] <khaosmarmoset> I get many more Non-monotonous DTS warnings and the youtube stream is way more corrupted than either recording to a local file or sending to youtube directly
[20:05] <khaosmarmoset> phone
[20:08] <khaosmarmoset> back.
[20:08] <khaosmarmoset> ok I guess I can do that.
[20:08] <khaosmarmoset> one moment.
[20:17] <khaosmarmoset> http://pastebin.com/uG2N8wH8
[20:18] <khaosmarmoset> so that's the output when I use -tee. when I watch the result on youtube I see a lot of grey areas in the image
[20:18] <llogan> add -flags +global_header
[20:18] <khaosmarmoset> where?
[20:19] <llogan> as an output option
[20:19] <khaosmarmoset> so that's before the -f tee, then right?
[20:20] <khaosmarmoset> can you see the end of the paste as well?
[20:20] <llogan> ffmpeg [global options] [input options] -i input [output options] output
[20:22] <khaosmarmoset> ok. I added that
[20:22] <khaosmarmoset> but I still get an awful lot of Non-monotonous DTS warnings
[20:29] <khaosmarmoset> well, the stream isn't as corrupted as before so perhaps that's okay. the audio seems to be pretty  much in sync even.
[20:30] <khaosmarmoset> although I'm not sure if that's because of the magic -itsoffset value or because it lets -async and -vsync do their thing
[20:31] <khaosmarmoset> regarding the last part of the pastebin, that's a command line I expected would work based on this: http://sonnati.wordpress.com/2011/08/30/ffmpeg-%E2%80%93-the-swiss-army-knife-of-internet-streaming-%E2%80%93-part-iv/7~
[20:31] <khaosmarmoset> but I get the Protocol not found error you see in the paste
[20:32] <khaosmarmoset> my plan was to separate upload to youtube from save to disk by a few dozen seconds so things wouldn't need to fit so tightly
[20:33] <khaosmarmoset> especially in case -tee doesn't work on site using the laptops there.
[20:34] <voip__> llogan, i need your help :)
[20:49] <llogan> voip__: i don't have an answer this time. try the ffmpeg-user mailing list.
[20:50] <voip__> ok
[20:57] <khaosmarmoset> why can't I send the flv file directly but get the Protocol not found error?
[20:58] <khaosmarmoset> afaict, I'm invoking ffmpeg exactly right there.
[21:16] <kippi_> Is there away to output if the audio level changes? I would like to plot this, I have seen volume detect however this only outputs at the end, I am looking for somthing that is realtime
[21:38] <Uber-Ich> can I use ffmpeg to convert .xcf to .png or is it limited to vidya only?
[21:40] <khaosmarmoset> could you not use gimp for that?
[21:41] <Uber-Ich> I could... I would just need to install xorg.
[21:42] <Uber-Ich> I might just switch over to my arch box and use gimp instead. Thanks khaosmarmoset!
[21:42] <khaosmarmoset> also try imagemagick.
[21:43] <khaosmarmoset> probably easier to script convert than use batch gimp
[21:43] <khaosmarmoset> he went away didn't he?
[21:44] Action: khaosmarmoset considers turning on all the noise
[21:55] <Mista-D> to just count "scene change" frames, what is a better method? -vf select='gt(scene\,0.4)' will output the frames, dont want to write them, just count.
[21:58] <khaosmarmoset> I count specific kinds of frames using ffprobe_show_frames
[21:58] <khaosmarmoset> erh. ffprobe -show_frames
[21:58] <khaosmarmoset> then grep and pipe to wc
[21:58] <kepstin-laptop> Mista-D: on an ffmpeg command, you can use as output "-c:v null -f null -" and it'll discard the video frames without encoding them and generate no output file.
[21:59] <Mista-D> kepstin-laptop: how do I know how many were found?
[21:59] <kepstin-laptop> hmm, that's a trickier question :)
[22:02] <khaosmarmoset> *sigh* my Protocol not found error only happens on the streaming laptops, but not on my dev box
[22:02] <khaosmarmoset> prolly missing some extra package
[22:03] Action: kepstin-laptop got the codec wrong anyways
[22:03] <khaosmarmoset> I used the same source and and exactly the same command line.
[22:03] <llogan> Mista-D: just look at the console output. frame= 53
[22:03] <khaosmarmoset> it's the same binary to boo.
[22:04] <khaosmarmoset> boot
[22:04] <khaosmarmoset> not any of it makes any kind of sense, because the streaming laptops can stream to youtube just find if they are grabbing from the webcam.
[22:09] <khaosmarmoset> aha
[22:09] <khaosmarmoset> bah
[22:09] <khaosmarmoset> my filename is cam-2014-10-30T23:22:24+0200.flv
[22:09] <khaosmarmoset> it tries to parse cam-2014-10-30T23 as the transport protocol
[22:09] <khaosmarmoset> on the dev laptop I was using /tmp/cam-... as the path
[22:33] <BtbN> put a ./ in front of it?
[22:38] <khaosmarmoset> yes
[23:22] <kippi_msa> hey
[23:39] <ArunC> Hi, is it possible to rotate AVFrame using sws_scale?
[23:53] <katayana> how do I play playlists (streaming, m3u) with ffplay?
[23:57] <WindozeNT> hi
[23:58] <WindozeNT> I'm having problems with FFMPEG
[00:00] --- Fri Oct 31 2014