[Ffmpeg-devel-irc] ffmpeg.log.20150306

burek burek021 at gmail.com
Sat Mar 7 02:05:01 CET 2015


[01:29:10 CET] <AStorm> hello, I'm having an issue with ffmpeg 2.5.3 and old flv
[01:29:42 CET] <AStorm> that flv is damaged (sorry, I cannot include it) and when ffmpeg hits the broken bit it skips ahead flv frames of random sizes
[01:29:51 CET] <AStorm> (and weird types)
[01:33:13 CET] <AStorm> the telltale sign of the broken packet is its size which is way different from the previous packet sizes
[02:22:15 CET] <Max-P> Hi, does anyone know the correct options to encode H.264 videos that are compatible with Windows 7 out of the box?
[02:23:03 CET] <Max-P> I've been messing around for over an hour with ffprobe and ffmpeg to try to find what the hell some videos play fine and some don't, can't find anything.
[02:24:06 CET] <klaxa> what kind of videos work and which don't? h.264 should generally be supported i think
[02:24:22 CET] <c_14> If you can give me the ffprobe of one that works and one that doesn't, I can probably tell you what could be causing it. If you want to be sure it works, use -profile:v baseline -level 3.0 -pix_fmt yuv420p
[02:24:46 CET] <klaxa> maybe it's even container-related
[02:25:05 CET] <klaxa> i think windows 8 was the first to support matroska out-of-the-box
[02:28:47 CET] <Max-P> klaxa, c_14: http://pastebin.com/VYXJGeyY
[02:29:29 CET] <Max-P> I'm using -profile:baseline and -level 3.0. I needed that video to work even on old Apple devices, so I went with everything as low as I can. WMP plays the audio but not the video.
[02:29:53 CET] <klaxa> haha what
[02:29:56 CET] <klaxa> that's quite funny
[02:31:03 CET] <klaxa> i honestly can't tell why windows won't play it
[02:31:04 CET] <Max-P> Err, -profile:v baseline
[02:31:23 CET] <Max-P> And it works in Windows 8.1 too, I don't understand either
[02:31:54 CET] <Max-P> I would expect the working video to be the one that doesn't work because it was encoded with more advanced settings.
[02:32:18 CET] <klaxa> indeed, the second video should be more compatible
[02:36:59 CET] <c_14> Can you check in 'windows media player>help>about windows media player>technical support information' if it lists anything interesting?
[02:37:11 CET] <c_14> Either under formats or video codecs or something?
[02:37:42 CET] <Max-P> LOL, wow
[02:37:49 CET] <Max-P> I think I found the problem
[02:38:02 CET] <Max-P> ... it doesn't play vertical videos.
[02:38:28 CET] <c_14> olo?
[02:40:50 CET] <Max-P> Yep, if I add "-vf transpose", it plays fine, no matter what settings
[02:41:56 CET] <c_14> Oh man.
[03:04:52 CET] <AStorm> Max-P: it's probably not vertical vs horizontal
[03:04:57 CET] <AStorm> check if you have square pixels
[03:05:06 CET] <AStorm> it might not support the "wrong" rectangular pixel
[03:05:31 CET] <AStorm> it definitely supports wider pixels than square, but maybe not narrower
[03:06:03 CET] <c_14> AStorm: both had square pixels
[03:06:10 CET] <AStorm> what the heck then :D
[03:06:33 CET] <AStorm> my win 7 wmp (latest updates) handles narrow h.264
[03:07:26 CET] <AStorm> maybe ffmpeg does something funny there? such as switching some orientation flag instead of changing resolution?
[03:08:37 CET] <Max-P> AStorm, Here are two sample files: http://d.max-p.me/temp/ffmpeg/
[03:09:22 CET] <Max-P> The rotation flag is explicitely cleared in the process as I use the transpose filter (which is to fix another bug with a specific browser playing the video sideways)
[03:09:44 CET] <AStorm> well, 403
[03:09:50 CET] <Max-P> Oh, hold in fixing it
[03:10:05 CET] <Max-P> There, fixed
[03:11:53 CET] <AStorm> DAR 9:16 - here's your problem
[03:12:29 CET] <c_14> The dar just means that it's vertical, the pixels are still square
[03:13:38 CET] <AStorm> yes but again, DAR field has nothing to do with actual aspect ratio
[03:14:30 CET] <AStorm> if you just munge this to something that is >= 1.0, (4:3, 16:9) you'll be fine
[03:16:49 CET] <Max-P> AStorm, Well I have nothing to lose. How do I change that?
[03:16:59 CET] <AStorm> not sure yet
[03:17:07 CET] <AStorm> definitely ffmpeg doesn't allow that
[03:23:33 CET] <AStorm> no, you can do that with -aspect <number> -vcodec copy after encoding
[03:23:43 CET] <AStorm> otherwise ffmpeg gets smart and scales
[03:25:21 CET] <Max-P> AStorm, Err, I can confirm WMP can play DAR 9:16 now
[03:25:32 CET] <AStorm> ok, so, what is wrong then?
[03:25:33 CET] <Max-P> 640x640 [SAR 9:16 DAR 9:16]
[03:25:38 CET] <AStorm> wrong!
[03:25:43 CET] <AStorm> SAR == DAR is different
[03:25:46 CET] <Max-P> This works. This is incredibly silly, but it works ._.
[03:26:01 CET] <AStorm> you essentially did <1.0 ratio pixels
[03:26:14 CET] <AStorm> but then did DAR that is the inverse
[03:26:20 CET] <AStorm> so, you got square pixels back
[03:26:38 CET] <AStorm> the bug seems to be about the *result* pixel shape
[03:26:55 CET] <Max-P> That video displays as 9:16
[03:27:16 CET] <AStorm> correct, it will, you gave it nonsquare pixels
[03:27:45 CET] <Max-P> So, what do I want exactly. Both at 1:1 ?
[03:27:56 CET] <AStorm> that will work, yes
[03:28:11 CET] <AStorm> or alternatively, DAR at something else common such as 16:9, 4:3
[03:28:22 CET] <AStorm> then it won't be scaled either (most of the time)
[03:28:34 CET] <AStorm> with SAR 1:1
[03:28:35 CET] <Max-P> Yeah, but then the video will be horizontal, and squeezed.
[03:28:47 CET] <Max-P> Unless I add black bars
[03:28:50 CET] <AStorm> no, it won't be horizontal at all, that is governed by resolution
[03:29:07 CET] <AStorm> because PAR is still 1:1
[03:29:35 CET] <AStorm> DAR is akin to "these pixels were PAR shaped when mastered on a display with this DAR"
[03:30:03 CET] <AStorm> at least this is how wmp interprets it
[03:30:46 CET] <AStorm> I'm not entirely sure if this interpretation is correct, let's check MPEG-4 standard
[03:31:23 CET] <Max-P> At this point I think I'll just hardcode IE to always use Flash and call it a day. If the user wanted an actual browser they'd use Chrome or Firefox anyway
[03:33:23 CET] <Max-P> The main problem is that IE refuses to play videos, because IE advertises being able to play MP4 videos so I load up the HTML5 player
[03:33:23 CET] <AStorm> just make sure to provide the download link
[03:33:43 CET] <Max-P> Turns out it's a bunch of lies
[03:36:13 CET] <AStorm> hmm? what does falsh(TF) do?
[03:36:53 CET] <Max-P> AStorm, what?
[03:37:13 CET] <AStorm> what do you mean by "bunch of lies"
[03:37:23 CET] <AStorm> ah, I see, I missed a line
[03:37:26 CET] <AStorm> never mind
[05:43:00 CET] <BJsgoodlife> Hi everyone, I have a very basic question. What directory should you be in when using ffmpeg?
[05:46:02 CET] <kapy> BJsgoodlife: have you downloaded the source for ffmpeg or the package ?
[05:46:28 CET] <BJsgoodlife> source, I compliled the source from ubuntuforums.org
[05:47:04 CET] <kapy> then you have to be in the source folder
[05:47:47 CET] <BJsgoodlife> Ok kapy thank you for your help.
[05:48:06 CET] <kapy> np
[09:52:06 CET] <cousin_luigi> Greetings.
[09:52:13 CET] <cousin_luigi> Can ffmpeg fully replace libmad?
[10:42:59 CET] <Seb_Sob> Hey
[10:43:41 CET] <Seb_Sob> Is it possible to have an image input, put watermark on it, and output as an image
[10:46:28 CET] <Seb_Sob> Anyone here?
[10:48:29 CET] <BtbN> yes
[10:48:34 CET] <Seb_Sob> hi
[10:48:38 CET] <Seb_Sob> can you help me?
[10:48:54 CET] <BtbN> Well, i can at least tell you that the answer is yes, that's definitely possible.
[10:49:13 CET] <BtbN> An image is just a special case of a video, with a single frame.
[10:49:32 CET] <BtbN> Just look for examples on how to watermark a video, and do the same, just with an image as input and output
[10:50:44 CET] <BtbN> But if you just want to overlay an image, maybe the imagemagick composite tool is better for that job?
[10:51:45 CET] <Seb_Sob> but its for an application
[10:51:57 CET] <Seb_Sob> i can only use ffmpeg
[11:23:14 CET] <cousin_luigi> does ffmpeg do everything that libmad does?
[11:23:54 CET] <Mavrik> What does libmad do?
[11:25:27 CET] <iive> mp3 decpding
[11:25:34 CET] <iive> decoding
[11:27:40 CET] <Mavrik> ah, isn't that another libmad that does streaming or some such
[11:29:21 CET] <iive> never heard of that one. google also doesn't show it on 1'st page.
[13:20:20 CET] <cousin_luigi> iive: I'm asking because it's oldish and I wonder if it's still good to keep.
[13:28:48 CET] <iive> yes, it is good to keep. the implementation is quite solid, even if not fast.
[13:29:05 CET] <iive> it supports mp1,mp2, and few variants of mp2.5
[13:55:01 CET] <DeadSix27> how would one archieve to hide the prefix when compiling? e.g so it wouldnt appear in the banner later on when running the self compiled ffmpeg. (prefix = the location to compile to)
[14:25:45 CET] <luc4> Hello! I'm using this command to extract one frame every 5 seconds: ffmpeg -i <file> -vf fps=1/5 out%03d.jpg. But it seems wrong.... If I extract a single frame using ffmpeg -i <file> -ss <time> -vframes 1 out.jpg I get a completely different frame. Comparing with the output of vlc it seems extracting one frame is the same, but the first one is not. Is my first command wrong?
[15:33:28 CET] <mc_ebu> hello, I have a problem with FFMPEG that cannot join a mutlicast group with a specific source (SSM, IGMPv3). vlc works but not ffmpeg
[15:34:09 CET] <mc_ebu> I use the following syntax: ffmpeg -i udp://10.101.23.2@232.200.1.3:5000
[15:35:01 CET] <mc_ebu> but looking at packets i see the query going out with the source specified:  10.107.1.2 > igmp.mcast.net: igmp v3 report, 1 group record(s) [gaddr 232.200.1.3 to_ex, 0 source(s)]
[15:35:53 CET] <mc_ebu> tested with deb-multimedia version and also with latest ffmpeg version (2.5.4). Has anybody any clue on this ?
[15:41:05 CET] <BtbN> udp.c says /* TODO 'sources', 'block' option */
[15:41:14 CET] <BtbN> so i'd assume support for sources is not yet implemented?
[15:44:30 CET] <BtbN> mc_ebu, try this url: udp://232.200.1.3:5000?sources=10.101.23.2
[15:47:38 CET] <mc_ebu> BtbN: it works !
[15:48:18 CET] <mc_ebu> BtbN: thanks ! Now I have to look if it works with mplayer because my original use is to play a multicast stream
[15:48:25 CET] <BtbN> no idea if this is documented somewhere.
[15:48:35 CET] <BtbN> That url format is quite ffmpeg specific
[15:50:33 CET] <mc_ebu> BtbN: effectively, it doesn't work: mplayer ffmpeg://udp://232.200.1.3:5000?sources=10.101.23.2
[15:50:50 CET] <BtbN> no idea what and how mplayer does with its urls
[15:50:57 CET] <BtbN> Why mplayer, if vlc works fine?
[15:52:21 CET] <mc_ebu> BtbN: because vlc doesn't handle jack as I need, I have many players in parallel and need to have them to use specific label name on jack. Only mplayer allow that, vlc choose random label
[15:53:05 CET] <BtbN> Is it audio only?
[15:53:16 CET] <mc_ebu> BtbN: looking on forum they where advising this syntax mplayer ffmpeg://udp://10.101.23.2@232.200.1.3:5000
[15:53:35 CET] <BtbN> Well, this syntax obviously didn't work.
[15:53:39 CET] <mc_ebu> BtbN: yes, audio only getting it on TS
[15:55:38 CET] <mc_ebu> BtbN: mplayer is convenient because I can use it for all cases, also taking audio from internet audio streams (mp3, aac,...) and I can do a good resampling
[15:55:43 CET] <BtbN> hm, no jack output filter in ffmpeg.
[15:56:13 CET] <BtbN> Can't you just output to alsa and get that output into jack?
[15:56:31 CET] <mc_ebu> BtbN: VLC would also be suitable but there's just this limited jack functionalities that would make it it difficult to jack connect the right source to the right sink
[15:57:07 CET] <BtbN> jack propably isn't a common usecase, and not well tested.
[15:57:54 CET] <mc_ebu> BtbN: would be a possibility using jack-stdio, we did this in the past but this was a bit less stable. Otherwise alsa to jack mapping using jackrc is ugly
[15:58:41 CET] <mc_ebu> BtbN: jack is great when you have multiple streams to handle and individually disconnect them, reroute them, change source without stoppping everything
[15:59:28 CET] <BtbN> I usualy use PA for that.
[15:59:49 CET] <mc_ebu> BtbN: it worked great with mplayer, also with multicast but we had to switch to IGMPv3. Another workaround is to force the ethernet switch to subscribe to the group but this is not very elegant also
[16:00:18 CET] <mc_ebu> BtbN: PA ?
[16:00:23 CET] <BtbN> I'd guess the mplayer ffmpeg url thing just doesn't propperly forward the URL parameters.
[16:00:26 CET] <kepstin-laptop> pulseaudio serves a bit of a differnt purpose,  it's not designed to hook applications up to eachother.
[16:00:38 CET] <BtbN> Still possible to do that though, kepstin-laptop
[16:00:46 CET] <BtbN> Usualy well enough for non-professional needs
[16:01:13 CET] <mc_ebu> BtbN: so it is a mplayer problem in the end ?
[16:01:29 CET] <BtbN> If it works with the ffmpeg cli tool, what else would it be?
[16:07:34 CET] <mc_ebu> BtbN: thank  you for your valuable help. I have forwarded the issue to mplayer channel
[16:07:43 CET] <luc4> Hello! Can anyway confirm that this command ffmpeg -i myvideo.avi -vf fps=1/5 img%03d.jpg takes 1 frame every 5 seconds starting from 0?
[16:08:34 CET] <BtbN> mc_ebu, ffmpeg with pcm output to stdout and jack-stdin doesn't work?
[16:09:08 CET] <mc_ebu> BtbN: yes, this would work
[16:20:16 CET] <mc_ebu> BtbN: I assume I can find the equivalent of -"af resample=32000:0:2" for mplayer
[16:20:56 CET] <BtbN> What does the :0:0 part do?
[16:20:59 CET] <BtbN> :0:2
[16:21:18 CET] <mc_ebu> BtbN: it specifies the kind of audio resampling (polyphase) and the quality
[16:21:35 CET] <BtbN> Well, you can just do -ar 32000
[16:21:49 CET] <BtbN> Don't think you can influence too much what it does
[16:22:48 CET] <mc_ebu> BtbN: audio test have show that it's better to use polyphase in floating point
[17:18:37 CET] <DeadSix27> 'in.avs: Unknown error occurred' by running: ffmpeg -i "in.avs" -crf 21.5 -r 120 -f avs -c:v libx264 -movflags faststart -x264opts bframes=0 -profile:v high -pix_fmt yuv420p -preset fast -c:a libopus -b:a 512k "out.mp4"
[17:18:38 CET] <DeadSix27> any idea?
[17:18:48 CET] <DeadSix27> ah prob the movflags
[17:19:07 CET] <DeadSix27> (nvm didnt fix it)
[17:42:56 CET] <DeadSix27> meh after ages i realised i used the wrong binary (which ix 64bit and avs ofc is 32bit)
[17:52:45 CET] <luc4> Hello! I'm trying to understand why ffmpeg is printing "deprecated pixel format used, make sure you did set range correctly". Is this related to the format of the input file or to the output?
[17:55:38 CET] <luc4> If I try to output to png instead of jpg that log is not printed at all, so I suppose it is related to the output.... but what's wrong about jpg?
[18:05:18 CET] <DeadSix27> another issue, probably avs related, im unsure. But ffmpeg -i test.avs -c:a flac out.mkv, makes the video slowmotion anyone ever came across this and has an advice?
[18:22:44 CET] <DeadSix27> uh this isnt even a problem with avs hm
[18:34:42 CET] <DeadSix27> so ye, my videos turn out slowmotion once i run ffmpeg over them, weird. the fps cant be it, source is 120 and output is 120 as well, tried 60fps source to 60fps output, same issue.
[18:35:00 CET] <DeadSix27> command line is basic in.mp4 out.mkv
[18:41:39 CET] <Bigpet> does ffmpeg compile fine on the Visual Studio 2015 CTP6 for anyone?
[18:42:03 CET] <Bigpet> I'm getting some configuration errors, I suspect the C runtime changes
[18:45:35 CET] <Bigpet> well looks like my first error was unrelated (the good ol' used MSYS and forgot to rename link.exe)
[19:07:40 CET] <Bigpet> is that "./configure" hand-written?
[19:47:28 CET] <Bigpet> ok, do I need to use the mailing list for pull-requests, or can I just make them on github?
[19:48:18 CET] <c_14> You can make them on github, but it's preferred if you do it on the mailing list. And if you do make a pull request on github also send the patches to the ml, that way more people can review it.
[19:48:43 CET] <Bigpet> it's a really trivial change, but alright
[19:51:27 CET] <TSM> i have the following stream probe info - Stream #0:1(eng): Video: dvvideo (dvc  / 0x20637664), yuv411p, 720x480 [SAR 8:9 DAR 4:3], 23016 kb/s, SAR 10:11 DAR 15:11, 23.98 fps, 23.98 tbr, 24k tbn, 29.97 tbc (default)
[19:51:37 CET] <TSM> what would you expect the output to be 4:3?
[19:53:18 CET] <kepstin-laptop> TSM: huh, that's a bit odd. 720x480 with SAR 10:11 is the more-or-less standard NSTC 4:3 setting.
[19:53:35 CET] <kepstin-laptop> (it's not exactly 4:3, because 720x480 is overscanned a bit)
[19:54:39 CET] <TSM> i am finding that lots of media has a DAR of 4:3 but would always look better in a 16:9 aspect, I can only conclude that the editors have failed to set the correct DAR info on the output file when they have supplied this to me, is that a possability
[19:55:09 CET] <kepstin-laptop> well, the DAR shouldn't be 4:3 or 16:9 on a 720x480 video, that would give wrong results.
[19:55:31 CET] <kepstin-laptop> instead, SAR should be set to give the correct rectangular pixel shape, then the player figures out the final size
[19:55:43 CET] <TSM> I have being doing my nut in to try and standarise video when converting it to mp4 previews so that it looks correct in the browser
[19:56:15 CET] <TSM> ok so I may not be taking into account the SAR value in the conversion
[19:56:34 CET] <TSM> also why are there two SAR/DAR values in that stream segment
[19:57:03 CET] <kepstin-laptop> TSM: probably one is the video stream's sar, the other is the container sar
[19:57:09 CET] <kepstin-laptop> the container one should be overriding the video one
[19:57:53 CET] <kepstin-laptop> TSM: for ntsc video, non-widescreen should be 10:11 SAR, widescreen should be 40:33 SAR in most cases.
[19:58:44 CET] <kepstin-laptop> (it's fairly common when doing digital video to crop to 704:480 to remove black bars on the sides; but the sar should remain the same if you do that)
[19:59:39 CET] <kepstin-laptop> pal is another story, and apparently people can't agree on a single value :)
[20:00:14 CET] <TSM> how annoying
[20:17:56 CET] <j5k> can anyone point me to a decent read on streaming over LAN with or without ffserver?
[21:34:39 CET] <SQUelcher> hello
[21:38:05 CET] <SQUelcher> I'm trying to stream aac_he encoded audio to an icecast server, it seems that -f adts and/or -content_type audio/aacp is not the right choice for it.
[21:38:27 CET] <SQUelcher> vlc as client yells "Multiple blocks per frame in ADTS not supported"
[21:38:47 CET] <SQUelcher> gstreamer "No caps found for contents within an ICY stream"
[22:30:40 CET] <rjp421> using ffmpeg 2.5 on linux, when transcoding an single stream to multiple bitrate rtmp streams, does the order of the outputs in the cmd line *need* to have the output which just copies the original stream set to the first output in the cmd line? or how can i put the copy of the input as the last output to be created, without forcing a reencode? i need the outputs to be created in order, from lowest to highest bitrates, using specific stream names for each o
[22:30:40 CET] <rjp421> utput.
[22:31:57 CET] <rjp421> i read if i tried using -c:v copy -c:a copy after transcoding, it would use a copy of the previously defined output in the cmd line
[22:32:30 CET] <rjp421> which would be a transcoded and lower quality stream than the source
[23:52:57 CET] <c_14> rjp421: as long as you use -map 0:v -c:v copy it should be fine
[23:55:03 CET] <rjp421> c_14, ok cool ty :)
[00:00:00 CET] --- Sat Mar  7 2015


More information about the Ffmpeg-devel-irc mailing list