[Ffmpeg-devel-irc] ffmpeg.log.20161223

burek burek021 at gmail.com
Sat Dec 24 03:05:01 EET 2016


[00:28:06 CET] <Maverick|MSG> efface: I tried it with -abort_on but no luck there
[00:29:08 CET] <Maverick|MSG> if I do set a timeout ffmpeg won't quit, but it does finish up writing the mp4 and it outputs to the console: "I/O error"
[00:49:12 CET] <teratorn> is sending an ffmpeg process SIGTERM the same as hitting "q" at the terminal?
[00:49:50 CET] <DHE> basically yes
[00:54:22 CET] <teratorn> SIGTERM and SIGKILL do the same thing, it seems... they print a message "Exiting normally..." but just hitting q does not print that message.. hm oh well, good enough for who it is ;)
[00:54:31 CET] <teratorn> err, SIGINT and SIGTERM
[00:58:47 CET] <DHE> yes, SIGKILL is one of two very special signals
[00:58:49 CET] <DHE> :)
[01:12:33 CET] <Mista_D> Encoding video and copying PCM stream  - AV out of sync. If encode both audio and video the sync is ok. MXF container, MPEG2video and PCM codec.
[03:13:00 CET] <madprops> hi i usually use something like "ffmpeg -i RJnUmRaFx.mp4 -crf 25 tost.mp4"
[03:13:27 CET] <madprops> can I add something to it so there's almost 0% chance of encoding fails (corruptions, no video, no audio, etc) ?
[03:13:44 CET] <madprops> input files would be either mp4 or webm
[03:14:12 CET] <DHE> ffmpeg is pretty good. if one of those things happens, it's the input file's fault
[03:14:25 CET] <k_sze[work]> I'm trying to build ffmpeg from source on an nVidia Jetson TK1 (it's an armv71), but I get "GNU assembler not found, install/update gas-preprocessor" from the configure script.
[03:14:28 CET] <DHE> but realistically you should specify the codecs you want, at a minimum
[03:14:58 CET] <madprops> DHE, like x264?
[03:15:08 CET] <DHE> also the audio codec
[03:15:33 CET] <madprops> which one do you suggest for audio?
[03:15:59 CET] <madprops> it's for displaying videos in a browser
[03:25:29 CET] <madprops> i would think mp3 is more compatible
[03:25:38 CET] <madprops> but im seeing it's maybe aac which is more compatible
[07:37:06 CET] <beauty> hello
[07:38:06 CET] <beauty> how to use custom IO?
[07:38:48 CET] <beauty> After I use custom IO, will lack five frame.
[07:43:09 CET] <fahadash> Can we have line breaks in filter complex script?
[07:54:46 CET] <nonex86> beauty: i am using custom io in my software without a problem
[08:02:26 CET] <beauty> nonex86: could you teach me ?
[08:02:33 CET] <beauty> hello
[08:03:49 CET] <nonex86> beauty: that depends what do you mean by saying "teach me" :)
[08:49:57 CET] <Diljeet> hello
[08:50:23 CET] <Diljeet> i am here to ask a question to anyone available here
[08:51:50 CET] <Diljeet> I want to ask, can i use a "C" class from ffmpeg source and convert it to "JAVA" and use it in commercially, i will share all modified class with eveyone, ami permitted to do that
[08:53:16 CET] <Diljeet> I want to ask, can i use a "C" class from ffmpeg source and convert it to "JAVA" and use it in commercial app, i will share all modified class with everyone, am i permitted to do that
[10:18:00 CET] Last message repeated 1 time(s).
[10:28:09 CET] <ikevin> Diljeet, you'll find your reply here: https://www.ffmpeg.org/legal.html
[10:30:35 CET] <Diljeet> Thanks i am reading it
[11:18:15 CET] <thebombzen> "This channel is publicly logged" - where can I find the logs?
[11:28:37 CET] <thebombzen> also
[11:28:53 CET] <thebombzen> the wiki lists libvorbis as being >= libfdk_aac. when did this happen? when did libvorbis get better than AAC?
[12:02:32 CET] <ehsanv> hi, i want to add a logo (use overlay) to my videos, and want my logo to be exactly 1/4 width of input videos (input videos have different widths). how can i do this
[12:50:25 CET] <durandal_1707> thebombzen: on ffmpeg.org web page
[14:09:52 CET] <nyuszika7h> how can I get ffmpeg to take the Display Aspect Ratio into account when taking a snapshot with "-frames:v 1"?
[14:10:23 CET] <nyuszika7h> well, it seems "-vf scale=1024:576" works
[14:16:31 CET] <nonex86> simple scale will not preserve aspect ratio
[14:16:59 CET] <nonex86> you should said this directly and perhaps pad the result image
[14:18:01 CET] <nonex86> read about force_original_aspect_ratio and pad
[14:18:45 CET] <nyuszika7h> I'll take a look
[14:18:49 CET] <nyuszika7h> the resolution is 720x576 and corrected to 1024x576 with display aspect ratio (a friend captured this from TV)
[14:20:59 CET] <nonex86> i assume you have SAR 1:1?
[14:21:39 CET] <nyuszika7h>     Stream #0:0: Video: h264 (Main), yuv420p, 720x576 [SAR 64:45 DAR 16:9], 25 fps, 25 tbr, 1k tbn, 50 tbc (default)
[14:22:17 CET] <nonex86> and the result SAR?
[14:23:00 CET] <nyuszika7h> the thumbnail?
[14:23:02 CET] <nyuszika7h> eam #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 720x576 [SAR 64:45 DAR 16:9], 25 tbr, 25 tbn, 25 tbc
[14:23:11 CET] <nyuszika7h> Str^ even
[14:23:32 CET] <nonex86> i dont understand, you said 720x576->1024x567
[14:23:34 CET] <nonex86> isnt it?
[14:23:39 CET] <nyuszika7h> 576 not 567
[14:23:56 CET] <nonex86> ok, lets correct
[14:23:59 CET] <nonex86> i dont understand, you said 720x576->1024x576
[14:24:00 CET] <nyuszika7h> I mean on playback it's 1024x576 because of the DAR
[14:24:10 CET] <nonex86> :)
[14:24:50 CET] <nyuszika7h> with some players if I take a snapshot it ends up as 1024x576, with other players and with ffmpeg it's 720x576 and looks stretched
[14:25:32 CET] <nonex86> so, as was said before
[14:25:37 CET] <nonex86> simple scale doesnt work
[14:25:39 CET] <nonex86> isnt it?
[14:25:50 CET] <nyuszika7h> it does work if I just scale when taking the thumbnail
[14:25:57 CET] <nyuszika7h> the resulting thumbnail looks right
[14:26:21 CET] <nyuszika7h> (I tried that after asking the question)
[19:15:16 CET] <thebombzen_> I don't remember if I got an answer for this before, but... why does -map_metadata -1 work?
[19:15:54 CET] <thebombzen_> if -map_metadata behaves like -map, then -map_metadata -1 would mean "discard metadata from the first input"
[19:16:05 CET] <thebombzen_> so doesn't that mean that you should use -map_metadata -0
[19:20:05 CET] <thebombzen_> another thing to note: -map_metadata -1 is not mentioned in: https://ffmpeg.org/ffmpeg.html#Advanced-options
[19:20:35 CET] <thebombzen_> at least not under the section documenting -map_metadata. maybe that's why so many people ask that question
[20:01:22 CET] <kerio> thebombzen: is metadata taken from inputs that are not the first, tho?
[22:51:55 CET] <mosb3rg> hey there, i have a question about the sirius xm style HLS AAC delivery and ripping with there cookie system
[22:52:57 CET] <mosb3rg> unlike what im customaryly used to seeing with the set cookie: value im seeing a few set cookie values coming from them even in the same playlist etc. could i provide an example and get some feed back on how ffmpeg can parse this set cookie value and correctly dump from the system ?
[23:15:31 CET] <blue_misfit> hey folks, if I have an anamorphic 720x480 source and I want ffmpeg to make it square pixels, and specify the output height but let ffmpeg calculate the appropriate square pixel width, how would I do that?
[23:16:31 CET] <blue_misfit> like for a 720x480 source with a 16x9 DAR I want to do something like -vf scale=w=-2:h=360 and have ffmpeg make a 640x360 output
[23:16:57 CET] <blue_misfit> and for a 720x480 source with a 4x3 DAR I want that same command to create a 480x360 output
[23:25:42 CET] <markvandenborre> blue_misfit: sorry, no idea
[23:25:52 CET] <blue_misfit> I'm getting pretty close
[23:25:57 CET] <blue_misfit> -vf setdar=16:9,scale=w=360*dar:h=360,setsar=1
[23:26:07 CET] <blue_misfit> -vf setdar=4:3,scale=w=360*dar:h=360,setsar=1
[23:26:10 CET] <blue_misfit> that seems to do it
[23:26:18 CET] <markvandenborre> ah, congratulations
[23:26:40 CET] <blue_misfit> I have to force in the beginning, but that's because my sources have slighly screwy DARs so there's some fuzzy logic upstream :)
[23:26:45 CET] <blue_misfit> closeenough.jpg
[23:27:49 CET] <markvandenborre> blue_misfit: do you happen to have any idea how I could solve my problem:
[23:28:03 CET] <markvandenborre> I have ffmpeg output to a udp multicast address
[23:28:22 CET] <markvandenborre> (a h264 encoded video stream, and an aac encoded audio stream)
[23:28:44 CET] <markvandenborre> I can play that stream using mpv
[23:28:48 CET] <markvandenborre> but not using ffplay
[23:29:30 CET] <markvandenborre> actually, scrap that, I can play that stream using ffplay
[23:30:08 CET] <markvandenborre> the problem is when I want to read from that udp multicast source
[23:30:29 CET] <markvandenborre> and send it out as an flv with codec copy
[23:30:36 CET] <markvandenborre> as an rtmp stream basicly
[23:30:40 CET] <markvandenborre> it refuses to
[23:31:12 CET] <markvandenborre> because it gets confused about the video stream in the udp multicast (mpeg ts)
[23:32:25 CET] <blue_misfit> hmmm
[23:32:32 CET] <markvandenborre> it says "could not find codec parameters for stream 0 (Video: h264 ([27] [0] [0] / 0x001B), none): unspecified size
[23:32:51 CET] <markvandenborre> Consider increasing the value for the 'analyzeduration' and 'probesize' options
[23:32:57 CET] <blue_misfit> let me see if I can replicate
[23:33:00 CET] <markvandenborre> but those are already ridiculously high
[23:33:07 CET] <blue_misfit> so make a multicast TS using h264 and aac is the first step right?
[23:33:18 CET] <markvandenborre> what I am using is
[23:33:34 CET] <markvandenborre> ffmpeg -re -i testvid/testvid.mp4 -c copy -f mpegts udp://227.1.3.1:9000
[23:33:46 CET] <markvandenborre> to create the multicast stream
[23:34:57 CET] <mosb3rg> hey gents anyone around with experience using cookies with HLS streams ?
[23:35:37 CET] <markvandenborre> mosb3rg: I've played with HLS streams
[23:35:43 CET] <markvandenborre> but the cookie thing, not really
[23:36:49 CET] <markvandenborre> mosb3rg: have a look at stuff like http://stackoverflow.com/questions/2936329/nginx-redirect-if-cookie-present
[23:36:52 CET] <markvandenborre> ?
[23:36:52 CET] <blue_misfit> ok so I have the multicast stream working markvandenborre
[23:37:01 CET] <blue_misfit> what's next
[23:37:18 CET] <blue_misfit> you want to rewrap to flv and deliver via rtmp?
[23:37:22 CET] <blue_misfit> or save as an flv file
[23:37:27 CET] <markvandenborre> sec
[23:38:36 CET] <markvandenborre> hm, I skipped a crucial step after a long day, sorry...
[23:38:47 CET] <blue_misfit> all good! Glad you sorted it out
[23:39:08 CET] <markvandenborre> nope, didn't
[23:39:21 CET] <markvandenborre> but in showing you what I was trying to do with the multicast...
[23:39:51 CET] <markvandenborre> https://github.com/FOSDEM/video/blob/master/software/ansible/playbooks/roles/voctop/files/scripts/sink-output.sh
[23:39:59 CET] <markvandenborre> is actually what prepares the output
[23:41:02 CET] <markvandenborre> so I guess you could replace the first bit by ffmpeg -y -re -i testvid.mp4
[23:41:40 CET] <markvandenborre> then keep the rest from "-threads:0 0 \"
[23:42:17 CET] <markvandenborre> and replace the ${MULTICAST_SINK} in the last line with something sensible, like udp://227.1.3.3:9000
[23:42:44 CET] <markvandenborre> let me write it out for you
[23:42:54 CET] <markvandenborre> sec
[23:44:39 CET] <markvandenborre> blue_misfit: something like http://pastie.org/10985370
[23:44:54 CET] <markvandenborre> should give you a fairly good replication of the problem I'm bumping into
[23:45:05 CET] <markvandenborre> let me try it on my machine to confirm too
[23:49:54 CET] Action: markvandenborre is trying if this yields the same error message
[00:00:00 CET] --- Sat Dec 24 2016


More information about the Ffmpeg-devel-irc mailing list