[Ffmpeg-devel-irc] ffmpeg.log.20181214

burek burek021 at gmail.com
Sat Dec 15 03:05:03 EET 2018


[01:34:04 CET] <barg> where is the public log for the channel?
[01:38:02 CET] <pink_mist> barg: from https://ffmpeg.org/contact.html there's a link to https://lists.ffmpeg.org/pipermail/ffmpeg-devel-irc/
[02:50:57 CET] <Keshl> OBS guys claim that OBS doesn't mangle things to adjust for the audio streams. Perplexing.
[02:51:20 CET] <Keshl> Not that this means you guys need to do anything now; just an FYI.
[06:44:39 CET] <late_in_the_day> Gday, I'm on linux. I'd like to automate video capture from the webcam and audio from several jack channels . I can't specify which jack connections to make in ffmpeg, so i need to use jack_connect. Although I cant write a script because ffmpeg wont background with &.  Any ideas?
[08:46:08 CET] <Keshl> late_in_the_day: You need to include -nostdin    . This is because ffmpeg takes interactive commands, and that flag disables them.
[08:46:24 CET] <Keshl> Beyond that, everything else is normal.
[08:47:31 CET] <Keshl> Also consider "-loglevel fatal" to make output quieter. -nostats too, if it's still too much.[1]    ([1] https://unix.stackexchange.com/questions/222847/how-to-stream-with-ffmpeg-in-a-separate-process )
[09:06:39 CET] <TheAMM> Good citation
[10:45:50 CET] <POPEYE> How can I get constant bit rate using ffmpeg (avconv) -b:a (-ab) and -b:v (-vb) doesn't help; avprobe still says that bitrate is variable, while Kdenlive (uses avconv) makes constant bitrate with -ab and -vb in it's profile
[10:46:27 CET] <ServiceRobot> hello?
[10:49:41 CET] <ServiceRobot> so I'm doing a personal project using ffmpeg, but I'm having a bit of trouble decoding video. I'm not sure if this is a good place to ask for help
[10:51:51 CET] <JEEB> yes, for the APIs this is the correct place
[10:52:02 CET] <JEEB> together with the libav-user mailing list on ffmpeg.org
[10:52:35 CET] <ServiceRobot> I like talking directly with people through chat more though. this place seems good then
[10:53:24 CET] <ServiceRobot> so I'm using a library that accesses ffmpeg's C code directly
[10:54:02 CET] <ServiceRobot> the problem I'm encountering however is an error directly from ffmpeg itself. I know, because the error code returned is right on their documentation
[10:54:28 CET] <ServiceRobot> but, I'm not exactly sure what the cause is. I should go into further details
[10:54:59 CET] <ServiceRobot> the error is: invalid data found when processing input (-1094995529)
[10:55:49 CET] <ServiceRobot> that negative number translates to INDA in hexidecimal
[10:56:01 CET] <ServiceRobot> #define AVERROR_INVALIDDATA        FFERRTAG( 'I','N','D','A') ///< Invalid data found when processing input
[10:56:19 CET] <ServiceRobot> but I have no idea what is giving me this error
[10:58:24 CET] <ServiceRobot> the documentation on this error is quite sparse. I can't find causes of it, how to prevent it, etc
[11:06:22 CET] <th3_v0ice> Why would a local UDP connection drop often?
[11:07:11 CET] <ServiceRobot> UDP is an unreliable protocol. it just sends data out and hopes it gets there. no checks on the other end
[11:07:30 CET] <ServiceRobot> in fact, that's what I'm using to send video and audio packets across the network
[11:09:20 CET] <JEEB> look at the doc/examples part
[11:09:28 CET] <JEEB> also make sure that whatever you're doing goes fast enough
[11:09:38 CET] <JEEB> I recommend you test with AVFormatContext reading from a file first
[11:09:51 CET] <ServiceRobot> me, or the other guy?
[11:10:01 CET] <JEEB> you're the only API user here active atm ;)
[11:10:42 CET] <ServiceRobot> well I'm using a library that uses ffmpeg. what I want to know is, if I send video and audio data across the network, what needs to be known to ffmpeg on the other end to play the video and audio back?
[11:11:09 CET] <ServiceRobot> I know height, width, and pixel format are needed for video
[11:11:19 CET] <ServiceRobot> but that's not enough apparently
[11:11:37 CET] <JEEB> ok, so you have another abstraction level in the middle?
[11:12:01 CET] <ServiceRobot> yes. it's written in another language, but it does use ffmpeg's library
[11:12:23 CET] <JEEB> that really doesn't matter as long as you're the one poking at the AVFormat/AVCodec contexts
[11:12:26 CET] <JEEB> etc
[11:12:30 CET] <JEEB> as in, you're in control of that layer
[11:12:43 CET] <JEEB> also for what data is needed, just use a container :P
[11:12:52 CET] <JEEB> one that sends out headers periodically
[11:12:57 CET] <ServiceRobot> that's what I'm worried about. that I'm not. or I'm not setting them up properly. I'm not exactly an expert at C
[11:13:03 CET] <JEEB> see doc/examples
[11:13:28 CET] <ServiceRobot> wait, you mean just send the entire container?
[11:13:29 CET] <JEEB> it has various examples. some are somewhat out-of-date, but they should still work
[11:13:48 CET] <JEEB> anyways, $dayjob time :P
[11:13:54 CET] <ServiceRobot> that doesn't exactly help since I'm using a different library
[11:14:03 CET] <ServiceRobot> well, an abstraction layer
[11:14:11 CET] <JEEB> but it's *your* abstraction layer
[11:14:23 CET] <ServiceRobot> eeeh, not exactly. it's someone else's
[11:14:25 CET] <JEEB> if it's not then there's nothing for me to do
[11:14:31 CET] <ServiceRobot> oh
[11:15:00 CET] <JEEB> but you could at least yunno check the order of things for various things :P
[11:15:13 CET] <JEEB> then check against what the abstraction layer and you are doing
[11:15:20 CET] <JEEB> but yea, $dayjob
[11:15:35 CET] <ServiceRobot> alright, see ya. I hope someone can enlighten me on this issue more
[11:23:56 CET] <th3_v0ice> ServiceRobot, JEEB: FFmpeg also fails at just reading the stream and sending it to /dev/null. That should be doing the work fast enough and still the stream is constantly disconnecting. My app which is using an API is probably not that fast but is also getting disconnects. Can something else be the culprit of the problem?
[11:28:23 CET] <ServiceRobot> your issue might be different to mine
[11:28:41 CET] <ServiceRobot> I'm getting an error when trying to decode media packets I send across the network
[11:29:09 CET] <ServiceRobot> so either I'm incorrectly instantiating the decoders, or the packets are corrupt
[11:29:18 CET] <ServiceRobot> or I'm not serializating the packets correctly
[11:29:23 CET] <ServiceRobot> no idea
[12:07:59 CET] <Skandalist> Can anybody help with constant bitrate?
[12:10:22 CET] <sur3> Hi I'm trying to use my amd gpu with openmax and compiled openmax into mesa, when doing configure ffmpeg will show me omx accel and h264_omx but when doing make no h264_omx is compiled, what is the problem?
[13:43:28 CET] <d3bug> having an encoding problem with hevc_vaapi... the error encountered is "Impossible to convert between the formats supported by the filter 'graph 0 input from stream 0:0' and the filter 'auto_scaler_0'"  which only seems to happen when I try to crop the picture
[13:44:46 CET] <d3bug> the actual line I am using is: ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format vaapi -i inputfile.mkv -vf crop=1920:800:0:140 -vcodec hevc_vaapi -qp 20 -map_chapters 0 -bf 0 -acodec libvorbis  outputfile.mp4
[13:45:24 CET] <d3bug> it works without the "crop=" part, but as soon as i try to crop it... boom - error
[13:48:04 CET] <d3bug> oh... Lubuntu 18.04 LTS, AMD RX480 GPU
[13:48:46 CET] <d3bug> kernel: 4.15.0-42-generic (buildd at lgw01-amd64-023) (gcc version 7.3.0 (Ubuntu 7.3.0-16ubuntu3))
[13:49:11 CET] <d3bug> any ideas what i am doing wrong here?
[14:19:55 CET] <FishPencil> I found a dynamic library that exports the following: https://paste.ubuntu.com/p/MG6k7yJ6Cg/ I'd like to get the version information along with the license and configuration, but I don't know if any of these functions provide that.
[14:22:42 CET] <FishPencil> I believe I need avcodec_license but I don't see that...
[14:29:41 CET] <furq> FishPencil: that stuff is all in libavutil
[14:30:31 CET] <furq> https://www.ffmpeg.org/doxygen/trunk/group__lavu__ver.html
[14:33:11 CET] <FishPencil> So am I just out of luck if this is all I have?
[15:16:18 CET] <th3_v0ice> What is the default timeout for UDP in ffmpeg?
[15:39:12 CET] <saml> what's cool project idea?
[15:42:51 CET] <hiihiii> hello
[15:42:59 CET] <saml> th3_v0ice, what kind of timeout? how long connection can be left open without no packets?
[15:43:54 CET] <hiihiii> one can do ffmpeg -i input.mp4 -c copy -f matroska output
[15:44:11 CET] <hiihiii> and ffmpeg will know that the output is to be in an mkv container
[15:44:29 CET] <hiihiii> I'm trying to do the same for jpeg/png/gif formats
[15:44:50 CET] <hiihiii> but I'm not sure what to specify for the -f option
[15:45:13 CET] <hiihiii> I checked ffmpeg list of formats but still not sure
[15:45:15 CET] <Mavrik> hiihiii: ffmpeg -formats lists all muxers
[15:47:43 CET] <saml> hiihiii, are you trying to convert a jpeg to mkv?
[15:48:23 CET] <hiihiii> no
[15:48:35 CET] <hiihiii> I'm trying to pipe the output
[15:48:46 CET] <hiihiii> so I need the format for images
[15:48:50 CET] <hiihiii> something like
[15:49:00 CET] <Mavrik> png = "apng", jpeg = "singlejpeg", gif = "gif" ? :)
[15:49:06 CET] <Mavrik> You can also just leave it out
[15:49:11 CET] <hiihiii> ffmpeg -i test.mp4 -c copy -f matroska pipe:1 | gzip -
[15:49:12 CET] <Mavrik> And it'll be autodetected by file extension
[15:49:16 CET] <hiihiii> but for imges
[15:49:29 CET] <saml> image2 ?
[15:49:42 CET] <saml> what are you trying to do at large?
[15:49:48 CET] <saml> gzip an image?
[15:50:40 CET] <hiihiii> no. that would be stupid cuz it's already compressed. I used gzip just as an example
[15:50:57 CET] <hiihiii> I'm trying to pipe the output to another program
[15:51:05 CET] <saml> what does the program expect?
[15:51:14 CET] <saml> bitmap?
[15:51:20 CET] <hiihiii> image data
[15:51:32 CET] <hiihiii> it can be anything as long as it's an image
[15:51:56 CET] <saml> try -f image2 - | yourprogram
[15:54:38 CET] <saml> ffmpeg -i a.png -c copy -f image2 - > b.png   ;   a.png and b.png are identical
[15:58:00 CET] <hiihiii> thanks
[15:58:13 CET] <hiihiii> it seems to work also with gif and png
[15:58:47 CET] <hiihiii> which save me the headache of having to change -f for each format
[16:15:02 CET] <HickHoward> so
[16:15:16 CET] <saml> yeah HickHoward
[16:15:22 CET] <saml> did you finish it?
[16:15:30 CET] <HickHoward> finish what
[16:16:03 CET] <HickHoward> oh yeah i remember saying something along the lines of "i'll upload more samples soon"
[16:21:06 CET] <HickHoward> although i also said i needed some help for some ps2 .int format as used in a few Ubisoft games i have here on my HDD
[16:28:47 CET] <th3_v0ice> saml: Yeah, how long until ffmpeg breaks the connection?
[16:32:53 CET] <HickHoward> right
[16:32:54 CET] <HickHoward> https://www.sendspace.com/file/nek2ig
[16:35:11 CET] <HickHoward> one of these files can work quite well with ffmpeg
[16:35:50 CET] <saml> th3_v0ice, https://www.ffmpeg.org/ffmpeg-all.html#udp  there is timeout option. i don't know the default
[16:37:49 CET] <saml> th3_v0ice, https://github.com/FFmpeg/FFmpeg/blob/c6e1966c1a1a8987de0834462d7a76377eaa76bc/libavformat/udp.c#L723-L735   looking at this, maybe timeout is 0 (no timeout) if no option is set
[16:54:04 CET] <HickHoward> https://www.sendspace.com/filegroup/hZDPKUYZ8jMzHd9LCR07Rw
[16:54:12 CET] <HickHoward> aaaaaaaaaaaaand another one
[16:54:30 CET] <HickHoward> if you vave amy questions about these samples just ask
[18:26:59 CET] <ariyasu> HickHoward you should upload them to the ftp
[18:27:01 CET] <hiihiii> is there a way to format the output filename so that it notes milliseconds or similar
[18:27:16 CET] <ariyasu> also ffmpeg doesn't handle ps2 videos very well
[18:27:18 CET] <hiihiii> -f image2 -strftime 1 "out_%H:%M:%S.jpg" only gets me up till seconds
[18:27:47 CET] <hiihiii> or can I output a random file name?
[18:27:54 CET] <ariyasu> you could generate a timestamp prior to running your ffmpeg command
[18:27:58 CET] <ariyasu> and store it as a var
[18:28:06 CET] <ariyasu> then have your output be that var
[18:28:17 CET] <CoreX> .batman
[18:28:19 CET] <ariyasu> i do that to get unique filenames
[18:28:53 CET] <hiihiii> do you mind if I ask how. I'm a windows dummy
[18:28:58 CET] <ariyasu> sure
[18:29:29 CET] <mixfix41> would there be a way to pipe ffmpeg screen capture to a local system specifically like from a wayland desktop to x11
[18:29:31 CET] <hiihiii> I did try that other syntax $(date +".%m-%d-%Y-%H:%M:%N.mp4")  but it doesn't except it
[18:29:58 CET] <mixfix41> or forward i mean
[18:30:36 CET] <ariyasu> https://paste.ofcode.org/5aFsVTHZg5ykBPHCgvGFy3
[18:30:56 CET] <ariyasu> i do this to generate a YYYYMMDDhhmmss timestamp in a var called %timestamp%
[18:31:10 CET] <ariyasu> then you can just have your output be %timestamp%.mkv or what ever
[18:31:22 CET] <ariyasu> i have some other code if you want a unix timestamp
[18:32:20 CET] <hiihiii> but
[18:32:34 CET] <hiihiii> I'm going to run a script over some images
[18:32:46 CET] <hiihiii> obviously it'll run fast
[18:32:58 CET] <ariyasu> https://paste.ofcode.org/zS9yZYR6jDTMHCfwhEtBuk
[18:33:03 CET] <ariyasu> yeah thats fine
[18:33:11 CET] <ariyasu> just save it as timestamp.bat
[18:33:26 CET] <ariyasu> then do "call timestamp.bat" before each command
[18:33:33 CET] <ariyasu> then it's always updated
[18:34:41 CET] <hiihiii> won't two commands run in less than a second? they'd both have the same timestamp, no?
[18:35:16 CET] <ariyasu> if you think thats a possibility you can add a line like
[18:35:22 CET] <ariyasu> timeout 1
[18:35:32 CET] <ariyasu> and it would wait 1 second before executing the the next command
[18:35:44 CET] <hiihiii> ok thank you.
[18:35:46 CET] <ariyasu> or if you don't care about the time stamp
[18:35:54 CET] <ariyasu> and just want unique name
[18:35:59 CET] <ariyasu> you could do
[18:36:15 CET] <ariyasu> set /a timestamp=%timestamp% + 1
[18:36:33 CET] <ariyasu> and it would add 1 to the number each time and would be quicker than waiting 1 second
[18:38:25 CET] <hiihiii> thanks that will work
[20:09:01 CET] <HickHoward> >you should upload them to the ftp
[20:09:02 CET] <HickHoward> ...how?
[20:17:41 CET] <pink_mist> http://ffmpeg.org/bugreports.html <-- has a "Submitting Sample Media" section
[20:18:41 CET] <HickHoward> cool
[20:21:49 CET] <HickHoward> does that section assume i have some kind of knowledge about...something or...?
[20:31:16 CET] <pink_mist> it assumes you know what ftp is
[20:31:26 CET] <pink_mist> and how to use an ftp client
[20:31:37 CET] <pink_mist> if you don't, you can google tutorials
[20:32:17 CET] <HickHoward> what's that supposed to mean
[20:32:36 CET] <pink_mist> exactly what I said
[20:33:56 CET] <Foaly> ftp is a protocol for tranfering files basically
[20:34:48 CET] <HickHoward> yah i kinda know what an ftp is
[20:38:10 CET] <pink_mist> HickHoward: so what do you think is missing? they mention what details to use when connecting, they mention which directory to upload into, they mention the limits on uploading and how to work around them
[20:38:26 CET] <HickHoward> yes
[20:38:27 CET] <HickHoward> i know
[20:39:06 CET] <HickHoward> and i'll try to learn how to ftp when the need arises
[20:39:15 CET] <HickHoward> which means "not now"
[22:47:08 CET] <monery> test?
[22:47:09 CET] <monery> ah
[22:47:13 CET] <monery> greetings channel
[22:47:19 CET] <monery> forgot about nickserv
[22:47:39 CET] <monery> is anyone free to answer some questions about cuda and ffmpeg on debian?
[22:49:00 CET] <monery> having some trouble getting ffmpeg running on a box I made from scraps to speed up some encoding stuff
[22:51:43 CET] <monery> nvidia-smi reports 384.130 and the ffmpeg is from the debian repo's ffmpeg version 3.2.12-1~deb9u1
[22:52:14 CET] <monery> [h264_nvenc @ 0x55b3c30d5540] Cannot load libnvidia-encode.so.1
[22:52:56 CET] <monery> I am sure I did something stoopid or forgot to install a package.. if anyone knows... if not... I am just screwing around today... I'll keep trying google, but the closest thing I found was from 2015 :(
[22:56:58 CET] <monery> hello stephenwithav
[22:57:21 CET] <stephenwithav> Hello monery
[22:58:01 CET] <monery> do you help with tech issues? I have one involving ffmpeg and cuda that I haven't been able to solve... looking for some insight
[22:59:03 CET] <stephenwithav> Not cuda issues, unfortunately.  I'm here for assistance, too. :)
[22:59:12 CET] <monery> maybe I can help?
[22:59:22 CET] <monery> I know my way around linux and ffmpeg
[22:59:25 CET] <monery> use it daily
[22:59:41 CET] <monery> I'm just a user tho, not part of the channel
[22:59:46 CET] <stephenwithav> give me a sec to phrase my question, ty.
[22:59:55 CET] <monery> no promises tho :)
[23:00:04 CET] <stephenwithav> lol
[23:02:49 CET] <stephenwithav> My question is: ffmpeg v4/libavcodec includes ff_libx264rgb_encoder, but wraps it in an #if CONFIG_LIBX264RGB_ENCODER block.  How do I enable that?
[23:03:42 CET] <kepstin> stephenwithav: that's enabled automatically if the configure script detects that your libx264 supports rgb
[23:04:05 CET] <kepstin> it'll only be disabled if your x264 is too old, basically
[23:05:15 CET] <kepstin> Note that the rgb support is exposed as a separate encoder, you use "-c:v libx264rgb" on the ffmpeg cli instead of "-c:v libx264"
[23:05:16 CET] <stephenwithav> ty, kepstin.  Is there a place to go to upgrade my x264 libs for Debian or Ubuntu?
[23:07:30 CET] <stephenwithav> I'm trying to use libavcodec directly with RGB data, avoiding the sws_scale call if possible.
[23:11:30 CET] <leonardus> Anyone know how this audio effect could have been made? https://leonardus.me/static/output.mp3
[23:11:41 CET] <leonardus> probably involves a low pass filter but not sure what else.
[23:13:13 CET] <kepstin> stephenwithav: are you selecting the right encoder? Like i said, you have to use "libx264rgb"
[23:13:31 CET] <kepstin> the rgb support's been in libx264 for quite a while
[23:14:44 CET] <kepstin> even ubuntu 16.04 has it
[23:16:05 CET] <stephenwithav> I'm doing avcodec_find_encoder_by_name("mpeg4"), creating a context, setting ctx->pix_fmt = AV_PIX_FMT_RGB24, then avcodec_open2(ctx, codec, NULL) fails with error -22.
[23:17:20 CET] <stephenwithav> I'm running 18.04, so I should be good.
[23:28:21 CET] <kepstin> well, that's wrong
[23:28:29 CET] <kepstin> mpeg4 is a completely different codec from h264
[23:28:48 CET] <kepstin> and you need to look up a specific encoder "libx264rgb" by name to get the rgb-capable x264
[23:29:12 CET] <leonardus> isn't mpeg4 a container?
[23:29:42 CET] <stephenwithav> Ah.  ty for educating me.
[23:29:56 CET] <kepstin> here, i assume it refers to mpeg 4 asp, i.e. more or less what xvid/divx were
[23:30:43 CET] <kepstin> the 'mp4' container is something different.
[23:31:01 CET] <kepstin> although both are part of the same overall "MPEG 4" standard
[23:31:08 CET] <kepstin> although so is h264, fwiw...
[23:31:41 CET] <kepstin> mpeg 4 is more like a collection of related standards than a single thing
[23:38:45 CET] <birdboy88> test
[23:49:36 CET] <birdboy88> Hi. Hoping someone could help me with this problem I have. I am recording PC games with NVIDIA GeForce ShadowPlay, which produces videos with variable frame rate. I am trying to make a ffmpeg script that extracts subclips and merges them with 0.5 second crossfade. The problem is that the video/audio is sometimes out of sync. I've been reading that it probably has to do with variable frame rate but I dont know how to fix it. My scri
[23:49:40 CET] <birdboy88> https://pastebin.com/YkASj9A1
[23:50:15 CET] <birdboy88> Its in windows powershell by the way
[23:53:50 CET] <kepstin> birdboy88: in order to keep a/v sync, i'd normally recommend extracting the audio/video together to a single file in a single command
[23:54:16 CET] <kepstin> but yeah, getting crossfade in there would be a pain
[23:55:21 CET] <birdboy88> yes, I think thats the reason we did it in separate steps. I actually got help for most of the script commands
[23:55:48 CET] <birdboy88> I've been reading about -async and -vsync paramenters, but I dont know where to put them
[23:56:17 CET] <birdboy88> Do you recommend me to post this on stackoverflow for example?
[23:57:43 CET] <BtbN> I'd recommend just not to use that software. It's horrible.
[23:58:14 CET] <monery> u can try transcoding to a constant bitrate format
[23:58:23 CET] <monery> I know thats not usually ideal
[23:58:36 CET] <kepstin> constant bitrate is a different thing from constant framerate
[23:59:00 CET] <birdboy88> Its critical to not loose processing speed, so heavy re-encoding must be avoided
[23:59:00 CET] <monery> I thought it was audio that was causing the problem, my appolgise
[23:59:13 CET] <kepstin> but if you're posting to youtube or something, it might make sense to convert to constant framerate anyways, so you could try doing that as a first step before editing :/
[23:59:29 CET] <kepstin> how far out is the a/v sync getting?
[23:59:44 CET] <BtbN> if you do cross-fade, you'll need to re-encode anyway
[00:00:00 CET] --- Sat Dec 15 2018


More information about the Ffmpeg-devel-irc mailing list