[Ffmpeg-devel-irc] ffmpeg.log.20180209
burek
burek021 at gmail.com
Sat Feb 10 03:05:01 EET 2018
[00:01:07 CET] <Guest95076> guys dont you think irc is like 1995\
[00:02:01 CET] <DHE> oh IRC is older than that. I have 1993
[00:02:30 CET] <durandal_1707> stay ontopic or be banned
[00:02:41 CET] <tdr> 1988
[00:06:54 CET] <Guest95076> StAy On ToPiC oR bE bAnNeD
[00:21:58 CET] <Guest34615> asdf
[00:22:00 CET] <Guest34615> sup niggas
[00:22:12 CET] <Guest34615> im back bitches
[00:22:21 CET] <DHE> iive: hate to annoy you...
[00:22:46 CET] <Guest34615> wat
[00:23:11 CET] <Guest34615> jadsf
[00:23:12 CET] <Guest34615> asdf
[00:23:13 CET] Last message repeated 1 time(s).
[00:23:13 CET] <Guest34615> asd
[00:23:13 CET] <Guest34615> fasd
[00:26:33 CET] <iive> DHE, don't worry, always notify me.
[00:34:06 CET] <fjsanchez> Hi
[00:36:32 CET] <fjsanchez> I have compiled release/3.4 on Ubutu 16.04, I have used the same configure flags as the package on mc3man/xerus-media and it compiled properly after installing all the dependencies. The problem that I have is that I am not able to output with format v4l2, I am using v4l2loopback and that was working with the prepackaged build...
[00:37:18 CET] <fjsanchez> When I check ffmpeg -formats it has v4l2 as emuxer
[00:38:57 CET] <fjsanchez> I just get [NULL @ 0x55832290c340] Requested output format 'v4l2' is not a suitable output format /dev/video0: Invalid argument
[00:38:59 CET] <iive> fjsanchez, I have " E v4l2 Video4Linux2 output device" in `ffmpeg -formats`
[00:39:22 CET] <fjsanchez> that is what I have iive
[00:39:53 CET] <fjsanchez> I have recompiled even using --enable-v4l2 and I have also "D video4linux2,v4l2 Video4Linux2 device grab"
[00:40:06 CET] <fjsanchez> I mean with --enable-libv4l2
[00:40:10 CET] <iive> ok, it might be problem with the color format. e.g. if you convert from rgb, you might end up with yuv444
[00:40:45 CET] <fjsanchez> I am trying this ffmpeg -video_size 640x480 -framerate 10 -f x11grab -i :1+100,200 -f v4l2 /dev/video0
[00:40:59 CET] <iive> see if there is an example using pix-fmt
[00:41:23 CET] <iive> grab is definitely rgb.
[00:41:34 CET] <fjsanchez> the packaged version was working with this: ffmpeg -re -video_size 640x480 -framerate 10 -f x11grab -i :1+100,200 -pix_fmt gray16le -vf noise=alls=20:allf=u+t -f v4l2 -threads 1 /dev/video0
[00:42:24 CET] <fjsanchez> The problem is that I am interested in gray10le, but the code currently doesn't support it, so my idea was to add support but I cannot even get workign the code without changes...
[00:43:06 CET] <iive> so the above line doesn't work anymore ?
[00:43:15 CET] <fjsanchez> no, just the same error
[00:44:04 CET] <fjsanchez> before that I had an error saying that not suitable pix conversion was available, which makes sense as it is not defined a conversion from gray10le for V4L2 in the ffmpeg code
[00:44:29 CET] <fjsanchez> but now it just says that is not a suitable output format
[00:44:41 CET] <iive> i haven't used v4l2 loopback before
[00:45:34 CET] <fjsanchez> If I run this: ffmpeg -re -video_size 640x480 -framerate 10 -f x11grab -i :1+100,200 -pix_fmt gray16le -vf noise=alls=20:allf=u+t -f v4l2 -threads 1 /dev/video0
[00:45:39 CET] <fjsanchez> everything works...
[00:45:55 CET] <iive> new build and old packaged?
[00:46:15 CET] <fjsanchez> if I run the version in /usr/bin it works
[00:46:34 CET] <fjsanchez> if I run the version I built in /usr/local/bin it doesn't
[00:47:55 CET] <iive> is your build using static libraries? it might be slight problem if you have similar libraries installed on 2 different locations
[00:48:00 CET] <fjsanchez> Here you can see the output for ffmpeg for each build: https://pastebin.com/aF5LJLth
[00:48:21 CET] <fjsanchez> I think I am not using static
[00:48:26 CET] <fjsanchez> At least not intentionally
[00:49:39 CET] <fjsanchez> I think I might not be using the same version of the source code...
[00:49:56 CET] <iive> it seems your build is using a git version from this year
[00:50:20 CET] <fjsanchez> yes, I just made the clone today
[00:50:51 CET] <fjsanchez> I am going to try to checkout to that tag
[00:51:41 CET] <iive> 3.4 should be in its own branch
[00:51:59 CET] <iive> so you can make 3.4.1 from it.
[00:52:19 CET] <fjsanchez> I am building from the release/3.4 branch
[00:52:32 CET] <fjsanchez> git branch master * release/3.4
[00:53:22 CET] <fjsanchez> what are the nXXXX tags on GIT?
[00:53:33 CET] <fjsanchez> there is n3.4.1
[00:55:56 CET] <iive> yes, 3.4.1 is the latest release.
[00:58:31 CET] <fjsanchez> Ok, I am going to try to compile this one
[01:03:15 CET] <fjsanchez> Nah, just the same version: ffmpeg version N-89978-gcb97400
[01:04:18 CET] <fjsanchez> or maybe not...
[01:07:58 CET] <fjsanchez> According to the repo the version that I can install in Ubuntu from PPA was uploaded by Doug McMahon (2018-02-03)
[01:08:05 CET] <fjsanchez> quite recently
[01:13:51 CET] <fjsanchez> Nothing, is not working...
[01:20:17 CET] <fjsanchez> I got it working! I think I just forgot to install properly last time I compiled...
[01:20:25 CET] <fjsanchez> So it was still calling the previous bin...
[01:26:40 CET] <ZeroWalker> can ffmpeg output raw binary?
[01:27:09 CET] <ZeroWalker> as in, pixel array
[01:29:48 CET] <furq> yes
[01:29:49 CET] <furq> rawvideo
[01:30:13 CET] <ZeroWalker> does that work on an image as well?
[01:39:18 CET] <ZeroWalker> nevermind, found a solution, thanks
[01:42:51 CET] <furq> rawvideo works fine for images
[01:47:46 CET] <kerio> an image is just a very short video
[01:53:30 CET] <furq> depends how long you look at it for
[02:21:43 CET] <n0xlf> Pulling my hair out with this one: https://pastebin.com/B3GtrTbw
[02:22:42 CET] <n0xlf> Trying to put 3 vids side by side with combined audio and different starting positions for 2. I made the height the same for all 3 (required for hstack) and converted all audio to mono flac. It seems to not like something with the audio/amerge ...
[02:23:02 CET] <n0xlf> Any ideas?
[02:24:31 CET] <n0xlf> "Failed to inject frame into filter network: Cannot allocate memory" seems to actually translate to "you did something wrong, but I won't tell you what"...
[02:25:15 CET] <kepstin> hmm, amerge probably doesn't like streams with widely varying start times
[02:25:18 CET] <furq> i take it you're not actually running out of memory
[02:25:31 CET] <kepstin> that said, you probably want to be using amix, not amerge?
[02:25:33 CET] <n0xlf> No - Have 48GB free...
[02:25:53 CET] <furq> yeah amerge with three mono inputs will create an output with three channels
[02:25:55 CET] <n0xlf> Possibly - I played around with it a bit but didn't quite understand the syntax. Do you have a suggestion for that one?
[02:26:36 CET] <furq> just amix=3
[02:26:38 CET] <n0xlf> I thought "-ac 2" might make (duplicate) stereo..
[02:27:40 CET] <furq> it'll mix down to 2 channels
[02:27:50 CET] <furq> i'm not entirely sure what kind of results you'll get with a 3-channel input
[02:28:02 CET] <furq> idk how ffmpeg would even interpret that, presumably fl, fr and lfe
[02:28:16 CET] <kepstin> btw, why did you convert the input audio to mono? you can mix together stereo sources fine.
[02:29:02 CET] <n0xlf> One was stereo and two were mono...
[02:29:10 CET] <n0xlf> Was just part of testing...Tryting to get everything the same
[02:29:29 CET] <furq> amix would work fine with that but it's probably better to have them all mono
[02:29:35 CET] <n0xlf> Changing to amix is actually doing something! Woo hoo! We'll see what...
[02:30:36 CET] <furq> i wonder if that was because -ac 2 was interpreting the third channel as LFE but it contained high frequencies
[02:30:57 CET] <n0xlf> I tried it without -ac 2 and amerge and same error...
[02:31:05 CET] <furq> oh
[02:32:01 CET] <n0xlf> I shouldn't get too excited - 8 CPUs 100% and:
[02:32:02 CET] <n0xlf> frame= 0 fps=0.0 q=0.0 size= 0kB time=00:00:00.14 bitrate= 0.0kbits/s speed=0.00143x
[02:33:31 CET] <n0xlf> Although it did just spit out:
[02:33:31 CET] <n0xlf> [swscaler @ 0x55b7bbc50ae0] deprecated pixel format used, make sure you did set range correctly [Parsed_hstack_1 @ 0x55b7b8ff5860] [framesync @ 0x55b7b8316888] Buffer queue overflow, dropping. Last message repeated 4 times [Parsed_hstack_1 @ 0x55b7b8ff5860] [framesync @ 0x55b7b8316888] Buffer queue overflow, dropping. Last message repeated 95 times [Parsed_hstack_1 @ 0x55b7b8ff5860] [framesync @ 0x55b7b8316888] Buffer queue ove
[02:34:13 CET] <n0xlf> Should I drop -ac 2 with amix or will that produce the desired 2 channel output?
[02:35:33 CET] <furq> it'll give a mono output
[02:35:35 CET] <n0xlf> Other than it throwing thousands of errors/sec. it is outputting to testvideo.mp4, so it's a start:
[02:35:36 CET] <n0xlf> [Parsed_hstack_1 @ 0x55b7b8ff5860] [framesync @ 0x55b7b8316888] Buffer queue overflow, dropping.=0 speed=7.49x
[02:35:49 CET] <n0xlf> Mono, but both right and left channels, correct?
[02:35:57 CET] <furq> sure
[02:36:39 CET] <furq> why are you overlaying this btw
[02:45:01 CET] <n0xlf> Good point - That was left over from when I thought I'd want a black background, but I won't need it
[02:46:34 CET] <n0xlf> I think "+" on itsoffset isn't valid as well - Might be nothing for + and - for -...
[03:26:35 CET] <n0xlf> Thanks for the help @furq! Gives me something to work with now at least...
[03:28:06 CET] <n0xlf> Going to let it chew on this for a while and see how it goes: ffmpeg -i a1.mkv -itsoffset 208 -i p1.mkv -itsoffset 2025 -i p2.mkv -filter_complex "hstack=inputs=3;amix=inputs=3" -c:v libx264 -preset veryslow -profile:v high -level 4.1 -crf 23 -c:a aac -b:a 128k -ac 2 video.mp4
[03:48:47 CET] <kinkinkijkin> how do I set up audio for ffmpeg stdin? I just want to send a raw stream of audio to ffmpeg via stdin for ffmpeg to turn into a file for me
[03:53:37 CET] <fella> cat audio.raw | ffmpeg -i - -c:a libmp3lame out.mp3
[03:54:37 CET] <kinkinkijkin> no, I am writing a program to generate audio and I'm wondering how to set it up for ffmpeg, I don't know what metadata I need or such
[03:55:12 CET] <kinkinkijkin> I know how to use stdin
[03:56:41 CET] <furq> you don't need any metadata
[03:56:54 CET] <furq> just write pcm frames and then tell ffmpeg what the format is
[03:57:00 CET] <kinkinkijkin> oki
[03:57:09 CET] <kinkinkijkin> thank you
[03:57:15 CET] <furq> -f s16le -i -
[03:57:17 CET] <furq> or whatever
[03:58:08 CET] <furq> you can obviously write a wav header or something first but you don't actually need to
[08:18:45 CET] <user128> hi, it seems that opus is not really supported in a mp4 container. is there a container which supports both, opus stream and "chapters"?
[08:21:31 CET] <iranen> mkv?
[08:32:00 CET] <user128> iranen, I will check mkv out. I had only seen chapters with m4b / m4a files before. thanks!
[15:48:09 CET] <zerodefect> I'm using the C-API with the overlay filter; I'm having a tough time getting it to work as I envisaged. I'm able to composite an image from the overlay onto the main. The problem I'm having is with the size. I don't know the size of the image until I need to add it to the graph (and it may change size). So what happens...
[15:49:22 CET] <zerodefect> is that the buffersrc for the overlay is initialized to 720x576. Then let's say I get a 250x250px image to composite. I tell the buffersrc of the new size, and then add the 250x250 image to the graph. Unfortunately, what seems to happen is that the image is stretched over the entire width/height of the main.
[15:50:07 CET] <zerodefect> I appreciate that not all filters handle dynamic changes. I can't quite determine if the overlay can/should or if its me doing something wrong.
[15:51:08 CET] <zerodefect> I've been trying to look at the src to try understand but no cigar. https://github.com/FFmpeg/FFmpeg/blob/master/libavfilter/vf_overlay.c
[15:52:45 CET] <durandal_1707> zerodefect: you cant change w/h after setup of buffersrc
[15:53:47 CET] <zerodefect> Even though this function is available: http://ffmpeg.org/doxygen/trunk/group__lavfi__buffersrc.html#ga398cd2a84f8b4a588197ab9d90135048 ?
[15:55:28 CET] <durandal_1707> zerodefect: once filtergraph is set you can not generally change w/h
[16:19:30 CET] <Maxz> hi, is there a libavformat hls maintainer here?
[17:52:54 CET] <saml> is there term for encoding spec such as width,height,framerate,codec,bitrate, ...
[17:56:51 CET] <iive> technically, it is meta-data, since it describes the "real" data.
[17:57:36 CET] <iive> sometimes it might be called extradata, since this is a common method for storing common codec parameters in the containers.
[17:58:34 CET] <iive> the extradata is given to the codec, so it is codec specific.
[19:23:59 CET] <Arrow> Hi
[19:31:52 CET] <SortaCore> is there a simplistic way to remove a section of a video that's not at start or end?
[20:13:57 CET] <Pollo_> Hi guys, how i can input a live video as ffmpeg input please ? like ffmpeg -i "onlineUrlWithoutDownloadingIt"
[20:14:59 CET] <JEEB> that's exactly how it works for live inputs
[20:15:05 CET] <JEEB> I use that daily with UDP or HTTP streams :P
[20:15:44 CET] <Pollo_> i would like to use a youtube live stream video as input but it's not working
[20:25:15 CET] <Pollo_> My goal is to make his working : ffmpeg -i "https://youtu.be/nKMuBisZsZI" -f flv output.flv but i have this error A TLS packet with unexpected length was received. https://youtu.be/nKMuBisZsZI: Input/output error, anyone can help me pls ?
[20:26:09 CET] <JEEB> umm
[20:26:17 CET] <JEEB> you need something like youtube-dl to give you the ACTUAL urls
[20:26:24 CET] <JEEB> and then pipe that to ffmpeg or so
[20:27:50 CET] <Pollo_> youtube-dl error : ERROR: Unable to download webpage: <urlopen error [Errno 8] _ssl.c:510: EOF occurred in violation of protocol>
[20:28:46 CET] <Pollo_> it should be fine with a "normal" video but seems to not working with live ones
[20:36:49 CET] <relaxed> Pollo_: https://youtu.be/nKMuBisZsZI works here with youtube-dl
[21:19:02 CET] <timbozeman> I am trying to run this command to add watermarks in random places https://pastebin.com/2ESjSVsb
[21:19:24 CET] <timbozeman> It seems to run fine on any 1080p files, but they don't show up on any 720p files :/
[21:22:23 CET] <saml> oh man i'm lonely
[21:22:29 CET] <saml> oh man wrong chan
[21:23:01 CET] <saml> that's a nice command
[21:23:10 CET] <therage3> Nice save
[21:23:34 CET] <saml> what's t?
[21:23:55 CET] <saml> why escape commas?
[21:24:16 CET] <furq> because you need to escape commas in filterchains
[21:24:28 CET] <saml> ah nice
[21:24:55 CET] <saml> is that flying text all over video ? or a static text at a fixed location?
[21:25:14 CET] <timbozeman> all over for a few seconds at a time
[21:25:34 CET] <saml> that's nice
[21:25:38 CET] <furq> i don't see why that would break
[21:25:54 CET] <saml> i saw google paper about removing video watermarks via machine learning
[21:26:16 CET] <saml> they recommend adding distortion to the text otherwise, it's easy to detect and remove
[21:26:51 CET] <furq> but it is probably worth pointing out that if() takes 0 as false and non-zero as true
[21:27:11 CET] <timbozeman> hmm
[21:27:15 CET] <furq> so if(eq(foo,0)) is the same as if(foo)
[21:27:23 CET] <furq> or er
[21:27:25 CET] <furq> ifnot(foo)
[21:30:42 CET] <timbozeman> I got that command off of stack overflow. I see there's a similar one in the examples.
[21:30:56 CET] <timbozeman> https://ffmpeg.org/ffmpeg-filters.html#Examples-52
[21:33:20 CET] <saml> ah the text jumps around. i thought it would animate smoothly
[22:16:43 CET] <timbozeman> furq, if I change the if to ifnot it sprays the text all over many times a second. I guess that's progress.
[22:18:43 CET] <timbozeman> I wish I knew how to debug why the if is returning 0 on 720p videos
[22:41:57 CET] <timbozeman> I think it's this part mod(t\,2) the delay thing
[22:43:51 CET] <Devrim> Is it possible for ffmpeg to choose the higest quality version of a HLS stream from a m3u8 file?
[22:53:55 CET] <timbozeman> hm it works if i base it on frames x=if(not(mod(n\,25))\,rand(0\,(W-tw))\,x)
[22:54:02 CET] <timbozeman> is that hacky?
[23:36:39 CET] <sikilikis> hey all. Is there an option or parameter for ffmpeg to tell it to exit if a warning message has repeated too many times?
[23:37:38 CET] <sikilikis> alternatively, an option to get it to exit if it encounters a specific warning message?
[23:44:10 CET] <relaxed> sikilikis: send ffmpeg's output to a file, check if the warning message appears using something and, if so, kill ffmpeg in a way that it exits
[23:46:11 CET] <sikilikis> is there no way to do that with ffmpeg itself? someway to treat a warning message as an error message or something?
[23:46:42 CET] <relaxed> yes, but I thought you wanted a specific error
[23:47:36 CET] <sikilikis> yes thats true. Are you saying that you can specify "treat all warnings as errors" but not "treat this specific warning as error" ?
[23:48:38 CET] <relaxed> maybe I'm mistaken, but I thought you could
[23:50:20 CET] <sikilikis> just a little more information, I'm specifically looking for the alsa buffer xrun message. It's not an error message so ffmpeg keeps running. I want it to stop running so I can restart it
[23:51:58 CET] <relaxed> shouldn't be too hard to script
[23:52:22 CET] <sikilikis> it shouldnt. I was just hoping to avoid having to do that.
[23:55:02 CET] Action: relaxed crushes hope
[00:00:00 CET] --- Sat Feb 10 2018
More information about the Ffmpeg-devel-irc
mailing list