[Ffmpeg-devel-irc] ffmpeg.log.20180331

burek burek021 at gmail.com
Sun Apr 1 03:05:02 EEST 2018


[08:48:40 CEST] <sysgrammer> hello,
[08:49:26 CEST] <sysgrammer> I am starting to convert all my media to a single format so taht I can stream it using icecast.
[08:49:53 CEST] <sysgrammer> I strated by installing ffmpeg on a debian box
[08:50:32 CEST] <sysgrammer> it reported that ffmpeg does not have installion candidate.
[08:50:54 CEST] <sysgrammer> I am running debian 3.16.51-3
[08:51:09 CEST] <sysgrammer> kernel 3.16.0-4
[08:51:46 CEST] <furq> debian isn't packaged in jessie unless you use backports
[08:51:48 CEST] <sysgrammer> any ideas?
[08:51:54 CEST] <furq> but jessie has been oldstable for a year so you should really just upgrade
[08:52:01 CEST] <furq> er, ffmpeg isn't packaged in jessie
[08:53:24 CEST] <furq> https://www.johnvansickle.com/ffmpeg/
[08:53:27 CEST] <furq> alternatively you can just use this
[08:53:54 CEST] <furq> i would probably recommend that over backports
[08:54:11 CEST] <furq> i know people have had issues with that backport in particular
[09:11:02 CEST] <sysgrammer> thanks for the info. What do you think about avconv vs ffmpeg?
[09:11:15 CEST] <sysgrammer> i.e., libav-tools
[09:12:54 CEST] <durandal_1707> sysgrammer: avconv is Libav - an FFmpeg fork
[09:17:46 CEST] <sysgrammer> thanks a ton!! I did not know that....
[11:00:30 CEST] <kerio> stop using bad distros that make it hard to use non-obsolete software
[11:04:25 CEST] Action: Mavrik goes to install Windows then.
[11:05:48 CEST] <kerio> D:
[12:19:17 CEST] <Swervz> Hey just wondering if higher bitrate is faster to encode with the same content, codec and preset etc
[12:19:26 CEST] <Swervz> the comparisons on google are for different presets
[12:19:33 CEST] <meins> hello
[12:19:36 CEST] <Swervz> instead of just bitrate
[12:21:18 CEST] <Mavrik> Hmm, it shouldn't really make a noticable difference.
[12:21:30 CEST] <Mavrik> Unless your HDD can't keep up
[12:23:31 CEST] <Swervz> Someone else was telling me that higher bitrate takes longer and google is saying lower bitrate takes longer because it does more compression but they seem to mention presets too
[12:23:38 CEST] <meins> does some here have experience with receiving a Stream over udp with ffmpeg?
[12:25:54 CEST] <furq> Swervz: with the same preset, lower crf takes longer to encode
[12:26:09 CEST] <furq> idk if makes any difference in abr mode
[12:26:15 CEST] <furq> it
[12:26:41 CEST] <Swervz> Would you happen to have a source for that so I can show this guy?
[12:27:06 CEST] <furq> you can just run ffmpeg and see it for yourself
[12:27:12 CEST] <furq> it's a noticeable difference
[12:27:36 CEST] <furq> also it sounds like the guy is in agreement with me
[12:27:39 CEST] <furq> lower crf is higher bitrate
[12:28:07 CEST] <furq> also i should stress it's a noticeable difference but not really a significant difference
[12:28:16 CEST] <furq> you're talking maybe a couple of percent
[12:28:32 CEST] <Swervz> ah okay thankyou
[12:28:58 CEST] <Swervz> Yeah I guess I was thinking of lower crf and confusing it with lower bitrate
[12:34:04 CEST] <furq> http://vpaste.net/PhCy0
[12:34:30 CEST] <furq> not exactly scientific, i'll admit
[12:36:11 CEST] <Swervz> ah ok yeah that makes sense seems as lower crf is higher bitrate
[12:36:26 CEST] <Swervz> Guess I confused it with cbr
[12:41:26 CEST] <Swervz> Haven't had time to sleep the past day or so and I got to finish some things so stuff doesn't make as much sense to me right now
[13:25:58 CEST] <axew3> maybe not the right channel to ask this, but after ffmpeg installed and full working asking myself: which would be the best way to not allow anybody access to ffmpeg. I mean, i would only that request are allowed within the server machine where it is installed ... what way is better i start to follow about this?
[13:40:23 CEST] <c_14> axew3: remove rx permissions for other and either use a group for access or posix acls
[13:47:50 CEST] <axew3> c_14 thank you ... as newbie about this, i will try to understand the full mean of your suggestion ...
[13:53:08 CEST] <sfan5> axew3: https://en.wikipedia.org/wiki/Modes_(Unix) should explain this sufficiently
[13:53:36 CEST] <sfan5> but note that even if you disallow your ffmpeg from being ran, what prevents users from downloading their own ffmpeg?
[14:46:24 CEST] <rjp421> http://streamingmedia.brightcovegallery.com/detail/videos/streaming-media-west-2017/video/5643859283001/d103:-how-to:-building-a-more-robust-cloud-encoder-with-ffmpeg-more?autoStart=true&page=1
[15:31:58 CEST] <Kam_> Hi, what is the difference between the -r and -framerate parameter?  Why should I choose one over the other for e.g. image2 input?
[15:32:38 CEST] <JEEB> welcome to the wonderful world of AVOptions
[15:33:01 CEST] <JEEB> which is what "framerate" is I think
[15:34:13 CEST] <JEEB> so "framerate is an AVOption in that specific image2 "demuxer"
[15:34:18 CEST] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavformat/img2dec.c;h=f3f52c83b355ad00f9cc2f49f67f2e47337d143a;hb=HEAD#l565
[15:34:32 CEST] <JEEB> while -r I think sets the frame rate of the AVCodecContext, but let me check
[15:34:37 CEST] <kepstin> Kam_: the "-framerate" option requests that the demuxer/input device (for some types of inputs) read at a specific framerate.
[15:34:42 CEST] <JEEB> as in, -r before -i
[15:35:15 CEST] <kepstin> Kam_: the -r option causes ffmpeg to rewrite the timestamps to constant rate after reading them from the input, which is sometimes useful but usually just causes audio desync.
[15:35:51 CEST] <kepstin> Kam_: both of those descriptions are for the input options. Output is different.
[15:35:59 CEST] <Kam_> oh, I see!
[15:35:59 CEST] <JEEB> and sometimes -r before -i just does nothing because something else overrides that value anyways :)
[15:36:58 CEST] <kepstin> Note that for demuxers/input devices that support the framerate option, it'll show up in the help - e.g. "ffmpeg -h demuxer=image2"
[15:37:13 CEST] <JEEB> yes, because it's an AVOption of that "demuxer"
[15:37:22 CEST] <JEEB> while -r is an ffmpeg.c option
[15:40:02 CEST] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=fftools/ffmpeg_opt.c;h=d7a7eb06628062d26322a77e97956727b23b8163;hb=HEAD#l789
[15:40:05 CEST] <JEEB> this is what -r seems to do
[15:40:21 CEST] <Kam_> thanks, that clarified it for me!
[15:40:23 CEST] <JEEB> it goes into the matching stream(s) and sets the ist->framerate value
[15:40:30 CEST] <axew3> sfan5 -> "but note that even if you disallow your ffmpeg from being ran, what prevents users from downloading their own ffmpeg?" the file need to be available in my case, after processed (and even before on browser after js stream capture is allowed to download his own video media file) ... if i would like to not let download the file after conversion process, i had think should be sufficient
[15:40:30 CEST] <axew3> to not return a value after conversion :)
[15:40:33 CEST] <JEEB> which might or might not get used
[15:41:46 CEST] <kepstin> axew3: oh, you're running ffmpeg from a web service? Then it's completely up to you to authenticate users in the web service before running ffmpeg.
[15:42:25 CEST] <axew3> obscure the way to achieve heheh ... at moment
[15:43:12 CEST] <axew3> i'm php js sql skilled maybe, not too much network admin or some else
[15:43:52 CEST] <axew3> so i will return over if not banned!
[15:44:50 CEST] <axew3> ah ok re -read
[15:45:04 CEST] <axew3> think i've understand
[15:49:31 CEST] <axew3> by the way the access to server could also be limited to internal calls via htaccess or apache conf ... isn't it?
[15:52:46 CEST] <axew3> sorry it was: the way to access to mmpeg via a server (because a file into the server execute mmpeg)
[15:53:14 CEST] <axew3> ffmpeg
[15:58:46 CEST] <axew3> in effect it is already blocked by any external call this file, due to a token released for logged in users that is passed via post ... mh i have an unresolved question about this if curious, stikll not answered (even if resolved on my scripts) https://stackoverflow.com/questions/49150032/php-cookie-and-assigned-js-var-with-same-value-mismatch-if-img-url-on-html-eleme  .... but to return to
[15:58:46 CEST] <axew3> the point, after an user have his token released and recognized, remain a point
[15:58:55 CEST] <axew3> this file could be curled
[15:59:05 CEST] <axew3> and i do not want this
[16:15:56 CEST] <axew3> mh i could associate to any call a refresh of the token that will change at any request ... maybe stored into db before to send out ... sorry for the out of topic ... mh i'll see if possible to resolve via php sql and just a cookie ... mumble mumble ... little confused
[16:20:59 CEST] <axew3> but it seem to be hard to avoid almost an external call, only via a php sql cookie check ... wait no ... if the token is released only when the strem process on browser finish ... by the way, should be of course the way to achieve the same just via htaccess. I will check.
[16:31:36 CEST] <axew3> i stop with this "non in topic", just following to get an idea, i see that is possible to Order deny,allow
[16:31:36 CEST] <axew3> Deny from all
[16:33:14 CEST] <axew3> adding the ip allowed to execute, but i giess isn't easy to mimic the ip of the server passing it
[16:33:17 CEST] <axew3> guess
[16:36:42 CEST] <axew3> guess was hope, sorry and bye for now ;)
[17:16:41 CEST] <tomtomb> hi all, been wondering, is there any way to overlay audio meters on a video with ffmpeg?
[17:19:04 CEST] <durandal_1707> tomtomb: yes
[17:19:36 CEST] <tomtomb> durandal_1707, what would be the best way to approach it?
[17:21:50 CEST] <durandal_1707> tomtomb: have you ever use ffmpeg?
[17:22:30 CEST] <tomtomb> durandal_1707, I have been for quite a while
[17:23:41 CEST] <durandal_1707> then use overlay filter with showvolume filter
[17:25:36 CEST] <tomtomb> durandal_1707, showvolume, will try that - thanks
[17:28:35 CEST] <tomtomb> durandal_1707, seems like showvolume shows a number value, can it generate a meter?
[17:30:29 CEST] <durandal_1707> tomtomb: it shows also meter
[18:15:41 CEST] <gjtorikian> hello friends, i have a question on generating PNGs from a video with subtitles. relevant commands and output is here: https://pastebin.com/vTywLY04 thanks in advance!
[19:06:28 CEST] <sfan5> gjtorikian: ffmpeg does not render subtitles, you need the "subtitles" filter
[19:10:43 CEST] <JEEB> oh I /wish/ ffmpeg.c didn't render subtitles
[19:10:58 CEST] Action: JEEB stares lonely into his sub2video patches
[19:52:47 CEST] <gjtorikian> sfan5 THANK YOU! I never would've gotten that. had to remake `ffmpeg` with `--enable-libass` but it's all smooth now
[21:57:34 CEST] <tomtomb> so showvolume works good for me, how can I reposition a video stream (change it's X/Y) ?
[21:59:50 CEST] <furq> reposition relative to what
[22:00:47 CEST] <tomtomb> sorry, when I overlay it on top of another stream, I tried [0:0][fg]overlay=x=100:y=100[out] but it didn't move
[22:01:35 CEST] <furq> if fg is the foreground then that should come first
[22:04:36 CEST] <tomtomb> furq, got it, thanks. one more question - is it possible to use the overlay filter with more than 2 streams?
[22:05:05 CEST] <tomtomb> let's say I have 3 video streams, I want to overlay them one on top the other, can I do it with one filter or do I need to overlay 1+2 and then [output]+3 ?
[22:05:36 CEST] <furq> yeah you need to do it twice
[22:05:45 CEST] <tomtomb> furq, got it, thank you
[22:11:55 CEST] <vlad_> I am trying to use the h264_nvenc encoder, but it fails with "Cannot init Cuda" (on ubuntu 17.10)
[22:14:05 CEST] <BtbN> you need the proprietary nvidia driver for it to work
[22:14:14 CEST] <BtbN> and the nvidia_uvm module needs to be loaded for CUDA to work
[22:14:58 CEST] <vlad_> BtbN: I have the proprietary driver installed
[22:15:18 CEST] <vlad_> but I do not have nvidia_uvm
[22:16:19 CEST] <BtbN> well, get it then
[22:28:02 CEST] <DHE> vlad_: it should be another loadable kernel driver with any decent version of the nvidia driver. just modprobe it and maybe you'll be fine
[22:28:46 CEST] <vlad_> BtbN, DHE: thanks! that worked
[22:29:13 CEST] <BtbN> DHE, a lot of distributions remove it from their default driver
[22:29:36 CEST] <DHE> really? CUDA and maybe opencl just doesn't work without it.
[22:29:42 CEST] <BtbN> yep
[22:29:50 CEST] <BtbN> At least on Ubuntu it's in a seperate package
[22:30:09 CEST] <BtbN> if the module exists, cuInit would otherwise just autoload it
[22:30:45 CEST] <DHE> if run as root...
[22:31:46 CEST] <BtbN> no, doesn't need to run as root for that
[22:32:16 CEST] <DHE> non-root can load a kernel driver? I disagree...
[22:32:32 CEST] <BtbN> The existing driver part will load it
[22:32:54 CEST] <DHE> I've been running folding at home on my GPU. it's never done that for me...
[22:33:13 CEST] <BtbN> I never manually loaded uvm, and it just wors. Something must be auto-loading it
[00:00:00 CEST] --- Sun Apr  1 2018


More information about the Ffmpeg-devel-irc mailing list