[Ffmpeg-devel-irc] ffmpeg.log.20160415

Sat Apr 16 02:05:01 CEST 2016

[03:31:21 CEST] <k_sze[work]> What are the relative merits between compiling FFmpeg on Windows with MSVC vs with MinGW?
[03:36:15 CEST] <furq> one of them doesn't require you to have MSVC installed
[03:53:14 CEST] <k_sze[work]> And I'm a bit confused about yasm and SDL.
[03:53:48 CEST] <k_sze[work]> Here it says to download yasm from a web site: https://trac.ffmpeg.org/wiki/CompilationGuide/MinGW#yasm
[03:54:34 CEST] <k_sze[work]> But here I'm told to install something related to yasm and SDL using pacman.
[03:54:46 CEST] <k_sze[work]> Does the pacman command already cover what I need?
[03:56:16 CEST] <furq> if you're using msys2 then sure
[03:56:34 CEST] <furq> original msys doesn't have a package manager
[03:56:47 CEST] <k_sze[work]> I meant this: http://ffmpeg.org/platform.html#toc-Native-Windows-compilation-using-MSYS2
[03:56:51 CEST] <k_sze[work]> I forgot to paste the link.
[03:56:57 CEST] <k_sze[work]> ok
[04:13:45 CEST] <DHE> as long as you get a copy of yasm of at least the minimum version it's fine
[09:48:38 CEST] <k_sze[work]> So if I understand correctly, x264 in ffmpeg 3.0.1 now supports threading?
[09:48:51 CEST] <k_sze[work]> (whereas it didn't in e.g. ffmpeg 2.5.3)
[09:48:53 CEST] <furq_> it's always supported threading
[09:49:36 CEST] <k_sze[work]> because if I run `ffmpeg -help encoder=libx264` with ffmpeg 2.5.3, it says "Threading capabilities: no"
[09:49:45 CEST] <k_sze[work]> zeranoe's build, btw.
[09:50:50 CEST] <furq_> i'm not sure what that refers to
[09:51:02 CEST] <furq_> i get the same with the zeranoe build but it definitely does multithreaded encoding
[09:52:01 CEST] <k_sze[work]> At some point I need to figure out how to build ffmpeg with mingw-w64
[09:52:19 CEST] <furq_> under msys or cross-compiling
[10:00:48 CEST] <k_sze[work]> hmm, this has slice_count, threads, and thread_type options. http://ffmpeg.org/ffmpeg-codecs.html#Codec-Options
[10:02:02 CEST] <furq_> you don't usually need to pass -threads to x264, it'll autodetect
[10:02:08 CEST] <furq_> and i don't think the other two apply to x264
[10:13:44 CEST] <brontosaurusrex> My terminal is sliglty bork when ffmpeg exists, what was the solution to this?
[10:14:55 CEST] <furq> ^L
[10:21:06 CEST] <BtbN> stty sane
[10:21:25 CEST] <BtbN> on a clean exit that shouldn't happen though
[10:22:31 CEST] <brontosaurusrex> stty sane < that means anything?
[10:23:11 CEST] <BtbN> see man stty if you want to know what exactly it means
[10:23:16 CEST] <BtbN> it fixes your terminal though
[10:23:25 CEST] <brontosaurusrex> oh, ok, so it is a command :)
[10:46:23 CEST] <k_sze[work]> furq: thread_type appears specifically under the x264 section as well: http://ffmpeg.org/ffmpeg-codecs.html#toc-Options-23
[10:48:18 CEST] <furq> fun
[10:48:27 CEST] <furq> i guess sliced threading is only useful for low-latency encoding
[11:19:30 CEST] <k_sze[work]> x264 does *not* do slicing by default?
[11:23:51 CEST] <k_sze[work]> And my computer has 4 logical cores. Why does ffmpeg do threads=6 by default?
[11:24:12 CEST] <c_14> I think it's 1.5 * numcpu
[11:25:17 CEST] <lukesan> morning all
[11:28:25 CEST] <furq> k_sze[work]: no, the default is frame-based threading
[11:28:46 CEST] <furq> which is faster but has worse latency
[11:29:10 CEST] <k_sze[work]> c_14 is that in case some threads are waiting for IO, the other threads can do work?
[11:30:06 CEST] <c_14> encoder threads usually don't do io
[11:30:30 CEST] <c_14> It's probably because switching to an inactive thread is faster than reinitializing the current thread
[11:30:44 CEST] <k_sze[work]> ok
[11:31:03 CEST] <c_14> *much io
[11:31:21 CEST] <c_14> Everything does io in some sense
[11:44:30 CEST] <hc2p> hey guys, i'm sending rtmp to nginx-rtmp, pushing it to ffmpeg -i rtmp://...  then connection handshake fails with
[11:44:35 CEST] <hc2p> [rtmp @ 0x37fc6e0] Erroneous C2 Message epoch does not match up with C1 epoch
[11:44:35 CEST] <hc2p> [rtmp @ 0x37fc6e0] Erroneous C2 Message random does not match up
[11:45:18 CEST] <hc2p> sorry, to be precise it's `ffmpeg -listen 1 -i rtmp://0.0.0.0:1935/live/test`
[11:46:58 CEST] <hc2p> i have the feeling i'm missing smth really obvious
[11:49:56 CEST] <hc2p> http://pastebin.com/fFb2nxc9 http://pastebin.com/Xy76zBEx
[12:22:35 CEST] <realies> would ffmpeg treat .ts files differently than .mp4
[12:22:43 CEST] <realies> eg. do I need a container?
[12:26:18 CEST] <k_sze[work]> Does anybody know what's the *default* -context when using ffv1 level 1? I can't seem to find it documented anywhere.
[12:26:29 CEST] <k_sze[work]> And it's not shown when I run the ffmpeg command.
[12:38:49 CEST] <DHE> realies: .ts is a container, it's called mpegts
[12:39:18 CEST] <realies> DHE so it encapsulates the h264 inside?
[12:39:33 CEST] <DHE> yes
[12:39:47 CEST] <realies> DHE, right, ty
[13:09:27 CEST] <hrw> hi
[13:10:30 CEST] <hrw> can ffprobe ignore what it does not know and continue? http://paste.fedoraproject.org/356045/71858214 is "ffprobe /dev/sr0" with some dvd movie which I want to convert to h264 file
[14:30:24 CEST] <BtbN> hrw, looks like it did continue?
[14:31:01 CEST] <BtbN> the unknown stuff is in stream 0, and it still listed the other stream
[14:31:42 CEST] <hrw> BtbN: I excepted audio tracks to be listed but it is $%#$%@ dvd disk so anything can be borken
[14:32:07 CEST] <BtbN> or the first track it found only has video
[14:32:24 CEST] <hrw> or
[14:32:41 CEST] <hrw> anyway it is last time when I rip dvd for someone d:
[14:34:18 CEST] <BtbN> there are better tools to rip a dvd than the ffmpeg cli.
[14:37:21 CEST] <hrw> BtbN: used several of them. all faild
[14:37:42 CEST] <BtbN> sounds like a broken disk to me
[14:40:46 CEST] <hrw> BtbN: 'mplayer -dvd-device /dev/sr0 dvd://' is capable of playing fine. mencoder with same info generates out of sync stuff. handbrake gives soup of pixels
[14:41:33 CEST] <BtbN> so of all tools available, you chose the ones least suited for ripping a DVD.
[14:41:39 CEST] <hrw> ;D
[14:42:13 CEST] <hrw> BtbN: dvd ripping during other work == not enough time to dig for all tools
[14:43:42 CEST] <Snoo20664> hello all
[14:44:00 CEST] <Snoo20664> ffmpeg -ss 00:01:00 -i video.mp4 -to 00:02:00 -c copy cut.mp4
[14:44:31 CEST] <Snoo20664> this ^ will cut out 1 minuit of video from 1m to 2m and save as cut.mp4
[14:44:52 CEST] <Snoo20664> how can i omit that and keep the proceeding and following sections of video
[14:45:10 CEST] <BtbN> uhm, don't do anything to the input file?
[14:45:14 CEST] <Snoo20664> so like i want to keep the first min, omit mins 1 to 2, and keep min 2 to 3
[14:45:24 CEST] <Snoo20664> i want to cut out an advert BtbN
[14:45:33 CEST] <Snoo20664> at 1min to 2min in this example
[14:45:56 CEST] <BtbN> make one video with only the first minute, and one starting at 2 minutes, and then concat those.
[14:46:04 CEST] <Snoo20664> yeah
[14:46:16 CEST] <Snoo20664> i was hoping there was a simpler/faster way
[14:47:31 CEST] <Snoo20664> like the opposite of -ss and -to would be great
[14:49:06 CEST] <BtbN> that's not easily possible, as the data that would product is non-contiguous.
[14:49:58 CEST] <hrw> BtbN: so which tool would you suggest to try?
[15:36:58 CEST] <Snoo20664> BtbN can you help
[15:37:40 CEST] <Snoo20664> ffmpeg -ss 00:14:46 -i test.ts -to 00:28:17 -vcodec copy -acodec copy temp\%seg%.ts
[15:38:00 CEST] <Snoo20664> instead of cutting from 00:14:46 to 00:28:17
[15:38:32 CEST] <Snoo20664> it cuts from 00:14:46, but doesn't end at 28:17, the video is just 28:17 long
[15:42:32 CEST] <Snoo20664> it's using the -to time as a duration like -t instead of an endpoint
[15:44:06 CEST] <Snoo20664> i think what it's doing is i set the start point as 00:14:46 with -ss 00:14:46
[15:44:45 CEST] <Snoo20664> and it jumps to that point and then set's it to 00:00:00, so when i set my end point of 00:28:17 it counts to that from it's new zero point
[15:44:57 CEST] <Snoo20664> there has to be someway round this
[15:47:04 CEST] <Snoo20664> got it
[15:47:12 CEST] <Snoo20664> -copyts to maintain timestamps
[15:47:20 CEST] <Snoo20664> ^should be a default setting tbh
[16:55:08 CEST] <yzT> is this diagram correct? http://i64.tinypic.com/289kfo1.png
[16:57:36 CEST] <Snoo20664> i don't think so
[16:58:25 CEST] <yzT> what's wrong?
[16:59:28 CEST] <yzT> mmmm well, the packager should connect to the CENC and the CENC to the website, right?
[17:20:52 CEST] <satinder___> how we can capture current image from /dev/video0
[17:21:01 CEST] <satinder___> please anyone can help
[17:26:41 CEST] <satinder___> hi please any know how we can capture a single current image from /dev/video0 with ffmpeg
[17:26:54 CEST] <satinder___> ??
[17:30:13 CEST] <Snoo20664> ffmpeg -i /dev/video0 -frames:v 1 out2.jpg
[17:30:31 CEST] <Snoo20664> should do it for you satinder___
[17:39:16 CEST] <satinder___> Snoo20664 -frame:v unrecognized option
[17:43:49 CEST] <satinder___> ??
[18:30:34 CEST] <realies> is it possible to playback 9 clips in a 3x3 grid simultaneously
[18:41:34 CEST] <benbro> how much bandwidth I need for live 720p at 30fps VP8 video?
[18:44:25 CEST] <DHE> depends on quality needs. 4 or 5 megabit/sec for video might be a good starting point
[19:08:09 CEST] <benbro> DHE:  I was hoping for 1-1.5 Mbps :)
[19:09:38 CEST] <durandal_1707> realies: yes
[19:16:58 CEST] <DHE> benbro: it depends on your content. cartoons might work out. a hockey game would not.
[19:20:46 CEST] <benbro> DHE: I'm shooting a meeting
[19:21:02 CEST] <DHE> ... 1.5 might work then. stationary camera and all that
[19:21:25 CEST] <DHE> all I can say is, give it a try. see how it turns out
[19:42:54 CEST] <Kuroonehalf> Yo guys. I currently have some really poorly ripped dvds that for some reason are quite large and I'd like to do something about. They're 480p, around an hour each, but somehow each one takes up around 1.2GB.  They were oririnally in avi and I reencoded to mp4 but they're still way larger in size than it looks like they should be
[19:43:02 CEST] <Kuroonehalf> is there perhaps some option ffmpeg has to deal with this?
[19:43:15 CEST] <blue_misfit101> re-rip them :)
[19:43:30 CEST] <Kuroonehalf> I was hoping that wasn't the only option :p
[19:44:10 CEST] <kepstin> well, you can probably just re-encode them using a lower bitrate/lower quality setting/better codec, if the source looks decent.
[19:44:48 CEST] <kepstin> obviously if you can start from the original dvd rather than this multiple-re-encoded file, that would be best.
[20:17:04 CEST] <realies> durandal_1707, how?
[20:17:51 CEST] <durandal_1707> hstack and vstack combo
[20:20:43 CEST] <realies> I need to look into this
[20:20:55 CEST] <realies> any ideas if it's resource intensive?
[20:21:50 CEST] <realies> I want to make a scrubable/seekable 3x3 grid, tried with CasparCG, but it seems to not be able to handle the situation effectively
[20:26:46 CEST] <durandal_1707> ffmpeg is for transcoding not playback
[20:27:16 CEST] <furq_> you can use hstack with ffplay
[20:27:22 CEST] <durandal_1707> you gona use libavfilter api?
[20:27:26 CEST] <furq> whether it performs acceptably i have no idea
[20:27:35 CEST] <realies> I thought ffplay functionality is embedded in ffmpeg
[20:27:38 CEST] <furq> and you won't be able to seek the videos individually
[20:27:51 CEST] <durandal_1707> not possible with ffplay
[20:27:56 CEST] <realies> furq the idea is for them to stay in sync
[20:31:08 CEST] <durandal_1707> mpv uses ffmpeg libs for playback
[20:31:38 CEST] <furq> oh
[20:31:54 CEST] <furq> so you can use hstack/vstack with ffplay but you can't provide multiple input files
[20:32:26 CEST] <furq> so that's no good
[20:33:41 CEST] <realies> any ideas for making that happen?
[20:34:06 CEST] <furq> and i don't think you can use filter_complex with mpv
[20:34:37 CEST] <realies> : /
[20:35:36 CEST] <furq> as far as using ffmpeg goes i think your only option is to transcode it and stream/pipe the output to a player
[20:35:56 CEST] <furq> or use the api
[20:36:36 CEST] <realies> I wonder what the latency would be when transcoding
[20:36:55 CEST] <furq> if you're piping it to a player then there's no reason to not use rawvideo
[20:37:03 CEST] <furq> so it shouldn't be too bad
[20:37:05 CEST] <realies> prob not pleasable for realtime operations
[20:37:22 CEST] <realies> I'll try it out though
[20:37:32 CEST] <furq> i guess you'll be limited by the decoding speed
[20:41:24 CEST] <realies> can the API be used with .NET?
[20:42:28 CEST] <furq> not to my knowledge
[20:42:40 CEST] <furq> googling turns up a few wrappers but they're all really old and alpha versions
[20:42:47 CEST] <realies> yeah...
[20:42:58 CEST] <realies> bummer haha
[20:42:58 CEST] <furq> which is par for the course for ffmpeg bindings
[20:44:01 CEST] <realies> hm... https://github.com/Microsoft/FFmpegInterop
[20:44:30 CEST] <furq> One of the advantages of this approach is that audio and video synchronization is handled by the Windows media pipeline.
[20:44:39 CEST] <furq> i don't think microsoft know what "advantages" means
[20:44:56 CEST] <realies> im a newb and don't get that
[20:45:07 CEST] <realies> wm pipelines suck?
[20:45:39 CEST] <furq> i assume that means directshow
[20:46:02 CEST] <furq> i was always told that if you can't say anything nice about something then you should say nothing at all
[20:46:12 CEST] <realies> :D
[20:46:34 CEST] <realies> what does ffplay create in it's window then?
[20:46:41 CEST] <furq> it's an SDL window
[20:46:42 CEST] <realies> *playback window
[20:47:09 CEST] <furq> it works exactly the same on every platform afaik
[20:47:13 CEST] <realies> my manager was mentioning something about sdl today...
[20:48:09 CEST] <furq> it looks like this ffmpeginterop lib only does basic decoding and playback stuff anyway
[20:48:30 CEST] <furq> based on me looking at the source directory for about 20 seconds
[20:48:37 CEST] <realies> haha
[20:50:03 CEST] <realies> I guess making a wrapper for the ffmpeg binary is also a thing
[20:50:10 CEST] <realies> a funny one
[22:02:26 CEST] <blue_misfit101> how do you specify a qpfile when using libx264 in ffmpeg?
[23:59:29 CEST] <snakeryslug> i want lossless gif to webm. is that possible? i've done `ffmpeg -i input.gif -c:v libvpx -crf 4 -b:v 32000K output.webm` but the output looks worse.
[00:00:00 CEST] --- Sat Apr 16 2016