[Ffmpeg-devel-irc] ffmpeg.log.20171218

Tue Dec 19 03:05:01 EET 2017

[00:00:53 CET] <prelude2004c> well i got it all going except when i add lnppi
[00:01:07 CET] <prelude2004c> some reason i am missing that and not sure how to get it.. still trying .
[01:20:13 CET] <kepstin> the nppi stuff is only needed for some video filters, iirc, you don't need it to use the hardware encoder
[01:31:49 CET] <prelude2004c> ok guys.. only thing i am missing now is libnpp
[01:31:53 CET] <prelude2004c> how do i enable that thing
[01:32:08 CET] <prelude2004c> says libnpp can't be found
[01:32:17 CET] <prelude2004c> i tried to install via the apt-get but i can't seem to get the right one
[01:48:02 CET] <prelude2004c> can anyone help?
[03:05:57 CET] <SortaCore> is it good practice to have DTS lower than PTS?
[03:06:51 CET] <iive> usually decoder time stamps are just reordered presentation time stamps
[03:09:28 CET] <SortaCore> what's the logic to the reordering?
[03:09:39 CET] <SortaCore> atm I'm just doing dts = pts = framenum
[03:11:35 CET] <DHE> when frames are decoded in a different order than they are rendered, dts <= pts. if they're in order, dts=pts
[03:11:47 CET] <DHE> most notably B-frames in the MPEG family of codecs do this
[03:12:43 CET] <DHE> just give the encoder the PTS you want on your AVFrames, and trust that the DTS and PTS it gives you is fine
[03:12:43 CET] <iive> SortaCore: b-frames are bi-directional, it means that they use parts from the previous frame, but also part from the next (future) frame
[03:13:07 CET] <iive> SortaCore: so you need to decode b-frames after the next frame.
[03:13:28 CET] <SortaCore> err, so is there anything wrong with how I'm doing it?
[03:14:05 CET] <DHE> don't choose your own DTS. the encoder will provide it for you.
[03:15:09 CET] <SortaCore> hmm, ok
[03:15:17 CET] <SortaCore> I'll turn off the setting and see if it dies
[03:15:32 CET] <SortaCore> *disable the code
[03:20:44 CET] <SortaCore> why does avformat_free_context() not take a AVFormatContext ** like the other frees?
[03:44:23 CET] <TheRock23> I got a video which is loaded in my library which uses ffmpeg.
[03:44:38 CET] <TheRock23> When the video is written while is being played it stops
[03:44:54 CET] <TheRock23> for example in vlc, the player just continues
[03:44:59 CET] <TheRock23> when there is more data
[03:45:11 CET] <TheRock23> is there a function to enable this?
[03:46:00 CET] <TheRock23> the video is loaded with a 14 second length, but while it's being played, it's 60 seconds (data written). it does stop at 14 seconds
[03:57:00 CET] <TheRock23> just figured out
[03:57:43 CET] <TheRock23> i need configure the chunk size that is loaded in memory somehow
[04:16:00 CET] <SortaCore> I'm losing ~4% CPU usage to System and interrupts
[04:16:07 CET] <SortaCore> that's bad as interrupts make my whole system lag
[04:18:14 CET] <DHE> ... one core or globally?
[04:22:33 CET] <SortaCore> err, I think globally?
[04:22:46 CET] <SortaCore> it's copying from web to file though, it shouldn't be interrupting
[04:24:23 CET] <SortaCore> hmm, might not be ffmpeg
[04:42:10 CET] <SortaCore> weird, dpc latency checker is showing no problem
[04:42:17 CET] <SortaCore> but my games are visibly lagging
[05:06:24 CET] <eahm> hi guys,  I need a tip with variables, deleting etc. all at once
[05:06:54 CET] <eahm> I have a folder of .mkv, I would like to change the container of all of them to .mp4/m4v and delete (trash) the .mkv files
[05:07:42 CET] <SortaCore> use rm with wildcard *.mkv?
[05:07:57 CET] <eahm> "ffmpeg -i video.mkv -c:v copy -c:a copy video.mp4" this one does one by one
[05:08:53 CET] <eahm> this command actually changes the container right? they will work on iTunes this way? trying to convert a folder for my wife
[05:09:50 CET] <eahm> SortaCore: sorry, never used ffmpeg with commands before, no idea what that is
[05:13:30 CET] <SortaCore> it would change the container, not sure if it will make it compatible
[05:13:40 CET] <SortaCore> ffmpeg won't delete the input file, you have to use a second command
[05:15:04 CET] <eahm> yes I mean, Apple is being picky with .mkv files, probably just a rename would do it really, video and audio codecs are on on these files I checked
[05:15:19 CET] <eahm> for deleting no problem, I just need a tip for the variable
[05:15:37 CET] <SortaCore> the variable?
[05:15:39 CET] <eahm> how do I tell ffmpeg to pick every file and repack with the same filename?
[05:15:49 CET] <SortaCore> oh, hmm
[05:15:53 CET] <eahm> some website say %f some %ff?
[05:16:03 CET] <SortaCore> I dunno about the variables
[05:16:08 CET] <SortaCore> you may want a FOR loop?
[05:16:11 CET] <eahm> tried %f but not really sure how to start with the initialfile
[05:16:12 CET] <SortaCore> batch for
[05:18:58 CET] <furq> eahm: i take it you're on windows
[05:19:04 CET] <eahm> yes
[05:20:20 CET] <eahm> tried this "ffmpeg -i *.mkv -c:v copy -c:a copy *.mp4" not working, probably obviously
[05:23:06 CET] <hendry> hi, what cli tool do folks here use to decrypt AES-128 encoded .ts segments?
[05:31:32 CET] <SortaCore> is there a reason my custom metadata is destroyed?
[05:35:59 CET] <eahm> found this that may work: for %f in ("*.mkv") do ffmpeg -i "%f" -vcodec copy -acodec copy "%~f.mp4"
[06:00:17 CET] <Johnjay> eahm: i see another earnest traveler in search of ffmpeg encoding has discovered the milestone
[06:00:41 CET] <Johnjay> next job is to figure out how to get ffmpeg to download mp2t streams and convert them!
[08:06:21 CET] <johnnny22> my ffmpeg says: "Guessed Channel Layout for Input Stream #1.0 : 7.1"   is there a way to specify what kind of channel layout it is in a simple way ?
[08:35:29 CET] <SortaCore> odd
[08:35:39 CET] <SortaCore> getsocketopt() has no error when the socket is closed by TCPView
[08:41:34 CET] <SortaCore> nor does ioctlsocket() for FION_READ, so I guess I'll have to just manually timeout?
[08:41:43 CET] <SortaCore> let's see if ffmpeg picks up on it first
[09:00:57 CET] <SortaCore> seems to be
[10:59:48 CET] <ledufakademy> hello,
[11:00:29 CET] <ledufakademy> is it possible to convert m4a (alac apple lossless) to flac ? if yes which is the command lien to use ? thnak you.
[11:01:14 CET] <sfan5> ffmpeg -i input.m4a -c:a flac output.flac
[11:01:24 CET] <sfan5> make sure you do not run this on AAC files though
[11:01:55 CET] <ledufakademy> i do that : ffmpeg -i "$sndFile" -c:a flac "$sndFileConverted"
[11:02:20 CET] <ledufakademy> but ffmpeg only convert one on two
[11:02:44 CET] <fritsch> that's not how it works
[11:02:51 CET] <fritsch> it's your bashism which is the problem
[11:02:54 CET] <fritsch> has nothing to do with ffmpeg
[11:03:13 CET] <ledufakademy> can you take a look ?
[11:03:42 CET] <fritsch> post your script to e.g. paste.ubuntu.com
[11:03:59 CET] <sfan5> you probably just need a for-loop
[11:04:11 CET] <ledufakademy> https://pastebin.com/ndFyQdUE
[11:06:43 CET] <sfan5> that looks like it should work, what does ffmpeg output
[11:06:44 CET] <sfan5> ?
[11:09:21 CET] <ledufakademy> the error is at teh begining : https://pastebin.com/fT3AYqRf
[11:09:50 CET] <fritsch> yeah
[11:09:59 CET] <fritsch> it's not properly espaped
[11:10:18 CET] <ledufakademy> sure but i can't see where ;-)
[11:10:20 CET] <fritsch> and whatever "ue" is in the front
[11:10:44 CET] <ledufakademy> yes this is this xxxx "ue" ... it must be /Disque/
[11:11:09 CET] <ledufakademy> ... and it work for first file but not second ...
[11:11:29 CET] <fritsch> $find /Disque/Interne/Music/ -name "*.m4a" | sort > listeToConvert.txt <- post that list
[11:11:32 CET] <fritsch> please
[11:13:01 CET] <fps> hmm, looking at http://ffmpeg.org/doxygen/trunk/group__lavu__log.html#ga14034761faf581a8b9ed6ef19b313708
[11:13:22 CET] <fps> i wonder where i can store some additional payload (e.g. a pointer to a class)?
[11:13:38 CET] <ledufakademy> thread safe ! ... hum !!!
[11:14:45 CET] <ledufakademy> yep : here (it's just sample , i'm working with only 4 songs) https://pastebin.com/gV4SeqFX
[11:14:49 CET] <fps> i'd like to reroute the logging output to a window in my application. not being able to pass an arbitrary payload to av_log_set_callback forces me to jump through hoops with global variables.. is there a way around that?
[11:15:57 CET] <furq> ledufakademy: ffmpeg doesn't modify the input file at all so there's no need to back it up
[11:16:27 CET] <ledufakademy> furq: ok , i will implement this.
[11:21:33 CET] <furq> ledufakademy: find . -name "*.m4a" -print0 | xargs -0 -n1 -P2 sh -c 'ffmpeg -i "$1" /tmp/${1%.*}.flac && rm "$1"' ''
[11:21:36 CET] <furq> that should be all you need
[11:21:56 CET] <furq> replace P2 with however many you want to run in parallel (the flac encoder is singlethreaded)
[11:22:22 CET] <furq> and uh
[11:22:24 CET] <furq> get rid of /tmp/
[11:22:45 CET] <ledufakademy> i will test that : thank u very much
[11:23:25 CET] <furq> find . -name "*.m4a" -print0 | xargs -0 -n1 -P2 sh -c 'ffmpeg -i "$1" "${1%.*}.flac" && rm "$1"' ''
[11:23:33 CET] <furq> i forgot some quotes as well for good measure
[11:25:29 CET] <sfan5> doesn't that mean sh is invoked as sh -c 'ffmpeg -i "$1" ...' filename.m4a ?
[11:25:38 CET] <bencoh> what's the trailing ''?
[11:25:45 CET] <furq> the trailing '' is $0
[11:26:08 CET] <bencoh> ah
[11:26:10 CET] <ledufakademy> why my code did not work ?
[11:54:58 CET] <ledufakademy> it was an issue WITH ffmpeg , issue when you're in loop bash : the solution is to had  -nostdin !!!!
[11:55:45 CET] <ledufakademy> https://stackoverflow.com/questions/13995715/bash-while-loop-wait-until-task-has-completed
[13:19:47 CET] <termos> is there a fast way to convert AVFrame's to YUV420P using sws library? I only find examples using sws_scale, but I don't wish to scale the image
[13:26:04 CET] <BtbN> then don't tell it to scale
[14:10:23 CET] <DHE> depending on how you're getting the AVFrame, is it possible to request yuv420p format in the first place?
[14:10:39 CET] <DHE> it might not be possible, but it might be worth pursuing
[15:38:40 CET] <termos> the frame is from the blackmagic decklink SDK, it's possible to down convert on the capture card in hardware but only from fullHD down to SD, not from fullHD down to HD (720p)
[17:11:43 CET] <pmjdebruijn> hey, I've been playing a bit with haldcluts
[17:35:33 CET] <durandal_1707> pmjdebruijn: and?
[17:35:41 CET] <pmjdebruijn> sorry for the delay, got distracted,
[17:35:53 CET] <pmjdebruijn> but I'm having trouble applying it to a video while transcoding
[17:36:30 CET] <pmjdebruijn> the following works fine: ffplay -vf "movie=ilfordfp4plus125.png, [in] haldclut myvid.mp4
[17:37:05 CET] <pmjdebruijn> but ffmpeg -i myvid.mp4 -filter:v "movie=..., [in] hardclut" -c:v .. -c:a .. out.mp4 doesn't work
[17:38:24 CET] <durandal_1707> pmjdebruijn: perhapps you want to use -filter-complex ?
[17:38:31 CET] <pmjdebruijn> not sure
[17:38:45 CET] <pmjdebruijn> I find the movie= syntax rather confusing in the first place
[17:39:00 CET] <pmjdebruijn> -vf "hardclut=mylut.png" would have made more sense to me :)
[17:39:21 CET] <alexpigment> durandal_1707: i believe you mean -filter_complex
[17:39:27 CET] <alexpigment> (unless your version also works)
[17:40:09 CET] <durandal_1707> dunno,  i use -lavfi which is same
[17:40:14 CET] <alexpigment> ah
[17:40:53 CET] <TheRock2> How long does it take to study the handbook of ffmpeg
[17:40:54 CET] <kepstin> pmjdebruijn: anyways, the issue is that the 'haldclut' filter takes two inputs, one is the video that you're filtering, the other either a still image or video containing the hald clut.
[17:41:14 CET] <Fenrirthviti> TheRock2: Probably 20 years of practical experience.
[17:41:36 CET] <pmjdebruijn> kepstin: that makes sense to some extent
[17:41:39 CET] <kepstin> pmjdebruijn: you don't need to use the movie filter, really, it would also work with multiple "-i" options, but you do need to use the -filter_complex option & syntax to handle the multiple-input filter
[17:42:06 CET] <durandal_1707> TheRock2: why would you study handbook of ffmpeg?
[17:43:50 CET] <pmjdebruijn> kepstin: is there reason why it's like that, it feels way to complex than it has to be :)
[17:44:23 CET] <kepstin> pmjdebruijn: doing it this way lets it use any input that ffmpeg can handle for the clut, and it can even be a video.
[17:44:37 CET] <kepstin> it would be a lot more complicated to have the filter open another file internally
[17:44:55 CET] <TheRock2>  I want to learn all commands of ffmpeg
[17:45:05 CET] <kepstin> (although there are a few filters that do that, the 'movie' and 'subtitle' filters are examples)
[17:46:33 CET] <kepstin> pmjdebruijn: but really, the simplest command to apply a hald clut is easy to understand: "ffmpeg -i movie.mp4 -i clut.png -filter_complex '[0][1]haldclut' output.mp4"
[17:46:56 CET] <pmjdebruijn> but that doesn't combine with regular filters does it?
[17:47:20 CET] <kepstin> pmjdebruijn: you can add more filters to the filter_complex parameter
[17:47:35 CET] <kepstin> but you can't use -filter_complex and -vf in the same command
[17:48:07 CET] <pmjdebruijn> that's what I meant
[17:49:52 CET] <pmjdebruijn> oh wait I does work
[17:49:55 CET] <pmjdebruijn> it*
[17:51:19 CET] <pmjdebruijn> I thought you had to keep messing with the multiple inputs and outputs throughout the pipe
[17:52:07 CET] <pmjdebruijn> kepstin: thanks!
[17:53:09 CET] <kepstin> pmjdebruijn: the , character in -filter_complex does the exact same thing as in -vf :)
[19:06:23 CET] <durandal_1707> TheRock2: you cant know all ffmpeg, you must always consult handbook
[19:23:29 CET] <thebombzen> I'm looking to convert a subtitle to UTF-8 text  (ffmpeg -i subtitle -c:s text) but I can't figure out what muxer to send it to with ffmpeg.c so it'll just write the text file
[19:28:03 CET] <kepstin> thebombzen: huh, i don't think that's ever come up before. Doesn't look like there's a way to save subtitles as untimed text. I'd suggest just using the srt output, that's easy enough to parse (you could use sed to remove the timing info, even)
[19:31:18 CET] <thebombzen> may I suggest that there be a "raw subtitle" muxer
[19:31:24 CET] <thebombzen> or even just a "text" muxer
[19:31:35 CET] <thebombzen> and .ass, .vtt, .srt, those should all use the text muxer
[19:31:53 CET] <thebombzen> rather than have an ass muxer, a webvtt muxer, and a subrip muxer?
[19:32:04 CET] <thebombzen> seeing as the code for those three muxers is practically identical
[19:32:12 CET] <thebombzen> and those three would just become aliases for the text muxer
[19:32:28 CET] <thebombzen> that way it wouldn't break existing code
[19:35:13 CET] <kepstin> they're different enough that this wouldn't really make sense, particularly since it's the muxer that's responsible for taking the timestamps from the packet and writing them in the correct format for the subtitle format used.
[19:40:31 CET] <thebombzen> ah
[19:41:09 CET] <thebombzen> kepstin: in other news, I'm trying to dump a frame metadata produced by libavfilter
[19:41:16 CET] <thebombzen> I'm getting this interesting error: Automatic encoder selection failed for output stream #0:0. Default encoder for format ffmetadata (codec none) is probably disabled. Please choose an encoder manually.
[19:41:23 CET] <thebombzen> now I didn't disable codec none on build
[19:41:52 CET] <thebombzen> https://0x0.st/sXIP.txt
[19:42:18 CET] <thebombzen> am I misusing this, or is there something else?
[19:43:49 CET] <kepstin> as the name suggests, "none" is the absense of a codec - i.e. the ffmetadata format doesn't specify a default codec. As a result, you are required to always specify a codec.
[19:44:00 CET] <kepstin> might I suggest using 'rawvideo' or something?
[19:45:33 CET] <kepstin> hmm. that said, i'm kinda surprised that doesn't work
[19:45:37 CET] Action: kepstin looks closer
[19:45:48 CET] <thebombzen> hm, rawvideo works, but ffmetadata expects CFR input
[19:45:52 CET] <thebombzen> and it's duping frames
[19:46:11 CET] <thebombzen> using -vsync passthrough stops duping but then I get spammed about application providing invalid pts
[19:46:47 CET] <thebombzen> er, complains about invalid, nonmonotonically increasing DTS
[19:47:24 CET] <thebombzen> all of which are two numbers which are equal, like this: [ffmetadata @ 0x55e855c8c760] Application provided invalid, non monotonically increasing dts to muxer in stream 0: 11019 >= 11019
[19:48:11 CET] <kepstin> hmm. you said you want *frame* metadata?
[19:48:50 CET] <kepstin> then ffmetadata muxer is the wrong tool, it only does stream-level metadata.
[19:49:36 CET] <thebombzen> yea, I do want frame metadata
[19:49:40 CET] <thebombzen> how do I extract frame metadata?
[19:50:57 CET] <kepstin> thebombzen: use the "metadata" filter with mode=print (it can take a filename to save to)
[20:06:19 CET] <garyserj> Hi, I have these two files, I show you the info from mediainfo.exe for each of them.. one is KC..mp4 the other is some125.mp4   https://pastebin.com/raw/C6MQgCaB   I want the some125.mp4 to have the same codecs as KC..mp4   What command do I need?
[20:14:36 CET] <kepstin> garyserj: they already both have the same codecs
[20:15:26 CET] <thebombzen> kepstin: appears to have worked. now I gotta use some perl regex magic to convert it to SRT so I can then use it in ffmpeg, but that did the trick. thanks
[20:21:43 CET] <faruk> hi guys. i want to run ffmpeg over windows xp. which build should i use ?
[20:22:51 CET] <pmjdebruijn> faruk: presumably win32 static, not sure if that'll still work, considering winxp has been end-of-life, you'll just have to try and see
[20:23:05 CET] <furq> faruk: the zeranoe builds don't work on xp because of libmfx
[20:23:09 CET] <furq> so you probably need to build it yourself
[20:23:19 CET] <faruk> hmm ok thx
[00:00:00 CET] --- Tue Dec 19 2017