[Ffmpeg-devel-irc] ffmpeg.log.20160822

Tue Aug 23 03:05:01 EEST 2016

[00:00:08 CEST] <azonenberg> Filtergraph 'fps=5' was specified through the -vf/-af/-filter option for output stream 0:0, which is fed from a complex filtergraph.
[00:00:08 CEST] <azonenberg> -vf/-af/-filter and -filter_complex cannot be used together for the same stream.
[00:00:20 CEST] <furq> put fps in filter_complex
[00:00:23 CEST] <azonenberg> how do I do that?
[00:00:32 CEST] <furq> ,fps=5
[00:00:56 CEST] <azonenberg> Does that change playback speed, drop frames, both?
[00:01:09 CEST] <azonenberg> My goal is to drop four out of every five frames, and play the surviving frames for 5x as long
[00:01:13 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#fps-1
[00:01:48 CEST] <furq> the speed will remain the same unless you use setpts
[00:02:17 CEST] <azonenberg> That's an output codec option right?
[00:02:23 CEST] <furq> that's a filter
[00:02:53 CEST] <furq> it sounds like you don't want to use it though
[00:03:50 CEST] <azonenberg> My goal is to make the video play just like the original
[00:03:53 CEST] <furq> right
[00:03:54 CEST] <azonenberg> but more jittery
[00:03:56 CEST] <furq> fps will do that
[00:03:57 CEST] <azonenberg> and blockier
[00:04:13 CEST] <azonenberg> While not using petabytes of storage if i want to keep a long tail of video
[00:04:18 CEST] <azonenberg> Basically, the way I see it
[00:04:29 CEST] <azonenberg> the probability of me needing to view security footage drops off the longer ago the event happened
[00:04:32 CEST] <furq> what h264 encoder are you using to get the original videos
[00:04:43 CEST] <azonenberg> But if I misplace a tool in the shop or something, it'd be nice to go back to the last time i used it
[00:04:45 CEST] <azonenberg> and see where i left it
[00:04:47 CEST] <furq> if it's some hardware encoder you could potentially save a bunch of space by just running them through x264 at a slow preset
[00:04:50 CEST] <azonenberg> but i don't need high quality 30 FPS for that :p
[00:05:04 CEST] <azonenberg> The original source is an IP camera that lets me configure bit rate and framerate
[00:05:18 CEST] <azonenberg> I *want* high quality for the live feed, and for the first day or so of footage
[00:05:25 CEST] <azonenberg> then have a cron job to squish older footage
[00:05:26 CEST] <furq> i'd imagine it's incredibly inefficient then
[00:05:37 CEST] <azonenberg> I'd rather have a month of 5 FPS low-bit-rate video
[00:05:44 CEST] <azonenberg> than a few days of 30 FPS cinematic quality
[00:05:50 CEST] <azonenberg> With maybe a day of great quality video at the end
[00:05:54 CEST] <furq> sure
[00:06:10 CEST] <furq> i'm just pointing out that you could potentially have a month of 30fps video of roughly the same quality
[00:06:25 CEST] <azonenberg> Well, i'm open to suggestions
[00:07:21 CEST] <furq> try reencoding a chunk with libx264 at default settings
[00:07:25 CEST] <azonenberg> http://thanatos.virtual.antikernel.net/unlisted/1436_38.mp4
[00:07:38 CEST] <azonenberg> This is one of my test chunks, a 30 sec capture with the camera pointing out my office window down the block at the neighbor's house
[00:07:47 CEST] <azonenberg> i have a directory full of a dozen or so of these
[00:08:00 CEST] <azonenberg> in a real deployment the chunks would be more like 30-60 minutes each
[00:08:25 CEST] <azonenberg> I can certainly try dropping the bit rate on the camera, this is whatever it shipped with
[00:08:53 CEST] <azonenberg> My capture right now is ffmpeg -i $RTSP_URL -r 25 -c:v copy
[00:09:06 CEST] <azonenberg> So it doesn't transcode, it just takes whatever the camera feeds me
[00:11:07 CEST] <furq> huh
[00:11:10 CEST] <furq> x264 doesn't like this at all
[00:11:37 CEST] <furq> neither does ffmpeg's h264 decoder, for that matter
[00:11:40 CEST] <azonenberg> the actual scene this camera is being deployed on is going to be an indoor shop environment, so actual compression ratios will vary
[00:11:47 CEST] <azonenberg> that was generated by ffmpeg
[00:11:56 CEST] <azonenberg> i get lots of warnings when i'm converting
[00:11:58 CEST] <azonenberg> but it plays
[00:11:58 CEST] <furq> [h264 @ 00000000003eee20] Exceeded next NALFF position, re-syncing.
[00:12:02 CEST] <furq> is it that warning
[00:12:11 CEST] <azonenberg> Invalid UE golomb code31.0 size=    5679kB time=00:00:52.80 bitrate= 881.1kbits/s speed=6.44x
[00:12:12 CEST] <azonenberg> [h264 @ 0x4fa82e0] pps_id 3199971767 out of range
[00:12:17 CEST] <azonenberg> Invalid UE golomb code31.0 size=    5679kB time=00:00:52.80 bitrate= 881.1kbits/s speed=6.44x
[00:12:17 CEST] <azonenberg> [h264 @ 0x4fa82e0] pps_id 3199971767 out of range
[00:12:21 CEST] <azonenberg> among many other things
[00:12:34 CEST] <azonenberg> My guess is the camera's codec is iffily compliant
[00:12:37 CEST] <azonenberg> but i kinda have to work with what i have
[00:13:13 CEST] <furq> i recompressed it with x264 and it came out double the size
[00:13:16 CEST] <azonenberg> lol
[00:13:19 CEST] <furq> which is unexpected
[00:13:32 CEST] <azonenberg> So you're saying the scene compresses poorly, the file is invalid, both?
[00:18:30 CEST] <azonenberg> Somewhat related, i'm having trouble setting the output bit rate
[00:18:51 CEST] <azonenberg> as an extreme example i'm passing -b:v 50 and getting 2 Mbps instead of 50 Kbps output bit rate
[00:19:00 CEST] <azonenberg> so i assume that means i should be specifying it somewhere in the filter_complex command?
[00:19:37 CEST] <furq> well 50 would be 50 bits per second
[00:19:42 CEST] <furq> so i'm not surprised it's having trouble hitting it
[00:20:01 CEST] <azonenberg> sorry i meant 50k
[00:20:09 CEST] <azonenberg> i was passing 50k
[00:20:56 CEST] <furq> i tried at 720p5 at crf 23 and it's still 3/4 the size of the original
[00:21:01 CEST] <furq> i have no idea how that's possible
[00:21:39 CEST] <azonenberg> is my input that badly malformed? lol
[00:21:58 CEST] <furq> i'm not enough of an h264 expert to say
[00:22:07 CEST] <furq> but suffice to say i'd have expected it to be a bit smaller than that
[00:30:48 CEST] <azonenberg> libx264 recompresses it at a terrible speed
[00:30:57 CEST] <azonenberg> like, it's taking several seconds to compress each frame
[00:30:59 CEST] <azonenberg> is that usual?
[00:31:07 CEST] <furq> it's pretty quick here
[00:31:21 CEST] <furq> also i think it's just an incompressible source because of all the blocking
[00:31:34 CEST] <furq> it seems pretty much fine after downscaling (a lot)
[00:31:39 CEST] <furq> but that's not really ideal
[00:31:49 CEST] <azonenberg> I should probably try installing the camera
[00:31:59 CEST] <azonenberg> those trees are a pain in the butt to compress
[00:32:02 CEST] <azonenberg> and there's no trees in the actual alb
[00:32:04 CEST] <azonenberg> lab*
[00:32:22 CEST] <furq> there's a lot of weird blocking happening on the roof which i think is throwing it off
[00:32:30 CEST] <azonenberg> Yeah
[00:32:33 CEST] <furq> although yeah that might be because the hardware encoder is running out of bits from the tree movement
[00:32:45 CEST] <azonenberg> yeah the input video is ~8 Mbps now
[00:32:49 CEST] <azonenberg> i told it to use 4
[00:32:58 CEST] <azonenberg> Well, i'll go install the camera in the actual ceiling indoors and see how it works
[00:33:03 CEST] <azonenberg> May or may not be back with more problems
[00:33:11 CEST] <azonenberg> Gotta find a spot to plug in my PoE injector...
[00:33:32 CEST] <furq> well yeah for mostly static scenes you should (emphasis on should) get a massive bitrate drop from reprocessing with x264
[00:33:37 CEST] <furq> without doing anything else
[00:34:10 CEST] <furq> if that's not enough then downscaling to 720p will probably give more benefit than dropping the framerate
[00:34:25 CEST] <furq> although it'll take longer
[00:35:27 CEST] <azonenberg> http://thanatos.virtual.antikernel.net/unlisted/1471764763.jpg
[00:35:35 CEST] <azonenberg> well, as long as i get better than this quality
[00:35:39 CEST] <azonenberg> at 1 frame per 3 sec
[00:35:46 CEST] <azonenberg> you can see why i'm replacing my existing cameras :p
[00:36:20 CEST] <azonenberg> 720p, horribly grainy, no IR capability so i have to leave the lights on all the time and waste electricity, and uselessly low framerate
[00:36:55 CEST] <azonenberg> So, the bar isn't that high lol
[00:37:03 CEST] <furq> the font is nicer though
[00:37:28 CEST] <azonenberg> Yeah this is my own imagemagick overlay
[00:37:36 CEST] <azonenberg> this is a bunch of usb webcams
[00:37:43 CEST] <azonenberg> and a hackjob of PHP shell scripts running on a linux box
[00:37:45 CEST] <azonenberg> with usb hubs in the ceiling
[00:37:48 CEST] <azonenberg> it's a disaster :p
[00:37:56 CEST] <furq> i'm glad you said it
[00:38:00 CEST] <azonenberg> the new setup is all PoE IP cams, i just need a postprocessor for the data
[02:42:12 CEST] <witchdoctor> hello
[03:10:00 CEST] Last message repeated 1 time(s).
[05:35:00 CEST] Last message repeated 1 time(s).
[11:29:49 CEST] <ozette> what's it mean exactly when the ffmpeg binary says: ./ffmpeg: /lib/libc.so.6: version `GLIBC_2.17' not found (required by ./ffmpeg)?? that this ffmpeg was build with a different version of glibc?
[11:55:47 CEST] <Rabeet> hello
[11:57:02 CEST] <Rabeet> anyone?
[11:59:11 CEST] <Rabeet> i'm using ffmpeg with screen-capture-recorder and recording my screen and pushing it to facebook using live_videos api. When i open up any game like counter strike 1.6 or quake3 which uses openGL nothing displays grey screen apears. also not working with full screen.
[12:14:51 CEST] <BtbN> You can't easily capture hardware-accelerated stuff like Games and Exclusive-Fullscreen-Applications.
[12:15:05 CEST] <BtbN> Why not use a tool like OBS, that's made for exactly that job?
[12:17:24 CEST] <nonex86> <nonex86> why dont you just use obs for example?
[12:17:25 CEST] <nonex86> :D
[12:26:06 CEST] <Rabeet> Yes i can use OBS but i have some different business model
[12:26:48 CEST] <Rabeet> Any work around fixing openGL issue?
[13:08:45 CEST] <BtbN> No.
[13:09:07 CEST] <BtbN> You need an entirely different approach, which is simply not implemented in ffmpeg, and most likely never will be.
[14:36:03 CEST] <Spring> is there a straightforward way to ensure that the trim end value is greater than the trim start?
[14:36:41 CEST] <Spring> as both can have a variety of different timecode formats and it's hard to directly compare values
[14:37:16 CEST] <Spring> ffplay for example doesn't even launch a preview if the trim values are identical, it just stalls in its 'streaming' mode
[14:38:25 CEST] <Spring> if it produced some error I could at least trigger a conditional operation
[14:43:06 CEST] <simong> Hi, got a compilation problem with libvpx on CentOS 6, is this a good place to ask?
[14:50:32 CEST] <spaam> simong: maybe #vp8 might be better for that :)
[15:00:13 CEST] <simong> spaam: thanks :)
[15:47:51 CEST] <viric> wouldn't it be trivial to patch firefox-on-linux to use an h264 hw decoder?
[15:48:44 CEST] <DHE> possibly, but you'd have to provide one. and hardware decoders tend to be platform-specific so you really want some kind of flexible or plugin system... and then you get outside the scope of ffmpeg.
[15:49:24 CEST] <viric> be it vdpau
[15:53:45 CEST] <viric> firefox simply calls some ffmpeg lib methods
[15:54:27 CEST] <viric> https://github.com/mozilla/gecko-dev/blob/master/dom/media/platforms/ffmpeg/FFmpegVideoDecoder.cpp
[15:55:49 CEST] <apus> hi, could somebody please tell me whether there are video codecs out there that produce sparse files?
[15:56:52 CEST] <apus> i have a filesystem issue where sparse files are concerned, which may or may not have been corrupted. i'd like to know if i need to worry about video/audio files
[16:08:24 CEST] <kdehl> Greetings. I'm trying to decode a H.264 stream using OpenH264. I call DecodeFrame2 on my decoder object, send it an input h264 buffer, an array of YUV output buffers etc, and it works perfectly.
[16:09:12 CEST] <kdehl> Question is what do I do with the second frame? I can't seem to find out what offset in the buffer to feed DecodeFrame2() again.
[16:10:09 CEST] <kdehl> I assume the codec reads as many NAL units as it needs for decoding the first frame. Maybe I should send a pointer to the subsequent NAL unit for the next call?
[16:10:36 CEST] <kdehl> But then question is, how do I know how many NAL units has the decoder read in the first call?
[16:11:06 CEST] <kdehl> Am I on the right track, or have I misunderstood it completely?
[16:11:23 CEST] <DHE> apus: only if you enable compression and have a lot of zero bytes in a row that cover a block as per the recordsize parameter
[16:13:47 CEST] <apus> DHE: hm i'm not familiar with codecs. aren't there optimizers run, that can do such a compression?
[16:14:12 CEST] <apus> is this typical in "modern" video files/codecs you have every day?
[16:14:41 CEST] <nonex86> kdehl, never used openh264, but i dont understand the problem, why you dont just feed your decoder with nalu? when the frame is ready it should return it to you, everyone is happy
[16:18:53 CEST] <kdehl> nonex86: It does with the first one. But how do I encode a second frame?
[16:18:53 CEST] <DHE> apus: it's pretty rare that codec output is itself compressible. codecs can't generate sparse files themselves, so you're relying on the compression of ZFS to punch your holes. and I'm pretty sure it's not possible for H264 to output such a huge zero string in an equally huge AVPacket
[16:20:10 CEST] <nonex86> kdehl, i am sorry, we talk about encoding or decoding?
[16:20:27 CEST] <kdehl> nonex86: Decoding. Sorry.
[16:20:34 CEST] <kdehl> Decoding h.264 to YUV.
[16:26:59 CEST] <apus> DHE: thanks for the help!
[17:02:28 CEST] <ozette> hey my target is running on ffmpeg v2.7.1, now i want to cross compile v3.1.2
[17:02:43 CEST] <ozette> when i print the version of 2.7.1 i can see the configure options
[17:03:15 CEST] <ozette> is it wise to use the same?
[17:05:56 CEST] <furq> if you want the same features available
[17:07:53 CEST] <ozette> the target actually disables a lot of stuff, like ffserver
[17:08:00 CEST] <ozette> but that's ok
[17:08:13 CEST] <furq> you should disable that anyway because it sucks
[17:08:28 CEST] <ozette> lol, what's it do?
[17:08:47 CEST] <ozette> (i only need ffmpeg and ffprobe)
[17:08:55 CEST] <furq> it doesn't do anything as far as i can tell
[17:09:05 CEST] <furq> other than generate doomed support questions in here
[17:09:34 CEST] <ozette> lmao
[17:09:40 CEST] <ozette> ouch
[17:23:02 CEST] <ozette> if i don't set the --prefix and --incdir for configuration, ffmpeg should install in the source directory, right?
[17:23:33 CEST] <ozette> and i don't think i need the --cross-prefix flag
[17:23:59 CEST] <ozette> i can just provide the compiler with the make command
[17:27:14 CEST] <DHE> it's more complex than that. sometimes you need to build a build-helper application and you need a distinction between host and target compilers. you may need an assembler to handle target assembly code, etc.
[17:28:50 CEST] <furq> if you don't set --prefix then i'm pretty sure it'll install in /usr/local
[17:29:13 CEST] <furq> if you just want it built in the source directory then don't run make install
[17:29:52 CEST] <ozette> i'd like it in the source
[17:30:04 CEST] <ozette> so.. i should provide a prefix then
[17:30:11 CEST] <furq> no
[17:30:14 CEST] <furq> read the other thing i said
[17:30:40 CEST] <ozette> ah rght
[17:30:53 CEST] <ozette> just a make..
[17:41:10 CEST] <Chloe> how can I get a AV_CODEC_ID from a fourcc tag?
[17:41:59 CEST] <alcros> Hey, i have a H264-encoded video and i want only a sequence in the middle. When i use the CLI-parameters -ss and -to, the new video starts at the next keyframe after the position specified with -ss and ends at the previous keyframe of the -to position.....What i want is, that the video starts at the previous keyframe from the -ss keyframe and ends at the next keyframe from -to position...is there a CLI-switch or similar to
[17:41:59 CEST] <alcros> archieve this? Thank you
[17:42:51 CEST] <durandal_1707> With -c copy?
[17:42:57 CEST] <alcros> durandal_1707, yes
[17:43:38 CEST] <durandal_1707> Chloe: by inspecting source code :)
[17:44:59 CEST] <durandal_1707> alcros: afaik without keyframe, its not possible with ffmpeg
[17:47:14 CEST] <alcros> durandal_1707: what do you mean with without keyframe?
[17:47:41 CEST] <alcros> Let's say i have a keyframe every 5 seconds (only a example, in my input video the keyframes are not regular)...when i use -ss 2 -to 12 i get 5-10, but what i want is 0-15
[17:47:49 CEST] <durandal_1707> keyframe at exact pos you cut
[17:48:33 CEST] <durandal_1707> ahh, than open feature request on Trac
[17:48:43 CEST] <kepstin> alcros: hmm. ffmpeg already has some of the logic for that, because when *not* in copy mode, it'll find the previous keyframe, then decode and throw out frames to the desired seek point
[17:48:53 CEST] <kepstin> it's probably a fairly simple change :/
[17:49:32 CEST] <ozette> hmm
[17:49:37 CEST] <ozette> i installed faac-devel
[17:49:45 CEST] <ozette> but configure says i don't have libfaac
[17:51:45 CEST] <ozette> can i just drop faac?
[17:51:56 CEST] <durandal_1707> libfaac gonna be removed anyway
[17:52:11 CEST] <kepstin> ozette: the ffmpeg built-in aac encoder in ffmpeg >=3 is probably comparable to or better than faac anyways
[17:52:21 CEST] <ozette> i see
[17:52:26 CEST] <ozette> thanks guys :0
[17:52:29 CEST] <ozette> :)
[17:54:08 CEST] <alcros> durandal_1707,kepstin: ok, thank you
[17:57:13 CEST] <Chloe> durandal_1707: great. thanks.
[17:57:34 CEST] <Chloe> I saw av_codec_get_id() but couldnt get it working, oh well.
[17:59:35 CEST] <ozette> this is such a trial and error process to me
[18:00:50 CEST] <ozette> removing --enabled libs as they're not found
[18:16:10 CEST] <Rabeet> hello
[18:16:52 CEST] <Rabeet> can anyone tell me which (opengl/directx3d screen recorders) ffmpeg supports?
[18:17:36 CEST] <__jack__> x11 is supported
[18:17:39 CEST] <__jack__> at least
[18:17:43 CEST] <__jack__> dunno about the rest
[18:18:50 CEST] <__jack__> Rabeet: ffmpeg -devices
[18:22:20 CEST] <Rabeet> __jack__: Yes asking about directShow recorders? i know about screen-capture-recorder and UScreenCapture but they are not working with openGL
[18:22:27 CEST] <Rabeet> can you suggest any other?
[18:23:16 CEST] <__jack__> I suggest x11grab, as it's the only one I vaguely know about
[20:18:55 CEST] <ozette> The v3.1.2 I built says: "ffmpeg: error while loading shared libraries: libavdevices.so.57: cannot open shared object file: No such file or directory"
[20:19:05 CEST] <ozette> but i configured with --enable-static
[20:19:43 CEST] <ozette> shouldn't that bundle the code of required libraries into the program?
[20:21:52 CEST] <flux> perhaps. are you certain this particular ffmpeg is the one you built now, did you build it from and did you disable shared libraries?
[20:22:24 CEST] <flux> "from clean" I meant to write
[20:23:19 CEST] <ozette> yes i'm certain, because it's the one i built, because i brought it over to my workstation and just now executed the program with ./ffmpeg
[20:23:34 CEST] <furq> pastebin the output of ffmpeg -version
[20:23:45 CEST] <furq> or the configure string you used if it doesn't get that far
[20:24:20 CEST] <ozette> furq: the error i just posted is the output of ffmpeg -version, one moment and i'll show you my configure options
[20:24:34 CEST] <ozette> flux: not sure if i disabled shared libraries .. now that you say..
[20:25:12 CEST] <furq> enable-shared is off by default
[20:26:19 CEST] <ozette> these are the options: https://paste.fedoraproject.org/412583/90363147/
[20:26:43 CEST] <furq> "--enable-shared --disable-static"
[20:26:43 CEST] <ozette> oh wow
[20:26:58 CEST] <ozette> so i should drop- --enable-shared?
[20:27:04 CEST] <furq> you should swap both of those around
[20:27:24 CEST] <furq> alternatively you could just move the .sos to the other machine
[20:27:40 CEST] <ozette> so.. disable-shared en enable-static?
[20:27:43 CEST] <furq> sure
[20:27:48 CEST] <furq> or just remove them both as that's the default
[20:27:56 CEST] <ozette> i see interesting
[20:28:22 CEST] <furq> you can also get rid of enable-encoders and enable-optimizations
[20:28:33 CEST] <ozette> i must have messed up somewhere, this config is so weird
[20:28:43 CEST] <furq> yeah you don't need 99% of that
[20:28:58 CEST] <ozette> it says --disable-static at one point and --enable static later on
[20:29:05 CEST] <ozette> yea well
[20:29:29 CEST] <furq> http://vpaste.net/d4j60
[20:29:38 CEST] <furq> that's the entirety of my cross-compile build flags
[20:30:10 CEST] <ozette> i have two workstations, one's on ffmpeg 3.1.2 already, the other one at 2.7.1, i acquired a toolchain for my platform and smashed their config options together and tweaked some thing
[20:30:26 CEST] <ozette> sweet
[20:31:13 CEST] <furq> that looks like it's been adapted from one of those weird distro packages which tries to remove everything patented
[20:31:49 CEST] <furq> i'm guessing fedora via centos
[20:31:53 CEST] <ozette> it only builds ffmpeg
[20:32:14 CEST] <ozette> so it's pretty fast
[20:32:32 CEST] <ozette> and it is a weird distro
[20:32:35 CEST] <ozette> it's a synology
[20:32:39 CEST] <furq> oh man
[20:32:43 CEST] <ozette> yea..
[20:34:33 CEST] <ozette> anyway thanks, new try!
[22:48:55 CEST] <flyBoi> Any node.js fluent-ffmpeg pros around?
[22:49:09 CEST] <flyBoi> I'm having trouble concatenating three video files into one :-(
[22:49:27 CEST] <flyBoi> .mergeToFile() doesn't seem to play nice with > 2 videos
[22:49:55 CEST] <BtbN> ffmpeg does not support any JavaScript, let alone node.js.
[22:50:10 CEST] <BtbN> So you should ask whoever made that mergeToFile
[22:50:19 CEST] <BtbN> Or use ffmpeg directly.
[22:50:57 CEST] <flyBoi> K will do.
[23:01:03 CEST] <Lal> Alright, so, I'd like to ask for help converting an MKV using the h.264 codec to an AVI using the mpeg4 codec. Can I have some assistance, please?
[23:24:30 CEST] <c_14> Lal: ffmpeg -i mkv -c:v mpeg4 -c:a copy out.avi
[23:32:51 CEST] <ozette> flyBoi: i'm actually using fluent-ffmpeg in my application, but you better ask the author he's very active, or #Node.js
[23:33:10 CEST] <ozette> i've never merged video files
[23:39:24 CEST] <blue_misfit> hey with libx264, I usually set scenecut=0 when I want to use fixed GOP
[23:39:31 CEST] <blue_misfit> I'm seeing some guides say scenecut=-1
[23:39:35 CEST] <blue_misfit> and had never seen that before now
[23:39:48 CEST] <blue_misfit> any difference between setting this to -1 vs setting it to 0?
[23:41:37 CEST] <ianbytchek> greetings all. trying to use ffmpeg libs in osx app, tried both objective-c / swift. profiling shows noticeable memleaks originating in av_mallocz, av_strdup, av_malloc& code is from official example, everything gets properly released / cleaned up. couldn't google a solution / explanation, but this seems a very common issue. is this a false positive or is there a way to fix
[23:41:37 CEST] <ianbytchek> this? profiling details  https://www.dropbox.com/s/p184qttoibduajd/Screenshot%202016-08-23%2001.41.12.PNG?dl=0
[23:42:20 CEST] <ozette> hmm
[23:45:14 CEST] <markg13> hey guys I am trying to convert 5 hours of avi files (32GB) to a small format, I used lbx264 and It has taken more than 12 hours on a iMac with Corei7 and 16GB ram, and still has not finished
[23:45:42 CEST] <markg13> here is the command I am using, ffmpeg -i /input/$f -c:v libx264 -crf 23 -threads 6 /output/$f.mp4
[23:45:43 CEST] <blue_misfit> markg13, this might be totally expected if you used slow settings
[23:45:50 CEST] <blue_misfit> ok
[23:45:54 CEST] <blue_misfit> yeah that uses fairly slow settings
[23:45:58 CEST] <markg13> @blue_misfit how can I make it fast settings
[23:46:04 CEST] <blue_misfit> if you want faster settings just add -preset superfast
[23:46:15 CEST] <markg13> is it gonna affect quality?
[23:46:19 CEST] <blue_misfit> yes
[23:46:22 CEST] <blue_misfit> but not HUGELY
[23:46:29 CEST] <blue_misfit> increase the CRF value to compensate :)
[23:46:34 CEST] <blue_misfit> like set crf 20 or something instead of 23
[23:46:38 CEST] <markg13> do I add it to the end of the line
[23:46:47 CEST] <blue_misfit> I'd suggest after -c:v libx264
[23:46:47 CEST] <markg13> is the codec I am using the best one ? libx264 ?
[23:46:53 CEST] <blue_misfit> what kind of avi files are you working with
[23:47:03 CEST] <blue_misfit> post the output of ffmpeg $inputFile.avi
[23:47:07 CEST] <markg13> they are old mpeg1 I think
[23:47:08 CEST] <blue_misfit> post the output of ffmpeg -i $inputFile.avi
[23:47:18 CEST] <blue_misfit> mpeg1 in AVI? that would be highly irregular :)
[23:47:45 CEST] <blue_misfit> https://trac.ffmpeg.org/wiki/Encode/H.264
[23:48:23 CEST] <markg13> @blue_mist https://gist.github.com/medyagh/d5ba789161725cf42860a149570e1ac2
[23:48:30 CEST] <markg13> thats the input format
[23:50:35 CEST] <blue_misfit> that shows me your ffmpeg build but not info about the file
[23:50:37 CEST] <blue_misfit> need more info
[23:51:12 CEST] <markg13> I did ffmpeg -i
[23:51:24 CEST] <blue_misfit> ffmpeg -i $inputfile.avi ?
[23:51:46 CEST] <markg13> oh hold it is inside a docker I need to attach it
[23:51:48 CEST] <markg13> sorry
[23:54:33 CEST] <markg13> @blue_misfit https://gist.github.com/medyagh/979e329a80c51f119f005a189dfebed3
[23:55:34 CEST] <blue_misfit> Stream #0:0: Video: mjpeg (MJPG / 0x47504A4D), yuvj420p(pc, bt470bg/unknown/unknown), 1440x1080, 15397 kb/s, 30 fps, 30 tbr, 30 tbn, 30 tbc
[23:55:44 CEST] <blue_misfit> k so it's MJPG, 1440x1080p30
[23:56:00 CEST] <blue_misfit> encoding that on your system should be somewhat close to realtime on the default settings
[23:56:27 CEST] <blue_misfit> do like ffmpeg -i $input -c:v libx264 -preset superfast -crf 20 -c:a aac $output.mp4 and you should be fine
[23:56:50 CEST] <blue_misfit> if the quality isn't good enough use a lower crf or a slower preset. the link I sent you earlier has details on the different preset options
[23:57:14 CEST] <Lal> Thanks, C_14
[23:57:22 CEST] <Lal> Err, c_14
[23:57:32 CEST] <blue_misfit> you owe me a beer now, @markg13
[23:57:37 CEST] <Lal> Seems to be doing well.
[23:59:51 CEST] <markg13> thanks @blue_misfiti
[00:00:00 CEST] --- Tue Aug 23 2016