[Ffmpeg-devel-irc] ffmpeg.log.20160830

[burek]

[00:00:53 CEST] <deweydb> llogan: gotcha! the -preset ultrafast seems to have helped quite a bit.
[00:02:31 CEST] <deweydb> llogan: http://pastebin.com/74V5RrnM
[00:03:30 CEST] <deweydb> not sure if this can be optimized further?
[00:04:44 CEST] <llogan> i mostly wanted to make sure you weren't using --disable-(y)asm. you could try a modern FFmpeg and see if it makes a difference.
[00:30:53 CEST] <Hitechcg> remind me to unalias ffmpeg if I ever ask for help
[00:31:08 CEST] <Hitechcg> I have all the ffmpeg commands aliased to (command) -hide_banner
[02:08:30 CEST] <Victorma> Hello! I am getting "Connection to ... failed: Error number 26 occurred" inside avformat_open_input. Anyone knows what does it mean?
[02:08:51 CEST] <memeka> hi; how can i convert AVFrame/AVPacket to annexB?
[02:16:31 CEST] <Kadigan> Hello, fair folk of the ffmpeg land. :)
[02:17:13 CEST] <DHE> memeka: if you're talking h264/265 for copying, there's a bsf for that...
[02:17:37 CEST] <DHE> Victorma: what OS?
[02:19:44 CEST] <Victorma> DHE: lets say none, it's happening inside a 3DS
[02:19:59 CEST] <memeka> DHE: yes i know how to do it with ffmpeg
[02:20:28 CEST] <memeka> but how to in code -> i am trying to feed a hw decoder only annexB frames (or else it's not decoding) ...
[02:21:14 CEST] <memeka> DHE: like, in here: https://github.com/mihailescu2m/FFmpeg/commit/cca21ca457da2c921d16b5ffb29d6b5eef3d2c73#diff-66050eef83e6968ba6c0146084e5c8e9R541 - how can i do it?
[02:23:41 CEST] <DHE> memeka: based on ffmpeg cli code, I'd start here: https://ffmpeg.org/doxygen/3.0/group__lavc__misc.html#ga4f87a57cd7c08c1d9b9bfe1a4bd611f3
[02:26:18 CEST] <memeka> DHE: so i need to do that actually outside of avpriv_v4l_enqueue_frame_or_pkt_or_buf, such that avpriv_v4l_enqueue_frame_or_pkt_or_buf will receive annexb already, correct?
[02:26:49 CEST] <memeka> i won't be converting frame in there, but have a filter higher up that converts the entire stream, before feeding it
[02:28:00 CEST] <DHE> well, I'm looking into the code some more and there are a few API changes in newer versions of ffmpeg, so the exact function sequence varies...
[02:28:24 CEST] <DHE> but you should be able to just pass in the AVPacket (or the payload thereof, still not sure which function to use) and you'll get the annexB version back out
[02:31:16 CEST] <memeka> DHE: thx, i'll try
[03:34:48 CEST] <kyleogrg> is yadif + mcdeint much better than just yadif?  is there a better free deinterlacer that can be used with ffmpeg?
[03:35:03 CEST] <BigWoop> arighgt fuck it (o°¡°	o5 ;
; ffmpeg -i  "$INDIR/$@" -i $LOGODIR/watermark.png -filter_complex 'overlay=x=main_w-overlay_w-5:y=main_h-overlay_h-5' -strict -2 "$OUTDIR/$@"
[03:35:17 CEST] <BigWoop> and I'm getting [png_pipe @ 0x3744920] Could not find codec parameters for stream 0 (Video: png, none): unspecified size
[03:35:17 CEST] <BigWoop> Consider increasing the value for the 'analyzeduration' and 'probesize' options
[03:35:36 CEST] <BigWoop> anyoen have a clue cause I'm grittin the teeth over here lol
[03:36:40 CEST] <deweydb> is there syntax for the ffmpeg concat demuxer that does not require a file. i.e. a commandline version? instead of -i somefiles.txt perhaps -i filename.mp4 -file2.mp4 etc?
[03:41:24 CEST] <Kadigan> deweydb: You can always string together multiple inputs and multiple outputs...
[03:41:51 CEST] <deweydb> you mean using a filter?
[03:43:11 CEST] <Kadigan> In the specific case of concat,
[03:43:33 CEST] <deweydb> yeah, but a concat filter is much slower than a concat demuxer
[03:43:34 CEST] <Kadigan> I'm seeing `-i "contact:file1.mp4|file2.mp4|file3.mp4"`
[03:43:37 CEST] <Kadigan> Ah.
[03:43:48 CEST] <Kadigan> concat: *
[03:44:10 CEST] <deweydb> its ok. i'm just gonna script it to create that file. and then delete the file afterwards.
[03:44:16 CEST] <deweydb> its a bit clunky but oh well
[03:44:24 CEST] <Kadigan> Have You tried < "" ?
[03:44:37 CEST] <Kadigan> I never did that stuff with ffmpeg before, to be honest...
[03:44:51 CEST] <deweydb> yeah, i'm waaaaay down the rabbit hole
[03:45:02 CEST] <deweydb> its funny cause what i'm doing ought to be pretty simple
[03:45:08 CEST] <deweydb> but its surprisingly difficult.
[03:45:15 CEST] <Kadigan> Funny, yes.
[03:45:17 CEST] <Kadigan> Funny.
[03:45:19 CEST] <Kadigan> >.>
[03:45:31 CEST] <Kadigan> "Been there, done that, got the scars." comes to mind
[03:45:37 CEST] <deweydb> lol
[03:45:57 CEST] <Kadigan> (if You catch the reference, I love You ;))
[03:47:50 CEST] <Kadigan> If You ever need another challenge like this one, try changing a video framerate w/o modifying frames (either dropping or duplicating) in a codec that doesn't support timestamps.
[03:52:17 CEST] Action: deweydb shudder
[03:53:29 CEST] <Kadigan> I mean, it's basically a case of flipping the FPS bit, so to speak, but ffmpeg seems unable to do it (maybe because it's a hack?)
[03:54:03 CEST] <Kadigan> Well, I base my theory on the fact that it took a second for multi-GB files in a small app once, but ffmpeg takes its sweet time to recode PTS.
[03:54:59 CEST] <ytan> hey guys
[03:55:31 CEST] <Kadigan> Hello.
[03:55:59 CEST] <ytan> I'm having problems trying to stream using TLS protocol
[03:56:10 CEST] <Kadigan> I'm not the person to ask.
[03:56:12 CEST] <ytan> Not sure if this is the place to ask...
[03:56:15 CEST] <ytan> hahha
[03:56:31 CEST] <ytan> been stuck for a few days now
[03:56:33 CEST] <Kadigan> No, seriously. I don't know anything about it. Wanted to save You the trouble of asking me specifically, since I'm active. ;)
[03:56:39 CEST] <ytan> I think my sanity is slipping away
[03:56:49 CEST] <ytan> no that's cool
[03:56:49 CEST] <Kadigan> Oh, THAT I know plenty about,
[03:56:53 CEST] <Kadigan> in conjuction with ffmpeg :D
[03:56:55 CEST] <ytan> I just needed someone to talk to
[03:57:10 CEST] <Kadigan> Go ahead, ask deweydb about his remuxing issues.
[03:57:12 CEST] <Kadigan> Just ask. ;D
[03:57:15 CEST] <ytan> and commiserate with
[03:58:19 CEST] <ytan> if you don't mind me asking, what do you use ffmpeg for?
[03:58:49 CEST] <Kadigan> Though I suppose this isn't a channel that takes kindly to commiserating, I suppose. But You can find help with various issues here, most of the time.
[03:59:10 CEST] <Kadigan> Me? Hm, mostly recoding video.
[03:59:11 CEST] <Kadigan> :D
[03:59:42 CEST] <Kadigan> But seriously -- mostly for processing of incoming media (so it's a process I control, not my software), and then producing deliveries when post is done.
[04:00:02 CEST] <Kadigan> That, at work.
[04:00:17 CEST] <Kadigan> At home, I just use it to put all of my media into one, standard format (typically mkv and flac).
[04:00:26 CEST] <Kadigan> (where it's sane to do so, anyway)
[04:00:45 CEST] <ytan> I see
[04:00:55 CEST] <ytan> have you considered using webm format?
[04:01:21 CEST] <Kadigan> MKV has ordered chapter support. I don't know about webm.
[04:01:42 CEST] <Kadigan> Can webm contain more than audio/video streams - stuff like subs, fonts, scripts etc.?
[04:01:49 CEST] <ytan> ordered chapter support? is that like bookmarking in a video?
[04:02:22 CEST] <ytan> good question. I don't know.
[04:02:25 CEST] <Kadigan> I think You could call it that, though it's not user-generated but rather pre-made (ie. you specify them at mux).
[04:02:40 CEST] <Kadigan> It also supports linked chapters,
[04:02:55 CEST] <Kadigan> so you can put a common part of multiple video files (like a TV series intro) into its own file and just link against it.
[04:03:49 CEST] <ytan> wow. that's neat.
[04:04:01 CEST] <Kadigan> So all in all, even despite the fact that I don't put such stuff in myself,
[04:04:16 CEST] <Kadigan> I do watch a lot of stuff that uses that, and I like to have a certain order to my media. So... MKV.
[04:05:18 CEST] <Kadigan> As for webm, I do sometimes find myself producing such files when I need to put something up on the web, in some player or other. But that doesn't happen all that often. I usually just put up an .mp4 file coded for streaming.
[04:05:36 CEST] <ytan> turns out webm is considered a profile in Matroska using VP8 and Vorbis: https://en.wikipedia.org/wiki/Comparison_of_video_container_formats
[04:06:08 CEST] <Kadigan> Seems webm is a subset of Matroska, yeah.
[04:06:28 CEST] <Kadigan> Or, based on.
[04:06:38 CEST] <Kadigan> Says so on webmproject.org.
[04:07:54 CEST] <ytan> so it seems
[04:08:15 CEST] <Kadigan> And now if You'll excuse me, I have some watching before work. :D
[04:08:38 CEST] <ytan> sure
[04:08:42 CEST] <ytan> nice talking to you
[04:12:45 CEST] <Kadigan> Likewise. :)
[04:23:26 CEST] <BigWoop> Does anyone understand [png_pipe @ 0x3744920] Could not find codec parameters for stream 0 (Video: png, none): unspecified size Consider increasing the value for the 'analyzeduration' and 'probesize' options
[04:23:45 CEST] <BigWoop> cause I'm a nub anything ffmpeg related and could use anyones help
[04:30:18 CEST] <memeka> hi, i'm converting each AVPacket to h264_annexb before decoding, but as a result I get:  Packet header is not contained in global extradata, corrupted stream or invalid MP4/AVCC bitstream
[04:31:30 CEST] <memeka> any idea why this happens?
[04:35:15 CEST] <BigWoop> Sall greek to me I've been trying to google out what I can. Doesn't seem like there is much alive here :(
[04:38:35 CEST] <mrelcee> beyond my use.
[04:48:09 CEST] <deweydb> with the -t flag, is there some trick to getting the end of the file and backing off from there.
[04:48:43 CEST] <deweydb> like: ffmpeg -ss 5 -t endOfVideo-5 -i myvid.mp4 myclip.mp4
[04:48:50 CEST] <deweydb> would take 5 seconmds off both ends
[04:57:04 CEST] <ytan> if you know how long the video is, you can specify the length of the video you want to crop
[05:01:16 CEST] <ytan> you can use ffprobe for that ffprobe -i input.file -show_format | grep duration
[05:02:46 CEST] <Victorma> hi, i'm trying to load hls streamings in my 3ds but i'm always getting "connection to tcp:// ... failed: error number 26 occurred". i cant find any reference anywhere... this happens all the time with tcp i think. however, when i put "hls+" before the url it starts to try over and over again, and in the end i get an "invalid data found when processing input" and it seems that the connections are made through http
[05:03:34 CEST] <Victorma> maybe I am missing something important to receive hls streamings?
[05:23:17 CEST] <miketo> hey, i'm having trouble with some old divx/xvid/mpeg4 files.  running "ffmpeg -i test.avi -c copy test.mkv"  fails with notice "Can't write packet with unknown timestamp"
[05:24:22 CEST] <miketo> i've tried a few different files and they all report the same.  is there something strange about the way old xvid-era video streams were encoded that prevents ffmpeg from finding timestamps ^^?
[05:52:01 CEST] <ytan> Hi, I'm getting the errror "Protocol not found" when I try to run ffplay tls://127.0.0.1:8554
[05:52:06 CEST] <ytan> any idea why that is?
[07:07:22 CEST] <Victorma> hi, i'm trying to open a hls streaming using avformat_open_input, is there any problem with leaving the format and options as NULL?
[07:11:17 CEST] <memeka> after converting a packet to annexb (bsf), i get [vd] Error while decoding frame! => any idea why this might happen?
[07:33:44 CEST] <ricemuffinball>  if google/youtube/gmail/maps/etc  cost  $10/year to use it,  would you still use it?
[07:37:02 CEST] <memeka> anyone can help me with the annexb filter??
[07:47:17 CEST] <BigWoop> Anyone alive in here?
[07:52:41 CEST] <miketo> probably not
[07:54:09 CEST] <ricemuffinball> miketo cheapstake
[07:55:42 CEST] <miketo> ricemuffinball: i was responding to BigWoop
[07:56:07 CEST] <ricemuffinball> miketo oh,  put a name in front next time
[07:58:09 CEST] <miketo> is cheapstake bad steak?
[08:05:01 CEST] <Gedrean> hi, I have a problem trying to work with ffmpeg and ffserver and I can find no information online that seems to match my problem.  I am running Ubuntu Server 16.0something LTS, and using the ffmpeg as included in that distro.  I run ffserver and it seems to take its settings, have a feed with avi and a feed with asf (as enabled by the default config).  When I run ffmpeg, input with an mp4 file, output to
[08:05:04 CEST] <Gedrean> localhost:port/feed1.ffm, it loads up fine, gives me some error about past duration .79992 too large, then proceeds to start encoding at something like 110-320 fps.  The server gives no indication of a problem until a client connects and then it's just dropped frame errors spamming the console until the client disconnects - client gets no data.
[08:06:45 CEST] <Gedrean> A friend suggested I try -framerate 60 but that option seems gone or deprecated or never existed.  I tried -r 60 for a reduction in rate, but that didn't change the fps at all.
[08:13:20 CEST] <squ> what if you can change framerate in the video itself :)
[08:15:04 CEST] <Gedrean> i'm not sure what you mean, squ
[08:15:35 CEST] <squ> try to stream video with other framerate
[08:16:04 CEST] <Gedrean> I'm still not following - where am I changing a framerate?
[08:18:47 CEST] <memeka> anyone can help me with the annexb filter??
[08:19:36 CEST] <Gedrean> A bit more searching using a few other options has revealed I could try -re - which seems to have put the output of ffmpeg at the same framerate as the feed settings (15 for now) in the ffserver config - but when connecting using a media player (VLC in this case) to the ffserver I still see a "too large number of skipped frames" error
[08:20:01 CEST] <Gedrean> Same if I set the feed framerate to 30
[08:20:22 CEST] <squ> Gedrean: you can encode a video file with specific framerate
[08:20:44 CEST] <squ> if framerate is your problem, which I don't know if it is
[08:21:43 CEST] <furq> fyi ffserver is going to be removed soon so you probably shouldn't use it
[08:21:53 CEST] <furq> you also probably shouldn't use it because it sucks
[08:22:20 CEST] <Gedrean> ahhh
[08:22:36 CEST] <Gedrean> squ: I'm not encoding a file the file is already encoded fine.  It plays fine just trying to stream it
[08:22:49 CEST] <squ> oh my
[08:22:53 CEST] <Gedrean> but if ffserver is deprecating which it sounds like, what's a better streaming server you'd suggest?
[08:23:07 CEST] <squ> you can encode a file with needed framerate and stream it
[08:23:09 CEST] <Gedrean> furq that is
[08:23:10 CEST] <furq> the only one i've ever used is nginx-rtmp
[08:23:27 CEST] <Gedrean> heh, except trying to avoid flash necessity here lol
[08:23:37 CEST] <Gedrean> well, thanks anyway. :)
[08:23:39 CEST] <furq> if you're connecting with vlc then you can do that just fine with rtmp
[08:23:44 CEST] <Gedrean> oh you can?
[08:23:47 CEST] <Gedrean> hmm
[08:23:47 CEST] <furq> it also outputs to hls/dash if you care about web browsers
[08:23:51 CEST] <squ> you can stream with vlc itself
[08:24:21 CEST] <furq> although if you're not using a web browser then rtmp is much better than those two hack jobs
[08:24:25 CEST] <Gedrean> yeah streaming with vlc is a headache on its own
[08:24:32 CEST] <squ> why
[08:24:42 CEST] <squ> last time I tried it worked ok
[08:24:48 CEST] <Gedrean> Because the documentation and usage is completely unclear
[08:24:59 CEST] <squ> clear as day
[08:25:03 CEST] <Gedrean> For every command line, it mentions "options" and doesn't say what those are
[08:25:27 CEST] <Gedrean> the wiki page for command line arguments is just a copy paste of the --help and that doesn't detail what things to put in
[08:25:45 CEST] <bencoh> --fullhelp
[08:26:03 CEST] <Gedrean> yeah that's all the wiki page is, and it's not clear
[08:26:20 CEST] <squ> maybe problem is with you
[08:27:09 CEST] <squ> I'm told you couple of times to try to encode video with framerate you want (15), and you don't get it
[08:27:27 CEST] <bencoh> well vlc definitely isn't a proper broadcast solution, but it can do the job
[08:27:49 CEST] <squ> I streamed hockey match for my friend with it
[08:28:17 CEST] <squ> he called me in emergency saying there is important match going on now
[08:28:22 CEST] <Gedrean> I'm going to drop the whole encoding, framerate, etc, bullshit because you're stukc on saying I need to re-encode the video which is not at all the damn problem.
[08:28:32 CEST] <bencoh> yeah and I know ISPs than stream a few hundreds channels with it
[08:28:35 CEST] <bencoh> but still :)
[08:28:36 CEST] <squ> I figured out how to stream it quite quickly
[08:28:48 CEST] <furq> squ: did you do so with the express written consent of the national hockey league
[08:28:49 CEST] <Gedrean> squ: Did you have a GUI?
[08:29:04 CEST] <squ> me? gui?
[08:29:08 CEST] <squ> well
[08:29:34 CEST] <bencoh> he probably meant guilt or something like that
[08:30:14 CEST] <Gedrean> no I meant GUI
[08:30:18 CEST] <Gedrean> Graphical user Interface
[08:30:29 CEST] <Gedrean> because every fuckign streaming tutorial out there for vlc is based upon using a windows gui
[08:30:30 CEST] <furq> as an aside, the whole framerate thing is an artefact of ffserver not letting you stream without reencoding
[08:30:38 CEST] <furq> because, as i might have said, ffserver sucks
[08:30:55 CEST] <furq> so you probably don't actually need to do that
[08:30:55 CEST] <bencoh> Gedrean: just kidding :)
[08:30:56 CEST] <Gedrean> furq: Wow.  Okay yeah that explains that.  Thank you.  You were FAR more helpful at explaining why something is.
[08:31:40 CEST] <Gedrean> I swear all I'm trying to do is take a video and begin streaming it so multiple stations can receive it, this is giving me an ulcer I fear.  I'm off to try to study the VLC documentation more.
[08:35:09 CEST] <ytan> furq
[08:36:07 CEST] <ytan> have you had experience streaming using encryption?
[08:36:31 CEST] <memeka> :( can anyone help me with my issue? :(
[08:37:38 CEST] <ytan> what's your issue?
[08:39:16 CEST] <memeka> ytan: my decoder can't handle frames that are not annexb; if i run ffmpeg to save a file after doing a h264_mpeg4annexb filter and play that file, it works, but my code to convert the AVPacket directly is not
[08:39:41 CEST] <memeka> where the code is: https://github.com/mihailescu2m/FFmpeg/commit/d012bc50211fbf8c2cd61624d71fc1dd52f421c3
[08:40:14 CEST] <memeka> basically replace packet data with the one resulted from the av_bitstream_filter_filter function
[08:41:18 CEST] <memeka> the file converted by ffmpeg will complain that "packet was not converted to annexb" and play (cause it's already converted); files that are not annexb will say "packet was successfully converted to annexb" but won't actually play :(
[08:51:58 CEST] <nonex86> why converting from avcc to anneb and vise versa is a problem for you? its quite easy task
[08:52:11 CEST] <nonex86> *annexb
[08:59:59 CEST] <ytan_> I don't know the answer to that. Sorry.
[09:01:55 CEST] <memeka> nonex86: well i can't see what's wrong wuith my code
[09:02:54 CEST] <nonex86> well, lets see how this should work
[09:03:01 CEST] <nonex86> you open the source file
[09:03:06 CEST] <nonex86> check the extradata
[09:03:31 CEST] <nonex86> if its in avcc format, get the size of length prefix
[09:03:46 CEST] <nonex86> than for every nalu in packet
[09:03:54 CEST] <nonex86> read the size of nalu
[09:04:08 CEST] <nonex86> put a start code to decoder
[09:04:13 CEST] <nonex86> put a nalu payload
[09:04:21 CEST] <nonex86> thats all i guess
[09:04:33 CEST] <nonex86> if the extradata contains annexb payload
[09:04:43 CEST] <nonex86> you even dont need to convert anything
[09:07:18 CEST] <nonex86> if the size of length is 4 bytes you can just replace it with start code "inplace"
[09:08:48 CEST] <nonex86> and if you dont see whats wrong with code i guess it would be good idea to check it in debugger :)
[09:13:46 CEST] <memeka> well, it all looks good
[09:13:56 CEST] <memeka> but mpv complains
[09:15:51 CEST] <memeka> nonex86: what i do: if codec = H264, create av_bitstream_filter_init("h264_mp4toannexb") ; for each packet, av_bitstream_filter_filter(s->bsf, avctx, NULL, &p_filtered, &n_filtered, avpkt->data, avpkt->size, avpkt->flags & AV_PKT_FLAG_KEY);
[09:16:16 CEST] <nonex86> i never used ffmpeg bitstream conversion filters
[09:16:26 CEST] <nonex86> i just converted payload myself
[09:16:29 CEST] <memeka> then replace avpkt->data with p_filtered and avpkt->size with n_filtered
[09:16:29 CEST] <nonex86> but again
[09:16:44 CEST] <nonex86> you can check in debugger what do you get
[09:16:56 CEST] <nonex86> after using conversion filter
[09:17:12 CEST] <nonex86> what exactly happends with your packet payload
[09:17:21 CEST] <nonex86> to be sure its the thing you want
[09:17:55 CEST] <memeka> i may need debug flags when compiling, now debugging doesn't work
[09:19:34 CEST] <nonex86> well, actually you can debug any code, even not a debug build :) but if you need, yes, rebuild it with debug information, its not a problem i guess
[09:20:17 CEST] <nonex86> one more question
[09:20:25 CEST] <nonex86> avpkt->flags & AV_PKT_FLAG_KEY
[09:20:39 CEST] <nonex86> ^^^ why do you need this parameter?
[09:20:56 CEST] <nonex86> what does it mean in context of conversion process?
[09:21:00 CEST] <nonex86> just curious
[09:22:06 CEST] <nonex86> its a keyframe flag, but why do it need to be supplied to conversion?
[09:23:13 CEST] <nonex86> ah, well, never mind
[09:23:19 CEST] <nonex86> i already check documentation :)
[09:28:54 CEST] <memeka> i have no idea
[09:29:11 CEST] <memeka> but the function had flag, so i supplied it
[09:29:21 CEST] <nonex86> yeah, i already check it :)
[09:56:34 CEST] <well> hey guys, i want to convert a bluray (mkv) into a 480p video with max size of 160mb. with other sources the results were really good but now i got a source where the results looks really ugly
[09:56:50 CEST] <well> can you maybe tell me what i have to adjust or recommend anything?
[10:06:31 CEST] <gauthier> Hi! I have trouble with avformat_new_stream().
[10:07:57 CEST] <gauthier> The comment for AVStream's member `codecpar` in avformat.h states that memory is allocated for `codecpar` in `avformat_new_stream()`
[10:09:00 CEST] <gauthier> Yet after calling `av_st = avformat_new_stream(av_oc, video_codec);`, which does not return any error, I get:
[10:09:17 CEST] <gauthier> av_oc->streams[0]->codecpar == 0
[10:09:55 CEST] <gauthier> my guess is that the ffmpeg source I am looking for does not match what I got installed on my system.
[10:10:36 CEST] <gauthier> But I don't know how to match the lib version with the source version.
[10:10:48 CEST] <gauthier> ldd on my application says:
[10:10:55 CEST] <gauthier> libavformat.so.56 => /usr/local/lib/libavformat.so.56 (0x00007f629ef8f000)
[10:10:55 CEST] <gauthier>  
[10:11:28 CEST] <gauthier> but what tag or commit should I look for in the repo?
[11:33:25 CEST] <manusitara> Hi
[11:33:27 CEST] <manusitara> all
[11:33:56 CEST] <manusitara> How can I extract the count of particular macro block types(I, P and S) from the frames in the video (H.264 and MP4)?
[11:35:14 CEST] <BtbN> S?
[11:39:52 CEST] <pihpah> Any idea how to fix that error: [aac @ 0x5621547692a0] element type mismatch 3 != 0
[11:40:17 CEST] <pihpah> I am trying to convert mkv video to mp4.
[11:41:13 CEST] <manusitara> @BtbN S stands for skipped macro block
[11:41:39 CEST] <pihpah> It looks like the problem is with the audio stream.
[11:48:11 CEST] <pihpah> Okay, seems like I can just ignore that error.
[11:48:56 CEST] <pihpah> Thanks for helping, feel free to grab some cookies in the jar.
[11:49:14 CEST] <manusitara> How can I extract the count of particular macro block types(I, P and S) from the frames in the video (H.264 and MP4)?
[12:27:52 CEST] <iive> manusitara: maybe something in ffprobe would help you in that...
[12:27:56 CEST] <c_14> manusitara: ffprobe -v quiet -show_entries frame=pict_type -select_streams v -count_frames -of flat ako_framed.webm | sed 's/[^=]\+="\(.\)"/\1/' | gawk '{array[$0]+=1} END {for (i in array) { print i, array[i] } }'
[12:28:59 CEST] <manusitara> thanks for the suggestions
[12:29:00 CEST] <manusitara> Also, I would like to extract the prediction residual of each of the P and B frames (computed from its corresponding reference frame) in the video from the encoded version itself while the package is decoding it.
[12:29:01 CEST] <iive> c_14: isn't that only frame types?
[12:29:42 CEST] <c_14> Oh, hmm. I don't think ffprobe will give you anything about macro blocks
[12:30:17 CEST] <c_14> you'd probably have to hook into the h264 decoder to get those
[12:30:43 CEST] <manusitara> any suggestions on how to do that?
[12:31:47 CEST] <BigWoop> anyone see anything wrong with this by chance  ffmpeg -i movie.mp4 -i logos/watermark.png -filter_complex 'overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10' -strict -2 output.mp4? I'm at a loss
[12:34:29 CEST] <BigWoop> I'm I asking the wrong question lol?
[12:37:12 CEST] <c_14> manusitara: ah, nvmd. ffmpeg -debug mb_type -i video -f null /dev/null
[12:37:18 CEST] <c_14> Have fun parsing the output though&
[12:38:27 CEST] <c_14> For the prediction residual you'd probably have to modify the decoder/parser in ffmpeg
[12:38:45 CEST] <BigWoop> Ewwww :(
[12:39:23 CEST] <c_14> BigWoop: hmm?
[12:40:07 CEST] <BigWoop> Nada I'm just having my own fit over here probably something silly but its killing me
[12:41:20 CEST] <BigWoop> http://pastebin.com/EKGfXv4N
[12:44:20 CEST] <c_14> try adding -analyzeduration 1G -probesize 1G before -i logos/watermark.png
[12:45:12 CEST] <manusitara> thanks c_14
[12:45:36 CEST] <c_14> But it looks like the png is missing its (or has a corrupted) header, you could probably rewrite it using anything that can open it (imagemagick, gimp w/e)
[12:45:51 CEST] <c_14> Also you shouldn't need -strict -2 anymore with teh version of ffmpeg you have.
[12:49:39 CEST] <BigWoop> k just s second lemme givec it go
[12:51:53 CEST] <BigWoop> http://pastebin.com/dm2KmQUS
[12:54:01 CEST] <c_14> Yeah, try rewriting the png with something like imagemagick or gimp or whatever
[12:54:21 CEST] <BigWoop> resaving the file?
[12:56:16 CEST] <BigWoop> Its just a normal PNG file thats worked before but I haven't the foggiest idea why.. resaved it image file seems fine though
[12:59:53 CEST] Action: BigWoop scratchs his head
[13:05:07 CEST] <BigWoop> c_14: any other ideas?
[13:05:33 CEST] <c_14> What dimensions does the png have?
[13:05:36 CEST] <c_14> widthxheight
[13:06:06 CEST] <c_14> try -s widthxheight -i logos/watermark.png
[13:06:39 CEST] <c_14> and/or -video_size widthxheight -i logos/watermark.png
[13:06:45 CEST] <c_14> not sure if those are aliases
[13:07:32 CEST] <BigWoop> hmm
[13:08:08 CEST] <BigWoop> http://pastebin.com/uLLQm9UN
[13:09:21 CEST] <c_14> ffmpeg -decoders | grep png
[13:09:58 CEST] <BigWoop> http://pastebin.com/Y7Szu4fC
[13:10:07 CEST] <c_14> Well
[13:10:10 CEST] <c_14> That explains everything
[13:10:17 CEST] <BigWoop> O.o
[13:10:28 CEST] <c_14> Does your system have zlib-devel installed?
[13:11:03 CEST] <c_14> or anything that provides zlib.pc (or at least zlib.h)
[13:11:38 CEST] <BigWoop> I don't believe i put that in just a moment while I sift thru this
[13:12:10 CEST] <furq> zlib1g-dev on ubuntu
[13:13:19 CEST] <BigWoop> yteah just apt cached it I give -dev a try since I was like nope I'm not doing yum here ;)
[13:13:19 CEST] <BigWoop> ok were installed on zlib
[13:13:47 CEST] <c_14> now rebuild ffmpeg
[13:14:00 CEST] <BigWoop> an new option in configure?
[13:14:10 CEST] <furq> it's enabled by default if it finds it
[13:14:11 CEST] <c_14> It should pick it up automagically
[13:14:22 CEST] <c_14> you can check the configure output to see if it's there
[13:15:04 CEST] <BigWoop> Usually go with ./configure --enable-gpl --enable-nonfree --enable-pthreads --enable-libx264 --enable-libfaac --enable-libmp3lame --enable-encoder=png
[13:15:27 CEST] <furq> is libfaac even still supported
[13:15:52 CEST] <Kakashi> hello. is there a precise way to RMS normalize a file with ffmpeg?
[13:15:58 CEST] <c_14> furq: afaik for now it's still there
[13:16:03 CEST] <BigWoop> *shrug* I don't too much things to keep track of :P
[13:16:08 CEST] <furq> well don't use it anyway
[13:16:20 CEST] <furq> if you want to use a halfway decent non-free aac encoder then use fdk
[13:16:24 CEST] <furq> otherwise just use the builtin one
[13:18:14 CEST] <BigWoop> I'll be honest I'm a moron with anything ffmpeg by comparision. I just need it to water mark things though I've tinkered around with some rtmp stuff but just trying to get my quest completed here lol
[13:18:41 CEST] <BigWoop> Bonking my head on the wall ... durpie dee
[13:18:42 CEST] <c_14> Kakashi: https://ffmpeg.org/ffmpeg-filters.html#dynaudnorm the r value sets the rms
[13:19:02 CEST] <Kakashi> c_14: thanks!
[13:19:36 CEST] <durandal_1707> thats not typically normalization
[13:19:55 CEST] <BigWoop> c_14: http://pastebin.com/YJb20VSi
[13:20:29 CEST] <Kakashi> c_14: so, can I specify dBFS?
[13:20:42 CEST] <BigWoop> c_14: it might be working just a moment on me
[13:21:00 CEST] <Kakashi> I can do so with normalize_audio, but it doesn't support raw audio
[13:21:39 CEST] <Kakashi> and I'd prefer not to have to transcode before normalizing
[13:22:35 CEST] <Kakashi> "This filter applies a certain amount of gain to the input audio in order to bring its peak magnitude to a target level (e.g. 0 dBFS)." << aha.
[13:28:22 CEST] <BigWoop> c_14: seems like its kind of working now... Now I just have to check and see if it has any problems along the way.. Thank you very much though!
[14:06:25 CEST] <gauthier> Hi, I'm very new to libav. I'm going through (the right version of, this time) doc/examples/output.c, and I'm a bit overwhelmed by the amount of setup needed to output a video.
[14:06:49 CEST] <gauthier> Is all that is there really necessary?
[14:07:07 CEST] <gauthier> Is there any other - more minimalistic - example?
[14:08:01 CEST] <gauthier> What I mean is that I'm surprised there isn't a simpler, higher level API
[14:08:08 CEST] <_jason_> hello
[14:08:29 CEST] <_jason_> i have cross complied build for windows on linux
[14:08:44 CEST] <DHE> gauthier: you make an AVFormatContext to store the file itself, you make AVCodecContext(s) to encode the video (and audio), and so on
[14:08:53 CEST] <_jason_> using https://github.com/rdp/ffmpeg-windows-build-helpers
[14:09:05 CEST] <_jason_> it created a ffmpeg_g file
[14:09:48 CEST] <_jason_> when i'm trying to execute it : ./ffmpeg_g.exe -f dshow -r 30 -i video="screen-capture-recorder" -s 1366x786 -vcodec libx264  output.mkv
[14:10:00 CEST] <DHE> gauthier: where is this output.c you find? The official examples are list at: http://ffmpeg.org/doxygen/3.1/examples.html (a bit version dependent)
[14:10:04 CEST] <_jason_> it gives : bash: ./ffmpeg_g.exe: cannot execute binary file: Exec format error
[14:10:16 CEST] <gauthier> DHE: that sounds simple enough. Which is why I wonder why doc/examples/output.c is so complex
[14:10:37 CEST] <gauthier> DHE: they are in the source.
[14:10:56 CEST] <DHE> ... not in mine
[14:11:11 CEST] <gauthier> DHE: I currently have commit 1985c2e checked out, since I want the lib version 56
[14:11:31 CEST] <DHE> oh wow, 2014
[14:11:55 CEST] <gauthier> yeah, it's the one distributed in Debian Jessie
[14:12:07 CEST] <DHE> ah, that'd make sense...
[14:12:59 CEST] <furq> gauthier: there's a newer version in jessie-backports
[14:14:22 CEST] <gauthier> Or I could just build from git and install (I made a short attempt at that, but only the static libs seemed to install, not .so)
[14:14:37 CEST] <furq> --enable-shared
[14:14:48 CEST] <gauthier> (so I decided to go for the ones in apt-get)
[14:14:57 CEST] <furq> the one in backports is pretty recent
[14:15:08 CEST] <furq> 3.0.1 iirc
[14:15:17 CEST] <gauthier> of course, you recommend to go for the latest version :) ...
[14:15:50 CEST] <DHE> well, as a guy who follows the development out of both personal and professional interest, yes I do
[14:16:05 CEST] <gauthier> but that's I had checked out when I was looking for examples, and the examples in master were not much simpler.
[14:16:13 CEST] <furq> i'm not really sure which version you are using
[14:16:18 CEST] <furq> it's not in the normal jessie repos
[14:16:37 CEST] <DHE> they're designed to be thorough. like the audio encoding examples will deal with the fact that different codecs produce different amounts of output and consume different amounts of input
[14:16:51 CEST] <gauthier> exactly!
[14:16:52 CEST] <DHE> and of course, checking all return codes for errors
[14:17:28 CEST] <gauthier> I don't mind my implementation being very narrow minded when it comes to what to output
[14:18:10 CEST] <gauthier> I need a video stream, I need either no compression or mpeg... I don't need the flexibility.
[14:23:18 CEST] <gauthier> Ok, I'll start over. Checking out n3.1.3, I'll install with --enable-shared. That will probably break other programs, we'll see
[14:23:29 CEST] <xmedeko> Hi, I have found a great option force_original_aspect_ratio in the http://ffmpeg.org/ffmpeg-filters.html#scale filter. But rarelly it causes ffmpeg error "height not divisible by 2". is there any option, how to make this option behave like -2 in scale width or height? ths
[14:25:33 CEST] <gauthier> by the way, building gives quite the number of warnings. Array bounds, deprecated, discarding const... Is it normal?
[14:28:48 CEST] <durandal_1707> Yes
[14:34:42 CEST] <xmedeko> looking at the source code i'm afraid force_original_aspect_ratio cannot make divisible height, width. it's pitty
[14:41:41 CEST] <gauthier> any reason why --enable-shared would not imply --enable-pic in configure?
[14:43:47 CEST] <durandal_1707> because its different option
[14:45:28 CEST] <gauthier> I'm just trying to understand why one would like to build shared libs without PIC. (really, this is not meant as an aggressive question)
[16:21:14 CEST] <inkbottle> Hi, I'm trying to do sth like that: ffmpeg -i test.wav -f avi pipe:1 | cat > test.avi, but from a webcam, and to a video player
[16:22:48 CEST] <inkbottle> this: ffmpeg -v verbose -r 5 -s 320x240 -f video4linux2 -i /dev/video0 pipe:1 | ffmpeg -
[16:23:00 CEST] <inkbottle> gives Unable to find a suitable output format for 'pipe:1'
[16:23:28 CEST] <furq> you had the answer in your first command
[16:24:38 CEST] <inkbottle> furq: really?
[16:24:45 CEST] <Spring> is it possible to merge two videos on top of each other with ffmpeg?
[16:24:47 CEST] <furq> add -f avi (and probably also -c copy) before pipe:
[16:25:05 CEST] <furq> or -f nut if the webcam outputs rawvideo
[16:25:27 CEST] <furq> it'll probably be mjpeg though in which case i guess avi is right
[16:25:49 CEST] <furq> Spring: what does "merge two videos on top of each other" mean
[16:25:55 CEST] <furq> overlay?
[16:26:45 CEST] <durandal_1707> vstack? blend?
[16:26:56 CEST] <durandal_1707> lut2?
[16:27:18 CEST] <Spring> vstack looks like it's what I
[16:27:24 CEST] <Spring> 'm looking for, thanks
[16:28:09 CEST] <durandal_1707> your videos need to be of same format and width
[16:28:28 CEST] <inkbottle> I must be doing sth silly: ffmpeg -v verbose -r 5 -s 320x240 -f video4linux2 -i /dev/video0 -f avi -c copy "pipe:1" | ffmpeg - => Unable to find a suitable output format for 'pipe:'
[16:28:47 CEST] <inkbottle> same without the quotes
[16:29:11 CEST] <inkbottle> I tryed "-f nut"
[16:29:15 CEST] <furq> if the player you're piping to is actually ffmpeg then you probably need -f avi on that side as well
[16:29:25 CEST] <inkbottle> ok
[16:30:13 CEST] <durandal_1707> replace pipe with -?
[16:33:54 CEST] <inkbottle> At least that works: ffmpeg -f v4l2 -framerate 25 -video_size 640x480 -i /dev/video0 output.mkv => and then playing output.mkv
[16:34:07 CEST] <inkbottle> as for the piping nothing worked yet
[16:36:45 CEST] <Spring> durandal_1707, by format you mean codec? Or everything, including bitrate?
[16:37:09 CEST] <Spring> also used one too many 'looks' in my previous post :p
[16:37:31 CEST] <durandal_1707> Spring: pixel format
[16:39:23 CEST] <Spring> that shouldn't be a problem then
[16:49:54 CEST] <inkbottle> I'll probably find the solution some other time
[18:09:26 CEST] <Lucretia> hi, I'm getting corrupt screengrabs, https://youtu.be/pFqhIGYLbDM can someone point me in the right direction of which component could be failing?
[18:16:36 CEST] <iive> Lucretia: unfortunately, this looks like xorg driver issue
[18:17:16 CEST] <Lucretia> iive: I was wondering whether it would be kernel, xorg, libdrm, mesa or ffmpeg
[18:17:58 CEST] <Lucretia> so many components
[18:27:15 CEST] <Lucretia> iive: the /usr/lib/xorg/modules/drivers/amdgpu_drv.so not the /usr/lib/dri/radeonsi_dri.so module?
[18:49:38 CEST] <Mista_D> Trying to set up libx264 to use static IDR with "-g" and use non-IDR intra-refresh frames for scene change. Any advice?
[18:53:01 CEST] <bencoh> is that even possible without patching it?
[20:28:03 CEST] <transhuman_> anyone know how to launch ffmpegYAG or ffmpeg-hi?
[22:32:28 CEST] <pxl__> hello hello
[22:46:02 CEST] <Kayy> odd question. Ive got a still image that im turning into a video but I cant manage to get it over 50kbs to save my life. I'm aiming for at least 200kbs to stress test a system with.
[22:46:13 CEST] <Kayy> http://pastebin.com/ccCHtATH
[23:49:52 CEST] <Denise0311> Hi, is there a way to access webcam raw video stream without modifying its firmware?
[23:54:25 CEST] <kyleogrg> does libvpx-vp9 have any tune options like libx264?
[23:54:41 CEST] <Jnorthrup> is there a filter to un-aperture quicktime codings?
[00:00:00 CEST] --- Wed Aug 31 2016