July 2016 - Ffmpeg-devel-irc - lists.ffmpeg.org

ffmpeg-devel.log.20160730
by burek 31 Jul '16

31 Jul '16

[00:29:12 CEST] <cone-673> ffmpeg 03Michael Niedermayer 07master:0219dc6c0725: avformat/rmdec: Clear extradata when extradata_size is cleared [00:29:12 CEST] <cone-673> ffmpeg 03Lucas Cooper 07master:bc9ce5f6bec0: avfilter: Add new format for PSNR stats log [02:12:42 CEST] <cone-673> ffmpeg 03Chris Cunningham 07master:52ec4cc09b5b: libavformat/matroskadec: Add test for seeking with codec delay. [04:01:48 CEST] <cone-673> ffmpeg 03Timothy Gu 07master:22df70e9edaa: doxygen: Add a note on libavutil's modular nature [04:01:49 CEST] <cone-673> ffmpeg 03Timothy Gu 07master:f198b8af72eb: doxygen: Fix PREDEFINED syntax [04:01:50 CEST] <cone-673> ffmpeg 03Timothy Gu 07master:6253e5676fb1: doxygen: Update Doxyfile for Doxygen 1.8.8 [04:01:51 CEST] <cone-673> ffmpeg 03Timothy Gu 07master:7204a6298be7: doxygen: Set tab size to 4 [07:37:55 CEST] <AnonBaiter> I have an issue with the .MPC file [07:38:09 CEST] <AnonBaiter> when I encode a .mpc file to any codec, the resulting file flickers for no reason [07:38:40 CEST] <AnonBaiter> I'm uploading the sample at... 20+ minutes or less [07:50:24 CEST] <AnonBaiter> https://www.sendspace.com/file/p1hcft [07:50:28 CEST] <AnonBaiter> here's the sample [08:57:50 CEST] <omerjerk> is there any way to specifically specify as to which encoder to use while converting wav to mp4 ? [08:58:11 CEST] <omerjerk> I merged an als encoder into the codebase but how to specify which command to use exactly? [09:20:12 CEST] <durandal_1707> -c:a als [09:21:02 CEST] <durandal_1707> omerjerk: see above! [09:22:06 CEST] <omerjerk> oh thanks a lot. It worked. :) [11:08:34 CEST] <omerjerk> Is this assignment of uint8_t* to int32_t* okay? - https://github.com/omerjerk/FFmpeg/blob/alsenc/libavcodec/alsenc.c#L2904 [11:08:48 CEST] <omerjerk> The compiler is giving a warning. [11:54:58 CEST] <kierank> omerjerk: needs a cast [12:01:42 CEST] <omerjerk> oh okay. thanks. [12:24:47 CEST] <kierank> durandal_1707: going to vdd? [12:54:50 CEST] <durandal_1707> kierank: nobody pays for significant contributions I did [13:04:03 CEST] <kierank> i would pay you to fix libavfilter api [13:04:21 CEST] <kierank> but i would pay for code, not time wasting on ml against neckbeards [14:34:28 CEST] <ubitux> i'll be away for about a week or two [14:34:54 CEST] <ubitux> i encourage anyone to look at the merge [14:35:27 CEST] <ubitux> maybe i'll do some more before going away, but starting tomorrow, nothing will happen for a while [14:41:35 CEST] <iive> it's usually bad idea to commit something right before you go AFK [14:55:07 CEST] <ubitux> it won't be my commits, just the merge [15:01:18 CEST] <durandal_1707> ubitux: nlmeans? [15:02:28 CEST] <ubitux> durandal_1707: i'm busy with the merges most of the time, and i unfortunately have other shit such as subtitles api to deal with [15:02:39 CEST] <ubitux> and now i'm away for a little while [15:07:08 CEST] <durandal_1707> but you said you will finish it... [15:18:34 CEST] <cone-296> ffmpeg 03Paul B Mahol 07master:61d302dd5215: doc/filters: mention HDMI output format for stereo3d filter [15:32:47 CEST] <omerjerk> hi [15:33:28 CEST] <omerjerk> I'm continuously getting "Internal error, put_bits buffer too small" [15:34:03 CEST] <omerjerk> It's coming from the code around here I suppose - https://github.com/omerjerk/FFmpeg/blob/alsenc/libavcodec/alsenc.c#L2806 [15:35:06 CEST] <omerjerk> I'm pretty much sure that the size is way more than enough than what is being written later [15:35:21 CEST] <omerjerk> any idea whatsoever? [15:35:34 CEST] <omerjerk> I'm not getting the problem. [15:42:13 CEST] <Compn> durandal_1707 : "+HDMI frame pack" is there somewhere that explains kinda what this does? does it auto scale or something ? [15:42:28 CEST] <Compn> sorry if its dumb question. i just confused by reading manual... [16:24:17 CEST] <durandal_1707> omerjerk: how you calculate packet size? [16:25:15 CEST] <durandal_1707> Compn: I will tell you when you add my codecs to mplayer [16:26:09 CEST] <durandal_1707> It's stupid mode for crappy 3d projectors [16:30:25 CEST] <durandal_1707> omerjerk: use number of samples you gonna use in frame [16:31:07 CEST] <omerjerk> okay!! Thanks. [16:32:15 CEST] <durandal_1707> omerjerk: also make sure you give init_put_bits correct size [16:32:30 CEST] <omerjerk> okay [16:36:48 CEST] <AnonBaiter> can I repost my problem here? [16:37:39 CEST] <Chloe> AnonBaiter: is it related to using ffmpeg or using the libraries? if so, then you should ask in #ffmpeg [16:42:26 CEST] <AnonBaiter> done [16:42:36 CEST] <AnonBaiter> it is related to both, I guess [17:15:20 CEST] <Compn> durandal_1707 : i was more hoping you would amend the manual entry to tell everyone, not just me. [17:15:26 CEST] <Compn> i dont have hdmi here so i dont need it :P [19:00:26 CEST] <omerjerk> Hi [19:00:48 CEST] <omerjerk> Can anyone please have a look at me write_frame logic ? - https://github.com/omerjerk/FFmpeg/blob/alsenc/libavcodec/alsenc.c#L2954 [19:01:17 CEST] <omerjerk> I don't get any errors when I run the encoding but at the same time, nothing gets written the output file either. :/ [19:01:41 CEST] <omerjerk> durandal_170: [19:23:44 CEST] <kierank> do you have an als muxer? [20:41:11 CEST] <durandal_170> omerjerk: paste full command you use [20:41:25 CEST] <omerjerk> ffmpeg -i int.wav -c:a als out.mp4 [20:54:02 CEST] <Timothy_Gu> Doxygen for av_gcd says "Returns: gcd of a and b up to sign" [20:54:07 CEST] <Timothy_Gu> what does "up to sign" mean [20:56:07 CEST] <durandal_170> omerjerk: hmm, do your encoder returns something? [21:06:27 CEST] <durandal_170> omerjerk: check avctx gop_size variable [21:06:58 CEST] <durandal_170> Perhaps your code never set avpkt size [21:12:46 CEST] <durandal_170> michaelni: do you have better fix than mine for alac encoder? [21:31:41 CEST] <KGB> [13FFV1] 15retokromer opened pull request #23: dead link (06master...06patch-2) 02https://git.io/vKjHo [22:07:45 CEST] <michaelni> durandal_170, it would be ideal to use the exact size and document how the value is reached but the patch is better than it is now [22:08:12 CEST] <michaelni> "exact worst case" [22:13:16 CEST] <Chloe> mateo`; if you wanted to have a look at porting ffmpeg*.c feel free, I cant seem to get it right [22:13:28 CEST] <Chloe> (to codecpar I mean) [22:16:14 CEST] <cone-175> ffmpeg 03Alexey Tourbin 07master:906ee4114117: avfilter/af_stereowiden: fix read/write past the end of buffer [00:00:00 CEST] --- Sun Jul 31 2016

1 0

ffmpeg.log.20160730
by burek 31 Jul '16

31 Jul '16

[00:36:06 CEST] <wallbroken> hi folks [00:36:26 CEST] <wallbroken> do you think is better to encode/decode with the last version of ffmpeg? [00:36:35 CEST] <wallbroken> if yes, why? [00:37:23 CEST] <klaxa> yes, possibly contains more performance tweaks and supports more codecs/specifics [00:37:50 CEST] <c_14> Also bugfixes [00:38:09 CEST] <wallbroken> just because windows xp supports ffmpeg since 2015 [00:38:17 CEST] <wallbroken> so, i need to move on a newer windows to doit [00:38:36 CEST] <wallbroken> i can do it, but if there are no differences, i can stay on 2015 version [00:39:34 CEST] <furq> ffmpeg should still run fine on xp, it's just the zeranoe builds which don't [00:40:23 CEST] <furq> https://ffmpeg.zeranoe.com/forum/viewtopic.php?t=2937 [00:40:29 CEST] <furq> there are some recent xp-compatible builds listed in there [00:40:35 CEST] <furq> or you could build it yourself [00:41:29 CEST] <wallbroken> but zaranoie is the official on windows [00:41:42 CEST] <wallbroken> and i don't want to build it myself [00:41:43 CEST] <furq> they're no more official than any other builds [00:42:37 CEST] <furq> and if you're concerned about security then you shouldn't really be using xp [00:42:54 CEST] <wallbroken> as i said, i can do it on win 10 [00:53:59 CEST] <cvax> how do i convert any audio file to mp3 in any suggested bitrate? [00:54:39 CEST] <furq> -i src -c:a libmp3lame -q:a 0 desc.mp3 [00:57:18 CEST] <cvax> thank you, sir [02:37:25 CEST] <wallbroken> 32bit version and 64 will produce the same output? [02:51:20 CEST] <codespells> I need to capture some noisy VHS video. Is there a filter in ffmpeg that can be used? [02:52:23 CEST] <furq> codespells: https://ffmpeg.org/ffmpeg-filters.html#hqdn3d-1 [02:52:24 CEST] <codespells> Was looking at https://trac.ffmpeg.org/wiki/FilteringGuide but there is not really much info there [02:52:28 CEST] <furq> other denoisers are available [02:52:43 CEST] <codespells> other? [02:53:31 CEST] <codespells> you mean like NLMeans etc [02:53:32 CEST] <furq> atadenoise, dctdnoiz, owdenoise, removegrain [02:53:39 CEST] <furq> nlmeans isn't in ffmpeg yet [02:53:51 CEST] <furq> you'd have to use avisynth/vapoursynth or something [02:54:22 CEST] <codespells> what would be normally used for an old VSH tape? [02:54:32 CEST] <codespells> VHS [02:55:18 CEST] <furq> i can't say i've ever tried [02:55:23 CEST] <furq> hqdn3d seems to be quite widely used [02:57:21 CEST] <codespells> I'll try it out. Does not seem to be any examples on that page. [02:58:05 CEST] <furq> -vf hqdn3d=3:2:2:3 [02:58:18 CEST] <furq> that's what the medium preset in handbrake uses [02:58:55 CEST] <furq> or just -vf hqdn3d [03:00:23 CEST] <codespells> Thank you. [03:07:35 CEST] <ljr1981> is there anything for doing audio-speech-to-text-to-SRT-CC? [03:07:43 CEST] <ljr1981> (automatic) [03:45:48 CEST] <kepstin> ljr1981: only thing I know of is YouTube, and it's, well, pretty bad. I can't think anything better exists. [03:46:07 CEST] <kepstin> You're probably better off just paying a human to do it [03:47:16 CEST] <codespells> uhh, was asked to write a denoising algorithm using wavelet transform using non-linear estimators using Bayesian theory for video. Anybody know of any GPL code out there that works along these lines? [03:48:00 CEST] <codespells> Don't really care how hard it taxes the hardware... [03:53:26 CEST] <codespells> Blah, its 4am here in Sweden. Gonna get a few hours of sleep. [04:41:17 CEST] <jpleau> What do I need to include to have an implementation of AVDictionary? libavutil/dict.h has 'typedef struct AVDictionary AVDictionary', but I'm getting an incomplete type error because that's not a definition.. [04:45:52 CEST] <kepstin> jpleau: you don't use it except as a pointer passed to various functions. It's an opaque object. [04:46:42 CEST] <kepstin> that header gives some examples of using it... [04:48:37 CEST] <jpleau> I see. I was looking at the doc for AVFormatContext, is contains a AVDictionary* metadata. I thought it was just usable as a normal "" array "", guess not [04:48:44 CEST] <jpleau> I'll look at that header, thanks kepstin [04:49:13 CEST] <jpleau> (defining that struct in my own file works, but this is completely the wrong way of doing it :( )_ [09:34:56 CEST] <fling> How to specify which audio stream is for which video angle? [10:49:19 CEST] <ZSU-23-4> which one device I should specify to capture sound? [14:30:42 CEST] <Penthus> Hey guys I am having an issue when I use 2-pass with libx264. It will go fine up to a random point and then just get stuck and not progress. It wont give an error or anything. Changing from libx264 to mpeg4 for the same file will work fine. Am I missing something? [16:42:24 CEST] <AnonBaiter> I have an issue with the .MPC file [16:42:24 CEST] <AnonBaiter> when I encode a .mpc file to any codec, the resulting file flickers for no reason [16:42:24 CEST] <AnonBaiter> I'm uploading the sample at... 20+ minutes or less [16:42:24 CEST] <AnonBaiter> https://www.sendspace.com/file/p1hcft [16:42:24 CEST] <AnonBaiter> here's the sample [16:43:23 CEST] <durandal_1707> What's mpc? [16:43:40 CEST] <AnonBaiter> it's an EA video format [16:43:44 CEST] <AnonBaiter> check it out for yourself [16:44:18 CEST] <durandal_1707> its pal8? [16:44:30 CEST] <AnonBaiter> looks like NTSC to me [16:48:27 CEST] <durandal_1707> and decoded one doesnt flicker? [16:51:46 CEST] <AnonBaiter> the decoded one flickered to me [16:52:35 CEST] <AnonBaiter> it goes like this: [16:52:36 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] ignoring SEQ_START_CODE after 101 [16:52:36 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] ignoring seq ext after 101 [16:52:36 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] Invalid mb type in B-frame at 33 0 [16:52:36 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] Warning MVs not available [16:52:36 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] concealing 400 DC, 400 AC, 400 MV errors in B frame [16:52:37 CEST] <AnonBaiter> [mpeg2video @ 056ddf60] ignoring SEQ_START_CODE after 101 [18:01:30 CEST] <wrs> hey, im wondering if its possible to perform multiple fast-seeks using the same input file [18:03:06 CEST] <wrs> eg with "ffmpeg -ss 10 -i input.mp4 -t 5 -ss 20 -i input.mp4 -t 5 output.mp4" I would 'expect' output.mp4 to have seconds 10-15 of input, then 20-25 of input. for a total of 10 seconds [18:32:47 CEST] <cider> Does FFmpeg hold any kind of non CMD/shell interface, regardless of being a official/unofficial plugin? I am looking for a easy to read encoding/formats list, have had too much trouble using NLE's like Adobe Premiere/AfterEffects. I do most work in TF Nuke but that is mostly .avi ? [18:34:36 CEST] <flux> cider, maybe there is, but perhaps ffprobe is useful to you, if you haven't noticed that one comes with ffmpeg [18:40:41 CEST] <cider> Yea i just read its description, i am not in dire need right now, i will take a look. Will take some time. I am trying to plot my pipeline because running 3 program subscriptions will end in a yearly 3200 USD at best. [19:05:54 CEST] <wrs> cider, if you're looking to extract information from media files: ffprobe -print_format json -show_format -show_streams input_file.mp4 [19:08:52 CEST] <cider> wrs, I was looking for a easy to use up-to-date encoding list, which would classify as free too. I do not think this is it but i am trying to take a look. [19:15:13 CEST] <cider> flex, wrs Thanks guys and sorry for the slow reply, but this is not what i am looking for. Too hard for me. I think i recall someone i once knew, using this program to skip Adobe Premiere/AE. [21:06:36 CEST] <medya> I keep getting WARNING: library configuration mismatch on ubuntu for ffmpeg, and it keeps bugging me many codecs are not installed, is there a way to install ffmpeg correctly on ubuntu (16.04) with all the codecs ? the offical repos seem to suck [21:26:11 CEST] <medya> I get this error when trying to ./configure from source " ERROR: gnutls not found using pkg-config" [21:26:57 CEST] <medya> I am trying to build it on ubuntu 16.04 [21:31:24 CEST] <jkqxz> Install libgnutls-dev (or whatever the package is actually named). [21:31:53 CEST] <furq> libgnutls28-dev [21:33:50 CEST] <furq> medya: if you're getting "library configuration mismatch" then compile with --enable-static or remove ffmpeg/libav* from apt [21:34:03 CEST] <furq> preferably the former [21:44:25 CEST] <medya> ok thanks [21:58:50 CEST] <Polochon_street> Hi ! I'm trying to update my code to the new version of ffmpeg; I changed my codec code to this http://sprunge.us/OjIP , but the codec_context is all wrong, without any error... Do you have any clue ? [21:59:08 CEST] <medya> my make failed on "common.mak:60: recipe for target 'libavcodec/libopenh264enc.o' failed" [22:03:41 CEST] <wallbroken> 32bit version and 64 will produce the same output? [22:07:38 CEST] <medya> ok I gave up on compiling it myself, I am thinking to use a docker with ffmpeg on it. [22:07:49 CEST] <medya> can I run ffmpeg on a docker on my host computer files? [22:07:52 CEST] <medya> if yes how [22:07:55 CEST] <Polochon_street> it's weird - I allocate and set the context properly, but every field of it is 0 [22:09:45 CEST] <furq> medya: what's wrong with the ffmpeg in apt [22:09:54 CEST] <furq> it has all the codecs i care about other than fdk, and there are no binary distributions with fdk [22:10:06 CEST] <furq> unless ubuntu have fucked it up, which wouldn't surprise me [22:10:10 CEST] <jkqxz> wallbroken: In general, yes. There may be minor differences due to nondeterminism of some things, but they shouldn't affect the quality of the output. (The speed of it certainly will change, though.) [22:14:56 CEST] <wallbroken> jkqxz, "same output" means same hash on both the files [22:15:37 CEST] <wallbroken> if i execute the same encoding two times in 32 bit mode, i get two identical files [22:20:46 CEST] <furq> it depends on the codecs and settings used [22:20:47 CEST] <jkqxz> It should be identical with most codecs, but I wouldn't be confident of saying that it is in all cases. You should just try it with the codecs you are interested in. [22:21:31 CEST] <furq> it's not really an ffmpeg question unless you're using internal encoders [22:27:45 CEST] <medya> furq: the one in ap doesnt work for 264, I did everything everyone said always breaks [22:27:49 CEST] <medya> and it is a very old ffmpeg [22:27:51 CEST] <medya> it is like 2 [22:27:52 CEST] <medya> nt 3 [22:31:14 CEST] <furq> 2.8.6 isn't that old [22:31:36 CEST] <furq> less than six months [23:26:05 CEST] <samsonr> hello, I'm trying to push a live stream to a http server using ffmpeg. All works fine using the following command: ffmpeg -i /dev/video0 -f mpeg1video http://localhost/pub?id=1 [23:26:37 CEST] <samsonr> however I was curious if it's possible to configure ffmpeg to make a new http request for each data chunk [23:26:54 CEST] <samsonr> atm it just keeps a long-lived connection and pushes the data [23:27:31 CEST] <samsonr> I thought the flag '-multiple_requests 0' might help, but it seems not [23:27:32 CEST] <samsonr> any ideaS? [23:28:30 CEST] <Mavrik> That's a very strange question and even stranger choice of video codec. [23:28:37 CEST] <Mavrik> Why do you need chunks? [23:31:09 CEST] <samsonr> https://github.com/slact/nchan/issues/208 [23:31:31 CEST] <samsonr> I'm doing mpeg live streaming via websockets [23:31:39 CEST] <samsonr> it actually works well for my needs :D [23:31:56 CEST] <samsonr> for more info: http://phoboslab.org/log/2013/09/html5-live-video-streaming-via-websockets [23:32:15 CEST] <samsonr> but the point is the server-side is written in node, also it's not really scalable [23:32:41 CEST] <Mavrik> Just... no. :/ [23:32:51 CEST] <Mavrik> It'll also look terrible due to MPEG-1. [23:33:06 CEST] <Mavrik> Use nginx rtmp plugin, stream to it via RTMP and let it distribute HLS/DASH [23:33:24 CEST] <Mavrik> All pure HTTP supported by browsers. [23:33:38 CEST] <samsonr> I can't use hls...I need the user to interact with the page [23:33:54 CEST] <samsonr> I already run Wowza in parallel so that's not the issue, there s a HLS stream handy anyway [23:34:51 CEST] <samsonr> anyway, that s beyond the point of my question :) So I assume it's not possible to push data in different reqs? [23:35:09 CEST] <samsonr> I d think the only overhead would be sending the same headers over and over [23:35:15 CEST] <samsonr> the handshake would be reused.. [23:36:10 CEST] <Mavrik> No, it's not really possible. [23:36:22 CEST] <Mavrik> What would ffmpeg split the data on anyway? [23:36:37 CEST] <Mavrik> It wasn't built for such use-case (unless the muxer supports it specifically) [23:36:52 CEST] <samsonr> what's it splitting it now on? [23:37:01 CEST] <Mavrik> It's not. [23:37:07 CEST] <samsonr> ffmpeg -i /dev/video0 -f mpeg1video http://localhost/pub?id=1 [23:37:10 CEST] <Mavrik> It's a transcoder, it outputs data in a stream. [23:37:24 CEST] <samsonr> when I do this, I get like 10 'data' events per sec [23:37:26 CEST] <samsonr> from node [23:37:33 CEST] <samsonr> it's a stream [23:37:41 CEST] <samsonr> right [23:43:02 CEST] <samsonr> then can you please explain what the "multiple_requests" flag does? I think its name doesn t make it clear [23:45:21 CEST] <Mavrik> Sets HTTP header to `Connection: Keep-Alive` when connecting to server. [23:46:09 CEST] <samsonr> thanks [23:46:12 CEST] <Mavrik> (or Close, depending on the value) [23:46:16 CEST] <Mavrik> Nothing else according to code grep. [23:46:31 CEST] <Mavrik> Meant as a client option I guess. [00:00:00 CEST] --- Sun Jul 31 2016

1 0

ffmpeg-devel.log.20160729
by burek 30 Jul '16

30 Jul '16

[00:02:27 CEST] <michaelni> Chloe, look at the change that was done to avconv*c [00:04:04 CEST] <Chloe> yeah, I was, but I think I've done too much now. I'm gonna have redo it from scratch :/ [00:05:26 CEST] <nevcairiel> one could do it in partial steps, first make sure encoding and decoding doesnt actually use AVStream->codec, but its own dedicated and cleanly created codecontext, and as second step then replace all remaining use of AVStream->codec with codecpar [00:09:13 CEST] <michaelni> i think spliting the change up would be a very good idea, also for bisecting any breakages [00:10:24 CEST] <nevcairiel> iirc libav also did these two steps [00:10:30 CEST] <nevcairiel> at least [00:36:50 CEST] <cone-553> ffmpeg 03Petru Rares Sincraian 07master:08c1b6bb72e4: fate: add test for silenceremove filter [00:36:50 CEST] <cone-553> ffmpeg 03Petru Rares Sincraian 07master:286368099a3a: fate: add test for stereotools filter [09:02:39 CEST] <cone-990> ffmpeg 03Stefano Sabatini 07master:156b5bb9bf75: lavf/mpegtsenc: add special case for handling timed ID3 packets [10:08:10 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:17e7c03e12d1: h264: only allow ending a field/starting a new one before finish_setup() [10:08:11 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:a63c73122e90: Merge commit '17e7c03e12d1e4490921e7bffaeaa6b46a7ada4e' [10:16:54 CEST] <cone-990> ffmpeg 03Josh de Kock 07master:238733f7417c: hapdec: remove unused memory.h include [10:18:50 CEST] <ubitux> Chloe: it shouldn't; look at the commit changing stuff in avconv.c [10:24:07 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:bcd91f1644b4: h264: move a per-field block from decode_slice_header() to field_start() [10:24:08 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:13872d2aa1f4: Merge commit 'bcd91f1644b46dd142c5355c8b742b27d9028903' [10:25:28 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:19446dc5fff6: h264: drop unused NAL_FF_IGNORE [10:25:29 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:cc0852bee360: Merge commit '19446dc5fff6a2f5ccd1a73db66e266255a5d768' [10:52:49 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:f651c6a259d4: h264: factor out setting frame properties / side data [10:52:51 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:7dda97fb86ff: Merge commit 'f651c6a259d4bc78f25db11d25df9256d5110bd3' [11:00:11 CEST] <ubitux> libavcodec/mips/h264chroma_mips.h:#include "libavcodec/h264.h" [11:00:13 CEST] <ubitux> libavcodec/mips/h264dsp_mips.h:#include "libavcodec/h264.h" [11:00:15 CEST] <ubitux> do we really need this?... [11:07:09 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:9df889a5f116: h264: rename h264.[ch] to h264dec.[ch] [11:07:10 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:15b26e88cb7c: Merge commit '9df889a5f116c1ee78c2f239e0ba599c492431aa' [11:20:15 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:251cbb44003c: h264: create a new header for common h264 definitions [11:20:16 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:ecf65c30cf75: Merge commit '251cbb44003caf179fb17afbb8a6c56643c2a646' [11:27:39 CEST] <cone-990> ffmpeg 03Anton Khirnov 07master:4e2f6212483a: svq3: stop using H264Picture [11:27:40 CEST] <cone-990> ffmpeg 03Clément BSsch 07master:a170365ae9b5: Merge commit '4e2f6212483ae1b2a4043cddf0a1cb001b476abc' [13:10:41 CEST] <michaelni> ubitux, both #include "libavcodec/h264dec.h" in the mips code are needed, it fails build otherwise [13:11:02 CEST] <michaelni> src/libavcodec/mips/h264dsp_mips.h:97: error: expected declaration specifiers or ... before ptrdiff_t [13:11:56 CEST] <ubitux> you don't need h264dec.h to get ptrdiff_t [13:12:01 CEST] <ubitux> unless i'm missing something [13:12:17 CEST] <michaelni> no but some include is needed [13:13:23 CEST] <ubitux> so stddef.h maybe? [13:14:41 CEST] <rcombs> should be, yeah [13:14:51 CEST] <michaelni> with stddef: src/libavcodec/mips/h264dsp_init_mips.c:138: error: expected =, ,, ;, asm or __attribute__ before void [13:17:14 CEST] <rcombs> needs libavutil/attributes.h [13:17:33 CEST] <ubitux> michaelni: this .c needs libavcodec/h264dsp.h [13:18:08 CEST] <ubitux> or if you prefer, h264dec.h h264dsp.h in the header [13:18:53 CEST] <ubitux> but imo you should have stddef in the header, and h264dsp.h in the .c, because somehow the mips prototypes do not need access to the H264DSPContext [13:19:45 CEST] <rcombs> this reminds me of a general question: is it safe to use intptr_t/ptrdiff_t as argument types for x86 ASM functions and assume that the top half of the 64-bit register will be cleared on ABIs that use 64-bit registers and 32-bit pointers? [13:20:12 CEST] <rcombs> (related: fuck said ABIs) [13:21:20 CEST] <michaelni> still fails to build 4 files with libavcodec/h264dsp.h and stddef.h in the headers [13:22:06 CEST] <michaelni> libavutil/attributes.h also doesnt help [13:22:16 CEST] <michaelni> src/libavcodec/mips/h264chroma_init_mips.c:24:5: warning: "HAVE_MSA" is not defined [13:22:30 CEST] <michaelni> config.h or so missing i guess somewhere [13:22:45 CEST] <nevcairiel> why are the mips optimizations always such a nightmare [13:23:13 CEST] <rcombs> because they come from people who don't care [13:23:32 CEST] <nevcairiel> maybe they should be purged with ENOMAINTAINER [13:23:58 CEST] <michaelni> with config.h it fails here: src/libavcodec/mips/h264chroma_init_mips.c:55: error: expected ) before * token [13:24:27 CEST] <michaelni> ill leave the include as is this is not my crussade [13:24:41 CEST] <michaelni> but happy to test if someone has a patch [13:31:10 CEST] <michaelni> rcombs, nevcairiel, if you think minimizing the set of recursive includes is imortant, contacting the authors/maintainers for the mips code is probably worth a try [13:31:34 CEST] <michaelni> they generally have fixed all issues i pointed out to them though not always quick [13:32:04 CEST] <michaelni> i could look at it too but not today [13:34:01 CEST] <rcombs> I usually avoid this sort of problem by not building anything for MIPS [13:34:35 CEST] <rcombs> since I've never done Nintendo 64 homebrew [13:34:55 CEST] <michaelni> but what problem is there to begin with ? h264 optimizations including the main h264 decoder header doesnt seem all that bad to me [13:35:00 CEST] <michaelni> ? [13:35:18 CEST] <rcombs> no clue [13:35:33 CEST] <rcombs> I'm just here to make fun of obscure architectures with giant convoluted optimizations [13:43:43 CEST] <JEEB> I think the only MIPS thing I built for was the Sony PSP [13:57:48 CEST] <ubitux> michaelni: they don't need the decoder; they only need h264 dsp context [16:14:58 CEST] <durandal_1707> is videolan still giving money away? [16:27:49 CEST] <mp1> Hi [16:27:56 CEST] <mp1> I have a question [16:28:18 CEST] <mp1> I want use ffmpeg in my script php [16:28:34 CEST] <mp1> for record video from my acquistion card [16:28:53 CEST] <mp1> No problem for starting a record with php [16:29:32 CEST] <mp1> but how stop record of ffmpeg with php whitout kill a process ffmpeg [16:34:26 CEST] <mp1> Would you help me please :) [16:39:45 CEST] <andrey_turkin> mp1: this question belongs to #ffmpeg channel [16:40:28 CEST] <mp1> ok :) [16:40:35 CEST] <mp1> thank you andrey_turkin [17:14:40 CEST] <omerjerk> hi everyone [17:15:00 CEST] <omerjerk> what could be an alternative of url_is_streamed - https://github.com/omerjerk/FFmpeg/blob/alsenc/libavformat/alsenc.c#L60 [17:15:01 CEST] <omerjerk> ? [17:15:06 CEST] <omerjerk> I couldn't figure it out. [17:17:39 CEST] <Chloe> omerjerk: "2cae980 / 8978fed introduce an AVIOContext.seekable field as a replacement for 1923: AVIOContext.is_streamed and url_is_streamed()" [17:17:59 CEST] <Chloe> from doc/APIchanges [17:20:25 CEST] <omerjerk> oh okay [17:20:33 CEST] <omerjerk> let me look into it. [18:05:24 CEST] <omerjerk_> what could be the solution to this error - 13 warnings generated. CC ffmpeg_videotoolbox.o LD ffmpeg_g Undefined symbols for architecture x86_64: "_ff_als_demuxer", referenced from: _av_register_all in libavformat.a(allformats.o) ld: symbol(s) not found for architecture x86_64 clang: error: linker command failed with exit code 1 (use -v to see invocation) make: *** [ffmpeg_g] Error 1 [18:05:42 CEST] <omerjerk_> "_ff_als_demuxer", referenced from: _av_register_all in libavformat.a(allformats.o) [18:05:54 CEST] <omerjerk_> Undefined symbols for architecture x86_64: [18:06:12 CEST] <omerjerk_> I've added the entry for the demuxer in allformats.c [18:08:54 CEST] <durandal_1707> but actual code? [20:14:56 CEST] <Timothy_Gu> + [20:16:20 CEST] <Timothy_Gu> oops [00:00:00 CEST] --- Sat Jul 30 2016

1 0

ffmpeg.log.20160729
by burek 30 Jul '16

30 Jul '16

[00:07:00 CEST] <insivumeh> Why is unsupported MEDIASUBTYPE_RGB24 e436eb7d-524f-11ce-9f53-0020af0ba770 in my video4linux device? [00:36:31 CEST] <insivumeh> insivumeh: OK, answer is here https://www.mail-archive.com/linux-media@vger.kernel.org/msg82920.html [02:09:43 CEST] <madprops> any tips on how to lower the file size of a mp4 or mkv as much as i can without affecting quality too much? [02:09:55 CEST] <madprops> 737mb is a bit much [02:10:32 CEST] <madprops> im willing to sacrifice some video quality [02:11:07 CEST] <furq> use a higher crf or a slower preset [02:40:36 CEST] <sponsx> Hi all, I have a video on my android phone that I was recording when the phone died and I'm trying to recover it. [02:41:25 CEST] <sponsx> I'm trying to use https://github.com/ponchio/untrunc but I can't get it to compile on my mac.. it can't find endian.h even though its there in /usr/include/machine [02:41:48 CEST] <sponsx> Can ffmpeg itself recover the video? Or any other ideas to get this working ? [04:57:09 CEST] <yenoh> hi all, I'm trying to build ffmpeg from source (in linux host) to get nvenc-h264 and nvenc-hvec encoder, as guided in trac: https://trac.ffmpeg.org/wiki/HWAccelIntro#NVENC [04:57:24 CEST] <yenoh> after building, I got: http://pastebin.com/C7bFwaMC [04:58:29 CEST] <yenoh> however, whenever I try to encode a sample video file with nvenc-h264, I have "unknown decoder" error: http://pastebin.com/tt9adHms [04:59:03 CEST] <furq> it's h264_nvenc [04:59:34 CEST] <yenoh> furq: thanks for your comment, but I still have Unknown encoder 'h264_nvenc' [05:00:02 CEST] <furq> er [05:00:06 CEST] <furq> nvenc_h264 rather [05:00:08 CEST] <yenoh> I also tried Unknown encoder 'nvenc_h264' [05:00:16 CEST] <furq> ffmpeg -codecs | grep nvenc [05:01:08 CEST] <furq> the "library configuration mismatch" implies that it's loading the wrong libraries [05:01:27 CEST] <furq> if you have ffmpeg installed through apt or whatever then get rid of it [05:02:13 CEST] <yenoh> I already removed it, maybe purge? [05:02:15 CEST] <yenoh> http://pastebin.com/MijAyTrF [05:02:18 CEST] <furq> try autoremove [05:02:41 CEST] <furq> alternatively just build with --enable-static [05:02:57 CEST] <furq> it's less hassle and won't cause conflicts with any other packages which depend on libav* (e.g. mpv) [05:03:20 CEST] <yenoh> I see. [05:03:27 CEST] <yenoh> thanks, lemme try :) [05:04:44 CEST] <yenoh> (just autoremove'd) [05:06:32 CEST] <yenoh> btw, after a ./configure I clearly see nvenc_hevc and nvenc_h264 in output [05:06:48 CEST] Action: yenoh goes make -j32 [05:06:53 CEST] <furq> that doesn't make any difference if it's loading the wrong libraries [05:10:12 CEST] <yenoh> seems right [05:10:21 CEST] <yenoh> after a configure and make [05:10:43 CEST] <yenoh> I got nvenc inside ffmpeg -codecs [05:11:34 CEST] <yenoh> ffmpeg doesn't complain about missing codecs any more \o/ [05:11:40 CEST] Action: yenoh iz happy [05:11:43 CEST] <yenoh> thanks furq [11:00:08 CEST] <brontosaurusrex> any cool denoisers? what would a good default atadenoise example be? [11:04:41 CEST] <streulma> hello is there a way to rebroadcast an mjpeg stream with lower framerate ? [11:42:45 CEST] <pi___> is there any option at resizing that san be use to round the automatic size ? [11:42:54 CEST] <pi___> [libx264 @ 0x1371e20] height not divisible by 2 (640x271) [11:43:03 CEST] <pi___> can* [11:44:42 CEST] <Mavrik> pi___, scale=640:-2 should work on newer ffmpegs [11:45:29 CEST] <pi___> let me test [11:45:31 CEST] <pi___> thx :) [11:46:41 CEST] <pi___> perfect [13:37:10 CEST] <ozette> i'm on ffmpeg 2.4.11, is -f segment supposed to print "[segment @ 0x2212d20] format hls not supported." to the console with my version? [13:37:23 CEST] <ozette> when i try to output a m3u8 playlist [13:40:40 CEST] <ubitux> 2.4 is not maintained anymore, and 2.4.11 isn't even the last revision of that branch [13:41:00 CEST] <ozette> very possible, i'm on a eol fedora [13:41:37 CEST] <ozette> 3.1.1 is the latet ffmpeg? [13:42:55 CEST] <JEEB> master is the latest, and FATE shows if it's completely broken or not [13:44:23 CEST] <ozette> master is the latest stable? [13:44:57 CEST] <JEEB> master is the latest state and FATE (fatebeta.ffmpeg.org) shows how the tests are handling [13:45:04 CEST] <ozette> oh [13:45:08 CEST] <JEEB> if FATE is green for your architecture/OS it means it's "stable" enough [13:45:42 CEST] <ozette> thanks [13:57:24 CEST] <ozette> ffmpeg -i <file> out.m3u8, gives me a playlist with 4 segments, i assume this is the default, what flag to use to create smaller/more segments? [14:11:17 CEST] <DHE> ozette: I'd check https://www.ffmpeg.org/ffmpeg-formats.html#hls you probably want at least -hls_list_size and -hls_time set [14:14:34 CEST] <ozette> DHE: i read this, but i must be doing something wrong, because no matter the value i give, the output is still the same [14:15:15 CEST] <ozette> i.e. 4 segments [14:17:37 CEST] <DHE> paste[bin] your full commandline [14:35:49 CEST] <ozette> DHE: http://paste.debian.net/785957/ [14:36:42 CEST] <DHE> you neglected to include -hls_list_size [14:36:55 CEST] <DHE> for a static video (not live/realtime) you want to use a value of 0 [14:40:13 CEST] <ozette> hmm .. [14:40:17 CEST] <ozette> http://paste.debian.net/785959/ [14:40:45 CEST] <ozette> it makes no difference [14:43:47 CEST] <DHE> if out.m3u8 specifies out0.ts, then it seems to have worked [14:44:02 CEST] <DHE> you can set the hls_time to 2 seconds, but if the keyframes only come every ~5 seconds then ffmpeg is forced to use that [14:44:45 CEST] <ozette> i see [14:46:35 CEST] <ozette> i find it odd this 19 second video only have 5 keyframesthen [14:47:17 CEST] <ozette> there's a lot of movement going on, so how does that work, can i confirm this somehow? [14:48:06 CEST] <ozette> 4 segments* [15:06:26 CEST] <DHE> keyframes in h264 come in 2 types: I and IDR. HLS can only split on an IDR type. [15:46:18 CEST] <ozette> DHE: split into frames where there are no frames? [16:27:18 CEST] <mp1> Hi [16:39:21 CEST] <aep4Ayai> hi. I've been trying to use ffmpeg to split long mp3 files into smaller, disjoint files with gapless playback as follows: [16:39:42 CEST] <aep4Ayai> ffmpeg -i input.mp3 -ss 00:00 -to 01:00 out1.mp3 -ss 01:00 out2.mp3 [16:39:50 CEST] <aep4Ayai> this works for some input.mp3 but not for others [16:40:07 CEST] <aep4Ayai> when it doesn't work, there is a noticeable "jump" between the tracks [16:40:36 CEST] <aep4Ayai> is there a way to make ffmpeg produce clean, disjoint outputs without reencoding [16:40:48 CEST] <aep4Ayai> I forgot that I also use -acodec copy for every output [16:40:54 CEST] <aep4Ayai> in the example above [16:42:17 CEST] <aep4Ayai> it's fine if the timestamps move a bit as long as the output is clean [16:52:27 CEST] <durandal_1707> short answer: no [16:53:10 CEST] <mp1> Hi :) [16:53:24 CEST] <mp1> I want use ffmpeg in my script php [16:53:31 CEST] <mp1> for record video from my acquistion card [16:53:36 CEST] <mp1> No problem for starting a record with php [16:53:42 CEST] <mp1> but how stop record of ffmpeg with php whitout kill a process ffmpeg [16:53:53 CEST] <mp1> Would you help me please :) [16:56:31 CEST] <aep4Ayai> durandal_1707: I don't know how the mp3 format works. Does it have something like keyframes at which a clean split would be possible? [16:56:53 CEST] <aep4Ayai> and if so: is it possible to have ffmpeg output the positions of those frames? [16:56:55 CEST] <DHE> every "frame" is a keyframe, but the length isn't a clean number [16:57:42 CEST] <DHE> there are ways with ffprobe -show_* commands which some other users have tried. I can't comment on the success level though [17:15:22 CEST] <aep4Ayai> is it possible to do something like -ss and -to with frame numbers? [17:15:33 CEST] <aep4Ayai> instead of timestamps [17:18:42 CEST] <kepstin> aep4Ayai: if you want to split mp3 files such that they can be rejoined later, you're better off using a dedicated tool that manipulates the mp3 structure directly rather than ffmpeg. [17:20:29 CEST] <kepstin> (keep in mind that the split files, when done like this, may have audible artifacts at the start/end of the segment; in order to get gapless playback, you actually have to concatenate them back into one file and decode, rather than decode them separately) [17:23:12 CEST] <aep4Ayai> I see. I've worked around this for this particular file by first decoding into wav, then splitting, and then reencoding as mp3 [17:47:02 CEST] <aep4Ayai> are other formats such as opus and vorbis more split-friendly? [17:49:14 CEST] <kepstin> not really - all audio formats use overlapping blocks, so you can't just split at a block boundary and get a clean break [17:49:53 CEST] <kepstin> opus and vorbis support indicating a number of pre-roll samples, so you can include an extra block-before or after and indicate - with sample accuracy - exactly where you want the cut [17:50:22 CEST] <kepstin> (this needs specialized tools, ffmpeg can't do this) [18:10:52 CEST] <aep4Ayai> alright thanks [19:17:48 CEST] <mp1__> Have you a solution for my problem ? ^ [19:37:18 CEST] <speedcuber> I have some analog video8 tapes I am using ffmpeg to capture video and sound with. This cmd works ffmpeg -f v4l2 -i /dev/video0 -c:v rawvideo -f alsa -i pulse video.mkv [19:38:00 CEST] <speedcuber> but I would like to capture lossless and later compress both video and sound. How would I go about doing that? [19:40:24 CEST] <speedcuber> It seems that rawvideo is being compressed with x264 and the sound is converted to vorbis. [19:50:46 CEST] <bp0> mp1__, send SIGINT once [19:52:03 CEST] <bp0> speedcuber, what format is the output of the video capture card? [19:53:33 CEST] <speedcuber> No video capture card. This camera is really old. Just RCA video outputs that go to a "easycap" usb using the usbtv007 chip. [19:54:24 CEST] <speedcuber> looks like I can choose between 720x576 PAL and NTSC [19:54:31 CEST] <bp0> did you try -c:v copy and -c:a copy? [19:54:33 CEST] <speedcuber> from that chip [19:55:56 CEST] <bp0> i only wondered if the card was sending mpeg or something already [19:56:24 CEST] <speedcuber> /dev/video0 but doing ffplay /dev/video does show video. Wonder if I can do a "live" ffprobe on /dev/video0 and get some raw info [19:56:26 CEST] <bp0> some usb captures do that, I've used -c:v copy with my pci capture card and got rawvideo in the mkv [19:57:07 CEST] <speedcuber> or just cat out /dev/video0 then probe it [19:57:24 CEST] <bp0> well, ffmpeg should tell you something about the input stream, Stream #0:0: Video: rawvideo (YUY2 / 0x32595559), yuyv422 etc. [19:57:37 CEST] <bp0> if it says rawvideo, then -c:v copy should be the solution [19:58:14 CEST] <speedcuber> yup see rawvideo -> x264 [19:58:32 CEST] <speedcuber> nice! Gonna give it a shot! [19:58:39 CEST] <furq> use -c copy [19:58:45 CEST] <furq> -c:v copy will still convert the audio [19:59:00 CEST] <speedcuber> ahh thx [19:59:10 CEST] <bp0> unless also using -c:a copy :) [19:59:17 CEST] <bp0> but yes, that is prolly better [19:59:19 CEST] <furq> right [19:59:30 CEST] <furq> you might also want to use some kind of lossless compression, e.g. -c:v ffv1 [19:59:34 CEST] <furq> rawvideo is pretty huge [19:59:50 CEST] <speedcuber> Have 18TB on this NAS =P [20:00:59 CEST] <speedcuber> OK it works! Is there a way to deinterlace this video? [20:01:17 CEST] <furq> -vf yadif -c:v rawvideo [20:01:36 CEST] <furq> there are other deinterlacers but yadif is probably the best one in ffmpeg [20:01:45 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#yadif-1 [20:02:03 CEST] <speedcuber> think I remember that one not being so destructive [20:02:22 CEST] <furq> probably -vf yadif=mode=1 [20:03:02 CEST] <furq> although that'll double the size with rawvideo [20:03:13 CEST] <speedcuber> So cool you can use ffplay right off the head on /dev/video0 ... [20:03:20 CEST] <furq> i assume your nas can deal with 30MB/s writes though [20:03:39 CEST] <speedcuber> It is a 6 disk ZFS NAS. [20:03:50 CEST] <speedcuber> reads at about 500MBs [20:04:09 CEST] <furq> if that's 6x6TB mirrors then you should be fine [20:04:10 CEST] <speedcuber> so yeah =) [20:04:52 CEST] <speedcuber> 6 WD Reds at 3TB each. In RAIDZ-2. [20:05:39 CEST] <furq> i have no idea how raidz2 performs but it should probably still be able to handle 30MB/s [20:05:50 CEST] <speedcuber> It can =) [20:06:10 CEST] <speedcuber> Not that super when it comes to IOPs tho but that is another story [20:06:19 CEST] <furq> my raidz1 can saturate my 1G nic anyway [20:06:31 CEST] <speedcuber> thats all you need. [20:07:27 CEST] <furq> fwiw you can use vapoursynth if you want nicer deinterlacing and you have a halfdecent cpu [20:07:32 CEST] <speedcuber> Truth be told I spent all my cash building this NAS and could not afford the 10G NiCs and switch ..haha [20:07:51 CEST] <furq> but if yadif mode 1 gives nice enough results then it's probably not worth the hassle [20:08:22 CEST] <speedcuber> Giving it a few min and gonna take a look. Seems to work tho.' [20:10:48 CEST] <bp0> is the easycap usb2.0? [20:11:22 CEST] <bp0> 720x576 rawvideo, hmm [20:12:40 CEST] <speedcuber> bk... [20:12:40 CEST] <mp1__> thx bp0 [20:13:10 CEST] <speedcuber> yeah I know what you thinking bp0 me too... its a usb 2 device [20:13:42 CEST] <speedcuber> Is this right? ffmpeg -f v4l2 -i /dev/video0 -vf yadif -c:v rawvideo out.mkv [20:13:51 CEST] <furq> -vf yadif=mode=1 [20:13:55 CEST] <furq> if you want 50fps output [20:14:01 CEST] <furq> which you probably do with yadif because it'll look nicer [20:14:19 CEST] <speedcuber> thx [20:14:25 CEST] <furq> oh and you want -c:a copy as well [20:15:24 CEST] <speedcuber> yeah the audio I am pulling from a line-in. [20:16:47 CEST] <speedcuber> bp0 its the usbtv007 version of the easycap. [20:17:50 CEST] <bp0> ok, I just wondered if that was too much for usb2.0, but maybe it isn't. [20:18:43 CEST] <kepstin> nah, raw SD video is no problem over usb2.0 (at least if it's 4:2:0 or 4:2:2 subsampled) [20:19:13 CEST] <bp0> alright [20:19:37 CEST] <speedcuber> good to know. Thought it would be to much too [20:21:49 CEST] <speedcuber> Stream #0:0 -> #0:0 (rawvideo (native) -> rawvideo (native)) [20:22:36 CEST] <speedcuber> looks good so far but I do get this at times - [video4linux2,v4l2 @ 0x249e140] Dequeued v4l2 buffer contains corrupted data (691200 bytes). [20:22:43 CEST] <speedcuber> what does that mean? [20:26:23 CEST] <speedcuber> furq yadif=mode=1 is REALLY nice! Thx [20:48:07 CEST] <speedcuber> Seems to not like the way I am tying to capture audio. Something with the -vf filter. Any tips? [20:48:11 CEST] <speedcuber> -f v4l2 -i /dev/video0 -vf yadif=mode=1 -c:v rawvideo -f alsa -i pulse -c:a TAPE5NO_RAW.mkv [20:52:16 CEST] <furq> -c:a copy [20:56:08 CEST] <speedcuber> furq wich the audio was presented in /dev/video0 but its not. There is some weird sound card thingy in this. [20:56:23 CEST] <speedcuber> arecord -l shows [20:56:25 CEST] <speedcuber> card 1: usbtv [usbtv], device 0: USBTV Audio [USBTV Audio Input] [20:56:25 CEST] <speedcuber> Subdevices: 1/1 [20:56:25 CEST] <speedcuber> Subdevice #0: subdevice #0 [20:57:59 CEST] <speedcuber> -f alsa -i pulse seems to grab it but it is being converted to vorbis [20:59:58 CEST] <furq> the command you pasted doesn't have any argument after -c:a [21:00:06 CEST] <bp0> as furq: -c:a copy, the "copy" is missing from the commandline you posted [21:00:26 CEST] <bp0> but aside, did you try using -f alsa -i hw:1 instead of pulse? [21:00:26 CEST] <speedcuber> ffmpeg -f v4l2 -i /dev/video0 -vf yadif=mode=1 -c:v rawvideo -c:a copy TAPE5NO_RAW3.mkv [21:00:42 CEST] <speedcuber> sry about that. This is the cmd I tried [21:00:47 CEST] <bp0> direct from the usb audio instead of via pulseaudio [21:01:17 CEST] <speedcuber> I would love to do that! Hate pulse. Gets in the way of everything! [21:01:25 CEST] <speedcuber> just not sure how to go about it. [21:02:49 CEST] <bp0> well your arecord output shows it as card 1, so -f alsa -i hw:1 instead of -f alsa -i pulse should do it [21:03:46 CEST] <bp0> details here: https://trac.ffmpeg.org/wiki/Capture/ALSA [21:04:55 CEST] <furq> -i hw:1,0 [21:04:58 CEST] <furq> card 1, device 0 [21:05:40 CEST] <furq> actually nvm hw:1 should pick the same device [21:18:44 CEST] <speedcuber> So close! Video works and sound works as soon as I put in -c:a copy like this [21:18:50 CEST] <speedcuber> -f alsa -i hw:1,0 -c:a copy -f v4l2 -i /dev/video0 -vf yadif=mode=1 -c:v rawvideo TAPE5NO_1.mkv [21:19:17 CEST] <speedcuber> I get Unknown decoder 'copy' [21:19:43 CEST] <speedcuber> if I leave out -c:a copy I get sound but it is converted to vorbis [21:20:14 CEST] <speedcuber> and a HUGE thx for the -i hw1,0 tip! [21:23:13 CEST] <kepstin> speedcuber: you have the '-c:a copy' in the wrong spot, it should be after all the inputs (put it beside the -c:v rawvideo) [21:23:53 CEST] <speedcuber> kepstin, I tried that but the filter started to complain and said to move it [21:24:08 CEST] <speedcuber> the -vf yadif filter that is [21:25:12 CEST] <kepstin> ... that makes no sense, unless you made a typo somewhere else in the command [21:25:27 CEST] <kepstin> i'd have to see the full command line and error to tell you what was actually wrong [21:26:03 CEST] <speedcuber> It works!!!! Thx guys! [21:26:07 CEST] <speedcuber> ffmpeg -f alsa -i hw:1,0 -f v4l2 -i /dev/video0 -vf yadif=mode=1 -c:a copy -c:v rawvideo TAPE5NO_1.mkv [21:56:04 CEST] <randng> Hello, I'm having trouble trimming a .MOV file: ffmpeg -i 2016_0729_145724_025.MOV -ss 00:03:48 -t 10 -async 1 RH.mov [21:56:19 CEST] <randng> when I run that command, I get: frame= 0 fps=0.0 q=0.0 size= 0kB time=00:00:00.00 bitrate=N/A speed= 0x [22:02:16 CEST] <furq> it'll take a while to seek [22:02:50 CEST] <furq> you can move -ss before -i if you want faster/less accurate seeking [22:04:18 CEST] <randng> furq: Yes, thanks! that worked. for some reason, i've never felt it take very long when I put -ss after -i [22:26:12 CEST] <shincodex> which fing file [22:26:19 CEST] <shincodex> Do i search for optiosn registeration [22:26:35 CEST] <shincodex> cause timeout on rtsp isnt working and windows code is blocking forever on ff_poll [22:27:28 CEST] <ljr1981> Hello everyone! [22:28:32 CEST] <ljr1981> I am looking for something like an FFMPEG Cookbook, which might describe common CLI-calls for various tasks. For example: If I want to do a cube-rotation cross-fade. [22:28:53 CEST] <ljr1981> Does anyone know of such a Cookbook resource or perhaps even a book in general? [22:30:44 CEST] <shincodex> options [22:30:46 CEST] <shincodex> timeout [22:30:49 CEST] <shincodex> rtsp [22:30:56 CEST] <ljr1981> For example: I see a book on Amazon for the "Basics", which I will purchase this weekend. However, I am looking for more advanced material as well. [22:30:58 CEST] <shincodex> is exiting avformat_open wtf ever [22:31:07 CEST] <shincodex> stimeout / timeout [22:31:20 CEST] <shincodex> i think docs for "timeout" on rtsp is lie [22:31:31 CEST] <shincodex> and http "timeout" is more correct and should be same for both [22:44:43 CEST] <ljr1981> Any FFMPEG books other than the "Basics" book on Amazon? [23:04:00 CEST] <JEEB> ljr1981: FFmpeg is a thing that changes quickly enough that no book could possibly cover it well enough. your best bet is to read the documentation and ask things. books are useful for generic subjects around FFmpeg, of course. [23:04:09 CEST] <JEEB> such as digital video, colorspaces and video|audio compression [23:04:49 CEST] <ljr1981> Thanks, JEEB! [23:05:04 CEST] <ljr1981> So, FFMPEG is relatively unstable in terms of API? [23:05:43 CEST] <ljr1981> (e.g. the FFMPEG people are changing it enough to where use-cases for it become obsolete or break often? If so, how often do you think?) [23:08:33 CEST] <speedcuber> Wondering if I should be using x265 as far as future proofing my files. Is there any evidence that suggests that is where mainstream is heading? [23:09:12 CEST] <TD-Linux> future proofing for what? [23:10:03 CEST] <speedcuber> compressing old video8 videos that are 25 years old. Wonder where things will be in 10 from now [23:10:29 CEST] <speedcuber> "proofing" is not the right word but you get the gist. [23:14:16 CEST] <speedcuber> btw is there a decent anti-shake filter that can be used when capturing video in ffmpeg? [23:16:06 CEST] <furq> speedcuber: if you have the space for it then something lossless like ffv1 is your best bet for archival [23:17:25 CEST] <speedcuber> furq I got the space so that is not really an issue. The file you helped me out with turned out to be [23:17:30 CEST] <speedcuber> 163G Jul 29 22:45 TAPE5_RAW.mkv [23:18:21 CEST] <furq> i wouldn't be particularly confident that h265 will ever have the sort of popularity that h264 has at the moment [23:18:50 CEST] <ljr1981> @furq: Why? [23:19:06 CEST] <furq> people aren't very happy about the licensing and there are credible license/patent-free alternatives now or coming soon [23:19:17 CEST] <ljr1981> gotcha [23:19:26 CEST] <ljr1981> lic. can be a deal-breaker [23:19:27 CEST] <furq> AV1's goal is to actually be patent-free, although who knows how that'll work out [23:19:44 CEST] <furq> VP9 is free for use under the terms of some shadowy deal google did with mpeg-la et al [23:20:34 CEST] <ljr1981> Can someone point me at a good "cubed-cross-fade" example for ffmpeg cli? [23:20:46 CEST] <speedcuber> please tell me there is not licensing issues with x265....always some leech sucking the teats of others work! [23:20:47 CEST] <ljr1981> (e.g. 3D cube) [23:21:01 CEST] <furq> speedcuber: of course there are [23:21:09 CEST] <JEEB> furq: they don't have any deal with mpeg-la AFAIK [23:21:13 CEST] <furq> patent issues, not software license issues [23:21:38 CEST] <JEEB> as far as I know it's just pure "it's patent free, we promise" kind of thing [23:22:03 CEST] <speedcuber> DFSG is not gonna like that [23:22:16 CEST] <JEEB> that said VP9 as a format definitely is having less patent-related issues than HEVC :P [23:22:22 CEST] <furq> http://arstechnica.com/information-technology/2013/03/google-mpeg-la-agree-… [23:22:32 CEST] <furq> there's no reference to an actual deal there but i can't imagine that happened without money changing hands [23:22:40 CEST] <JEEB> a) that's VP8 b) loldunno [23:22:48 CEST] <furq> The agreement also covers any codecs derived from VP8, including the VP9 codec that's currently in development. MPEG LA will also end the patent pool creation process for VP8. [23:23:05 CEST] <speedcuber> I actually use VP8 webm and vorbis [23:23:06 CEST] <JEEB> anyways, the issues with HEVC don't even come from MPEG-LA [23:23:27 CEST] <JEEB> it is the fact that companies decided they didn't get enough from MPEG-LA which led to the mess with HEVC [23:23:50 CEST] <JEEB> I think we currently have three theoretical patent licensing groups/companies asking for your money [23:23:52 CEST] <speedcuber> who are these wankers? [23:24:02 CEST] <speedcuber> sry could not help myself [23:24:21 CEST] <furq> speedcuber: if you're not doing business in the US then it doesn't matter anyway [23:24:21 CEST] <JEEB> we have MPEG-LA, then we have unnamed entities under "HEVC Advance" and and then we have Technicolor I think ? [23:24:54 CEST] <furq> oh nice [23:24:56 CEST] <JEEB> esp. the almost trollish pricings of the middle one led to companies thinking about alternatives more seriously [23:24:56 CEST] <furq> i thought it was just two [23:25:23 CEST] <JEEB> which thus currently makes VP9 with all of its uncertainty and the encoder being crappy the least bad alternative!? [23:25:34 CEST] <JEEB> as a format that is [23:25:45 CEST] <furq> as far as i can see the least bad alternative is to keep using h.264 [23:25:51 CEST] <JEEB> yes, that is true [23:26:08 CEST] <JEEB> I meant looking at the future [23:26:15 CEST] <JEEB> since HEVC isn't something you can deploy right now either [23:26:27 CEST] <JEEB> (you can, but with similar caveats to VP9) [23:26:51 CEST] <furq> it depends how little you care about mobile users' batteries [23:27:01 CEST] <JEEB> also the guy who optimized the VP9 decoder in FFmpeg noted he is doing a proprietary encoder [23:27:02 CEST] <speedcuber> I am actually started to look google more and more [23:27:17 CEST] <JEEB> which seems promising, unlike google's [23:27:40 CEST] <furq> it's a shame it's commercial [23:28:12 CEST] <JEEB> well, if VP9 takes off with it it will be a nice case of freetards' favourite format getting encoded with proprietary software [23:28:15 CEST] <JEEB> har har har [23:28:37 CEST] <furq> maybe av1 will be released next year in a usable state [23:28:46 CEST] <furq> which would be amazing except we'll all have had heart attacks from shock [23:28:52 CEST] <JEEB> yes [23:28:59 CEST] <JEEB> esp. given the major role of Google still in that one [23:29:05 CEST] <JEEB> and Google not giving two fucks about rate control [23:29:33 CEST] <JEEB> (Google representatives will outright tell you that they're totally focusing on it, but on the other hand libvpx doesn't even seem to have it on the roadmap :P) [23:29:37 CEST] <furq> isn't it pretty much vp10 with lapping transforms from daala [23:29:46 CEST] <JEEB> not sure they took that in [23:29:54 CEST] <JEEB> since that would require major rework [23:30:11 CEST] <JEEB> they took some simpler-to-implement things from daala though, I think [23:31:27 CEST] <speedcuber> Heard at work that intel is working on something that is GPL and even better then what is out there now. Gonna use you will need a Intel cpu for tho. [23:31:33 CEST] <furq> i thought the whole point was to avoid the kind of patent mess that DCT is mired in [23:32:04 CEST] <furq> speedcuber: intel are part of the aomedia group that are working on av1 [23:32:12 CEST] <furq> as are amd [23:32:26 CEST] <speedcuber> av1 ? [23:32:30 CEST] <furq> also gpl doesn't really matter, x264 and x265 are gpl [23:32:53 CEST] <furq> you still technically need an h.26[45] license [23:32:58 CEST] <furq> or three h.265 licenses apparently [23:33:18 CEST] <speedcuber> oh shit not that kind of story again... [23:33:21 CEST] <furq> https://en.wikipedia.org/wiki/Alliance_for_Open_Media [23:33:37 CEST] <speedcuber> reading [23:34:36 CEST] <furq> i guess they're big enough to just pay off any patent holders [23:35:22 CEST] <speedcuber> I love the US dong get me wrong but all this patent stuff ..... well. [23:35:54 CEST] <speedcuber> dont* *Friday night beers at work here in Sweden* [23:37:41 CEST] <speedcuber> Was nice that Firefox caved. Been using IceWeasel for years. Was kinda nice/weird to see it in the GNU/Linux Debian repos. [23:39:18 CEST] <speedcuber> From one thing to the next. Kinda liking the filters in ffmpeg. Anyway to add some cropping when I capture? [23:39:33 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#crop [23:39:52 CEST] <speedcuber> oh dang! the magic never stops [23:42:16 CEST] <JEEB> https://www.ffmpeg.org/ffmpeg-all.html is a simple way of having a single page you can ctrl+F through although it doesn't contain all the things (some things just don't have documentation) [23:48:29 CEST] <speedcuber> all right so I would guess this will crop 10% off ? crop=9/10*in_w:9/10*in_h ? [23:56:11 CEST] <kepstin> speedcuber: that'll give you the top left 90% [23:56:30 CEST] <kepstin> it's probably easier just to specify the size directly in pixels [23:59:54 CEST] <speedcuber> kepstin, yup. [00:00:00 CEST] --- Sat Jul 30 2016

1 0

ffmpeg-devel.log.20160728
by burek 29 Jul '16

29 Jul '16

[00:05:29 CEST] <jamrial> Chloe: looks like some de/muxers replaced it with codecpar's video_delay [00:06:28 CEST] <nevcairiel> video_delay is literally has_b_frames, the old name is rather crappy [00:11:06 CEST] <Chloe> thanks, and hopefully last question, what did flags become? codec_tag? [00:11:26 CEST] <nevcairiel> flags? [00:11:43 CEST] <nevcairiel> those are to control behavior of the decoders and such, dont exist in codecpar [00:16:39 CEST] <Chloe> I guess I wont be able to fully replace avctx then [00:26:39 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:60873bf992ea: avformat/utils: Fix find_stream_info not considering the extradata it found [00:26:40 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:9cb30f7a8805: avformat/hls: Fix regression with ranged media segments [00:26:41 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:4d85069e5dff: avformat/hls: Sync starting segment across variants on live streams [00:26:42 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:9884f17e343b: avformat/hls: Use an array instead of stream offset for stream mapping [00:26:43 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:83db3c84fa72: avformat/hls: Move stream propagation to a separate function [00:26:44 CEST] <cone-107> ffmpeg 03Anssi Hannula 07master:04964ac311ab: avformat/hls: Fix missing streams in some cases with MPEG TS [00:53:40 CEST] <iive> nevcairiel: actually video_delay is worse name, because audio_delay exists and it signal delay in time. (e.g. seconds), while if video_delay is like has_bf_frame, then it should be holding number of packets for reordering. [02:13:24 CEST] <cone-107> ffmpeg 03softworkz 07master:70c1647a3501: avformat/matroskaenc: Write duration early during mkv_write_header (Rev #3) [03:01:48 CEST] <cone-107> ffmpeg 03Burt P 07master:fb91143ef150: af_hdcd: Report PE as being intermittent or permanent [05:03:04 CEST] <Timothy_Gu> hmm, now CUVID is renamed to "NVDEC" [09:30:14 CEST] <cone-822> ffmpeg 03Clément BSsch 07master:4791716c1dca: lavfi/hdcd: mark pe_str as static and const [09:30:15 CEST] <cone-822> ffmpeg 03Clément BSsch 07master:37abc8cca256: lavfi/hdcd: fix style [10:31:27 CEST] <cone-822> ffmpeg 03Paul B Mahol 07master:d7ae4f79d364: avfilter/f_drawgraph: add another slide mode [12:14:06 CEST] <cone-822> ffmpeg 03Zhao Zhili 07master:65b2feb890db: avutil/mem: fix memleak [16:12:39 CEST] <cone-822> ffmpeg 03Matthieu Bouron 07master:83b9909398c6: Changelog: add entry for MediaCodec hwaccel [16:26:22 CEST] <ubitux> michaelni: so 17e7c03 should be a noop now or you see changes to take? [17:45:29 CEST] <michaelni> ubitux, yes i think it should be a noop you may or may not want to adjust indention [17:46:29 CEST] <michaelni> its cleaner to leave indention as is of course [18:04:39 CEST] <Chloe> I cant work out why avctx is null at utils.c:1214, would appreciate a little help :) https://github.com/joshdekock/FFmpeg/commits/codecpar also, general comments if you think my approach is wrong or whatever [18:56:07 CEST] <CFS-MP3> michaelni could you comment on this? http://ffmpeg.org/pipermail/ffmpeg-devel/2016-July/197159.html [19:09:46 CEST] <michaelni> CFS-MP3, does it work if you exchange the 2 mpegts_find_stream_type() cases ? [19:09:55 CEST] <michaelni> if so send a patch with that [19:10:38 CEST] <CFS-MP3> michaelni for that specific example yes, don't know if it breaks something else though... but will do [19:11:01 CEST] <CFS-MP3> do you need a patch over those, or a new one with everything from today's head? [19:11:04 CEST] <michaelni> CFS-MP3, ok, please keep the call xchange in a seperate patch [19:11:32 CEST] <michaelni> i mean first the exchange patch then the patch adding the new case [19:11:42 CEST] <michaelni> so noting is broken between them [19:12:19 CEST] <CFS-MP3> OK [19:12:21 CEST] <CFS-MP3> thanks [21:01:52 CEST] <Chloe> ubitux: I'm actually not sure how to convert ffmpeg.c to codecpar, as everything still needs to set a codecctx because from what I can see much of libav* still needs it [21:07:39 CEST] <Yagger> Hello. I have problems compiling ffmpeg on MacOSX (64bit) I get errors regarding direct and sse/mmx about the symbols only being in x86. [21:07:45 CEST] <Yagger> dirac [21:08:57 CEST] <Yagger> And I have no idea how I can fix that. [21:10:08 CEST] <jamrial> Yagger: that's something for #ffmpeg. this channel is for development of ffmpeg only [21:10:56 CEST] <Yagger> Okay. Will try there. Thank you. [23:06:27 CEST] <nevcairiel> Chloe: it still needs a avctx for decoding and practically everything in avcodec, the trick would be to stop using the deprecated st->codec and instead create a fresh codec context from codecpar, or everything in reverse for encoding. There is no way that it will use no codec context at all, it just needs to use it when appropriate [23:29:52 CEST] <Chloe> nevcairiel: and what to do about how Input/OutputStream passes around a codecctx? Do I replace that with a codecpar [23:31:04 CEST] <nevcairiel> i have no idea how ffmpeg.c works internally, but the key concept should be pretty simple: keep one codeccontext for encoding or decoding respectively, and use codecpar to fill it from a demuxer or create codecpar to configure a muxer [23:31:21 CEST] <nevcairiel> the only thing that goes away is AVStream->codec [23:31:27 CEST] <nevcairiel> not codec context itself [00:00:00 CEST] --- Fri Jul 29 2016

1 0

ffmpeg.log.20160728
by burek 29 Jul '16

29 Jul '16

[00:08:03 CEST] <Pxl_> lol kad makes a good point [00:10:51 CEST] <Pxl_> i literally just commented out my line and just replaced it with ffmpeg --version and it got skipped over..... [00:11:36 CEST] <Kadigan_KSB> Pxl_: can You at least provide us with some general are of the call? Like, snip out just the loop (and a few lines before/after) and post? [00:11:40 CEST] <Kadigan_KSB> (well, pastebin) [00:11:44 CEST] <Kadigan_KSB> area* [00:15:42 CEST] <Pxl_> http://pastebin.com/bvxP4ZeX [00:17:37 CEST] <Pxl_> and here is all the log produces http://pastebin.com/t8VjqPDX [00:18:31 CEST] <Kadigan_KSB> Oh goodie, I might finally find out how the hell to use ffmpeg to encode mpeg2-ts [00:18:31 CEST] <Kadigan_KSB> xD [00:18:43 CEST] <Kadigan_KSB> Is it DVD-compatible, by the way? [00:18:53 CEST] <Kadigan_KSB> No. [00:19:46 CEST] <Chloe> Pxl_: none of those statements would be true, so of course it wont execute ffmpeg [00:20:04 CEST] <Pxl_> well i just put some filler to take out the real stuff [00:20:20 CEST] <Pxl_> but it does make it to the loop [00:21:59 CEST] <Chloe> which loop? [00:22:06 CEST] <furq> [ "${ENDPOINT}" = "foobar" ] [00:22:09 CEST] <furq> that's true [00:22:10 CEST] <Pxl_> or conditional [00:22:31 CEST] <Chloe> furq: oh rip, I misread [00:22:41 CEST] <Pxl_> yea, it makes it in there is what i meant xD [00:23:01 CEST] <furq> Pxl_: that stderr redirect only applies to the second pass iirc [00:23:11 CEST] <furq> so if the first command fails the log will be empty [00:23:22 CEST] <furq> s/command/pass/ [00:23:26 CEST] <Pxl_> but like i said, i even replaced that 2 pass encode with just "ffmpeg --version 2>>{$LOG_FILE}" and it skipped it [00:23:39 CEST] <Chloe> Pxl_: reread what furq said [00:23:46 CEST] <furq> ffmpeg --version doesn't output anything on stderr [00:23:49 CEST] <furq> or -version, even [00:23:56 CEST] <furq> -version and -help output on stdout [00:24:31 CEST] <furq> the first thing i'd do to debug this is get rid of the redirects and run it with a terminal attached [00:26:24 CEST] <furq> as a side note, you don't need -x264opts to set keyint (-g), ref (-refs) or bframes (-bf) [00:26:33 CEST] <furq> and it should be keyint-min, not min-keyint [00:26:55 CEST] <furq> you should also be using -r 30000/1001 instead of -r 29.970 [00:29:16 CEST] <furq> you also shouldn't be converting mpg -> h264 -> h264, but i doubt it makes much difference at 65kbps [00:29:18 CEST] <Pxl_> THANK YOU [00:29:30 CEST] <Pxl_> furq, you are the man [00:30:03 CEST] <furq> thanks i'm pretty great [00:39:41 CEST] <Kadigan_KSB> Yeah, anyone could actually enlighten me as to why /1001? [00:39:46 CEST] <Kadigan_KSB> As in, the reason for its existence, [00:39:58 CEST] <Kadigan_KSB> not why we use it instead of writing out decimals, that I get. [00:40:30 CEST] <Kadigan_KSB> I assume it's somehow related to how chroma subcarriers in NTSC and PAL came about, but... [00:41:32 CEST] <furq> it's a hack to prevent the chroma signal phasing with the sound signal [00:42:04 CEST] <Kadigan_KSB> Oooooooh. [00:42:35 CEST] <Kadigan_KSB> Here's a question: does 60p actually exist, or is it just shorthand for 60000/1001? [00:42:54 CEST] <Kadigan_KSB> I ask because I often see people say "60 fps (59.94fps)" like they're equivalent, [00:43:03 CEST] <Kadigan_KSB> but NLEs offer both options as separate settings. I'm confused. [00:43:11 CEST] <furq> i think some cameras use 60fps [00:43:21 CEST] <furq> it's not used anywhere in broadcast afaik [00:44:27 CEST] <furq> this is one of the advantages of living in a pal territory [00:44:43 CEST] <furq> it almost makes up for all our DVDs being 4% too fast [00:46:00 CEST] <Kadigan_KSB> It really makes me wonder why /1001 rates would be used anywhere apart from digitization of old, analog footage. [00:46:28 CEST] <furq> presumably for compatibility with legacy equipment [00:46:30 CEST] <Kadigan_KSB> I mean, if broadcast is digital start-to-finish now (commonly), then I don't see the merit, and it's most certainly a big, big headache. [00:47:44 CEST] <furq> it didn't go straight from all analogue to all digital though [00:47:48 CEST] <Pxl_> furq is right. qam boxes still make up the majority of the cable footprint [00:48:00 CEST] <furq> and yeah it's not just production you've got to worry about [00:48:21 CEST] <Kadigan_KSB> No, it didn't go straight there -- but I would assume that there's no point in digitally creating /1001 footage, like for example on output from AfterEffects or similar. [00:49:09 CEST] <Kadigan_KSB> And yet, I'm quite regularly handed such footage. :/ [00:49:20 CEST] <furq> well it's going to end up as 23.97 anyway [00:49:35 CEST] <furq> or 29.97 rather [00:50:01 CEST] <libdavid> hi. i've been here yesterday, because i had some issues with using http-streams on the command line with the "movie" filter: http://pastebin.com/spWkPK0n [00:50:04 CEST] <Kadigan_KSB> In Poland it seems to end up as 50p or 50i, I think... Unless DVBT still relies on old stuff? [00:50:11 CEST] <Pxl_> well furq and kadigan, its 5pm so time to clock out. Thanks again for at least getting me to a point where i could debug a bit [00:50:33 CEST] <libdavid> just wanted to say that it's solved: hat to escape the colon (`:`) with two backslashes and now it works [00:50:38 CEST] <Kadigan_KSB> Pxl_: have a good one. :) Good luck with it. And do let me know when You've found the issue. [00:50:55 CEST] <Pxl_> will do! you in here alot? [00:51:10 CEST] <Kadigan_KSB> I typically idle, so put it in a private message. [00:51:16 CEST] <Kadigan_KSB> Please, and thank You. :) [00:51:22 CEST] <Pxl_> sounds good [00:52:54 CEST] <Kadigan_KSB> Meh. I assume it'll never be a solid, uniform standard, because we'll find something better still before current stuff fully develops. Ah, well. [00:53:14 CEST] <furq> it would probably be more confusing to give someone a 24fps clip these days [00:53:28 CEST] <furq> or 30fps ;_; [00:54:13 CEST] <Kadigan_KSB> I actually work for a small studio, so our process is 25fps start to finish. Well, not entirely true - we shoot 50i and then 'stretch' that to 25p via yadif (giving suprisingly good results). [00:56:52 CEST] <Kadigan_KSB> I suppose it doesn't really matter, in the end, because we get to create fun stuff and show it to others. How it happens doesn't really matter, so long as it does happen. I'm just a bit pissed when I hand over 25p footage to a subcontractor to do some VFX, and get a 30000/1001 clip in return. >.< [00:58:27 CEST] <Kadigan_KSB> I mean, when someone gives me 30000/1001 to work on, I work on 30000/1001 and return 30000/1001 unless asked otherwise. /rant [02:12:08 CEST] <CFS-MP3> hopefully someone is still awake :-) [02:12:08 CEST] <CFS-MP3> ffmpeg -txt_page 888 -i Sky_Box-20160725151400-20160725151900.ts -an -vn -map s -f segment -segment_time 10 -segment_format webvtt -scodec webvtt -segment_list_type m3u8 -segment_list "/var/www/html/sky_wtt/sky_subs.m3u8" WPEC.vtt [02:12:21 CEST] <CFS-MP3> Goal is generate webvtt from that file [02:12:28 CEST] <CFS-MP3> with a valid m3u8 [02:12:30 CEST] <CFS-MP3> I get this [02:12:56 CEST] <CFS-MP3> Input #0, mpegts, from 'Sky_Box-20160725151400-20160725151900.ts': [02:12:56 CEST] <CFS-MP3> Duration: 00:05:00.76, start: 18128.102667, bitrate: 8444 kb/s [02:12:56 CEST] <CFS-MP3> Program 3850 [02:12:56 CEST] <CFS-MP3> Metadata: [02:12:56 CEST] <CFS-MP3> service_name : Sky News HD [02:12:56 CEST] <CFS-MP3> service_provider: BSkyB [02:12:56 CEST] <CFS-MP3> Stream #0:0[0x200]: Video: h264 (High) ([27][0][0][0] / 0x001B), yuv420p(tv, bt709), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 50 tbr, 90k tbn, 50 tbc [02:12:57 CEST] <CFS-MP3> Stream #0:1[0x294](eng): Audio: ac3 ([6][0][0][0] / 0x0006), 48000 Hz, stereo, fltp, 384 kb/s [02:12:57 CEST] <CFS-MP3> Stream #0:2[0x243](eng): Subtitle: dvb_teletext ([6][0][0][0] / 0x0006), 492x250 [02:12:58 CEST] <CFS-MP3> Stream #0:3[0x905]: Unknown: none ([5][0][0][0] / 0x0005) [02:12:58 CEST] <CFS-MP3> Stream #0:4[0x409]: Unknown: none ([5][0][0][0] / 0x0005) [02:12:59 CEST] <CFS-MP3> Output #0, segment, to 'WPEC.vtt': [02:12:59 CEST] <CFS-MP3> Stream #0:0(eng): Unknown: none [02:13:00 CEST] <CFS-MP3> Metadata: [02:13:27 CEST] <CFS-MP3> what's most absurd is that at some point it worked, but how is not in the history... probably a leading blank or something [02:30:34 CEST] <furq> fwiw i don't think you need -an -vn -map s [02:30:47 CEST] <furq> it should only select the subtitle stream if you give it a subtitle-only container [02:42:20 CEST] <CFS-MP3> same thing without those, furq [03:47:44 CEST] <k_sze[work]> Is the h264_videotoolbox encoder supposed to be faster than libx264? [03:48:29 CEST] <furq> depends on your hardware [03:48:29 CEST] <k_sze[work]> (on a Mac, with an NVIDIA GeForce GT 750M, for example) [03:49:00 CEST] <furq> i believe it uses quicksync if your cpu supports it [03:49:09 CEST] <furq> although this is based on some forum posts, so don't take my word for it [03:49:35 CEST] <k_sze[work]> Hmm, I thought the point of it is that it uses the NVIDIA hardware. [03:49:50 CEST] <k_sze[work]> I *think* the encoder that uses QuickSync is a different one. [15:33:33 CEST] <squirrel> i'm trying to convert a video file to match my tv's available options. i use > ffmpeg -hide_banner -i %1 -vcodec mpeg4 -acodec mp3 %tmp%\tmp.avi. but the resulting file is 2x smaller than the original file and the quality is utter garbage, very blocky. any idea what could be wrong? [15:33:54 CEST] <squirrel> while converting, ffmpeg gives the following warning: Video uses a non-standard and wasteful way to store B-frames ('packed B-frames'). Consider using the mpeg4_unpack_bframes bitstream filter without encoding but stream copy to fix it. [15:34:43 CEST] <hzut> HI [15:35:44 CEST] <hzut> Please, how can say to the '-t' option to go to the end? [15:37:46 CEST] <hzut> Hellp [16:07:22 CEST] <DHE> squirrel: you didn't specify a bitrate, quality parameter, etc. the defaults are pretty bad. add "-b:v 3M" (3 megabits in this example) [16:07:31 CEST] <DHE> also, mpeg4 is pretty old. no h264 support? [16:09:56 CEST] <squirrel> DHE: by bad quality i mean that it *looks* like the resolution is 100x50 (while it really is about 700x300) [16:10:19 CEST] <squirrel> i'm pretty sure something other than poor default quality is wrong [16:10:29 CEST] <squirrel> and, no h264 [20:13:45 CEST] <farfel> I would like to convert a sequence of 12 bit 1920x1080 TIF images into a 12 bit 444 .yuv file [20:13:58 CEST] <farfel> is this possible with ffmpeg ? [20:14:25 CEST] <farfel> TIF images are RGB [20:15:17 CEST] <JEEB> I recommend building with the zimg library and thus enabling the zscale filter instead of the internal one [20:15:31 CEST] <JEEB> that should enable various pixel format conversions in high quality and correctness [20:15:53 CEST] <JEEB> zscale filter docs are @ https://www.ffmpeg.org/ffmpeg-all.html#zscale [20:16:01 CEST] <antonva> hrmmm [20:16:13 CEST] <antonva> webm streaming just broke on me all of a sudden [20:16:23 CEST] <JEEB> zimg being https://github.com/sekrit-twc/zimg [20:19:11 CEST] <furq> farfel: -i img%03d.tif -vf format=yuv444p12le out.yuv [20:19:19 CEST] <furq> or -i img%03d.tif -vf zscale,format=yuv444p12le out.yuv [20:19:26 CEST] <furq> if you have zimg [20:20:23 CEST] <farfel> @JEEB @furq thanks! so I have to do a custom ffmpeg build with zimg [20:21:00 CEST] <farfel> oh, just read furq's answer more carefully [20:21:09 CEST] <farfel> I will try that , thanks [20:21:21 CEST] <JEEB> I really don't recommend the internal swscale-based scaler/conversion thing [20:21:41 CEST] <farfel> JEEB ok, thanks for the heads up [20:21:49 CEST] <farfel> I don't really care about image quality at the moment [20:21:53 CEST] <farfel> just need a test image [20:21:58 CEST] <farfel> in that format [20:22:27 CEST] <JEEB> you can check if your current build by doing `ffmpeg -filters |grep zscale` [20:23:25 CEST] <farfel> thanks, I have a pre-build windows binary [20:23:34 CEST] <farfel> doesn't look like it has zscale [20:23:45 CEST] <farfel> there is libswscale [20:23:54 CEST] <JEEB> oh windows [20:23:56 CEST] <furq> swscale can do it but zscale is supposedly higher quality [20:23:56 CEST] <JEEB> https://ffmpeg.zeranoe.com/builds/ [20:23:58 CEST] <farfel> :) [20:24:04 CEST] <JEEB> this should contain zscale [20:24:20 CEST] <furq> oh neat [20:24:23 CEST] <furq> i didn't realise zeranoe had added it [20:24:44 CEST] <farfel> thanks very much [20:24:52 CEST] <furq> hopefully he remembered to enable simd, since it's still off by default in the last release version [20:25:04 CEST] <JEEB> yeah :D [20:25:13 CEST] <JEEB> the dev is a troll so welp [20:25:24 CEST] <furq> i gathered that from him calling his library "z.lib" [20:25:50 CEST] <furq> at least he had the good sense to not have it build libz.so [20:25:53 CEST] <furq> or libz..so [20:26:08 CEST] <JEEB> I was actually surprised that he's been making actually useful things, I just remember him trolling on #mplayer like ~10 years ago [20:27:05 CEST] <furq> the simd thing has been fixed in git anyway which is nice [20:27:11 CEST] <JEEB> nice [20:29:52 CEST] <flyBoi> Can anyone explain why the first command fails, while the second one succeeds?: [20:30:28 CEST] <flyBoi> ffmpeg -i input0.mp4 -filter_complex "[0:v]trim=start=1:end=4,setpts=PTS-STARTPTS[a]; [0:v]trim=start=90:end=92,setpts=PTS-STARTPTS[b]; [a][b]concat[c];" -map [c] out.mp4 [20:30:31 CEST] <flyBoi> vs [20:30:44 CEST] <flyBoi> ffmpeg -i input0.mp4 -filter_complex "[0:v]trim=start=1:end=4,setpts=PTS-STARTPTS[a]; [0:v]trim=start=90:end=92,setpts=PTS-STARTPTS[b]; [a][b]concat[out1]" -map [out1] out.mp4 [20:31:16 CEST] <flyBoi> the first one fails for: [AVFilterGraph @ 0x7f9800f00160] No such filter: '' [20:31:41 CEST] <furq> you've got a trailing ; after [c] [20:31:51 CEST] <flyBoi> omg [20:31:53 CEST] <flyBoi> thank you [20:32:04 CEST] <flyBoi> i thought c was some reserved value, man. [20:32:20 CEST] <furq> i'll be honest, that took me a minute [20:33:08 CEST] <nisthana> I am new to FFMPEG - I want to use FFMPEG in python but i am unable to find an SDK that will allow me to use encoding/decoding functions. Is there anything available for Python? [20:33:18 CEST] <furq> there's PyAV [20:33:23 CEST] <furq> i don't know how usable it is though [20:34:30 CEST] <nisthana> thanks will check out, so is FFMPEG used only by C++/C developers mostly? [20:36:31 CEST] <flyBoi> Node.js has a pretty decent wrapper library [20:36:35 CEST] <flyBoi> that's what I've been using [20:38:22 CEST] <furq> insert joke about javascript and semicolons [20:38:52 CEST] <flyBoi> forgot to mention- i'm a total masochist [20:42:18 CEST] <nisthana> thanks. My development is all in Python so I am looking for some kind of library. If I can't find, then I need to figure out how to invoke C function calls from python code [20:44:59 CEST] <flyBoi> could you invoke bash commands? like by using the subprocess module? [20:48:00 CEST] <nisthana> that will be too slow for my needs, I need to support millions of calls to convert AAC to LINEAR16 for a speech project that will use Machine learning to detect what user is talking about in a recording [20:52:51 CEST] <flyBoi> would it really be significantly slower than using a wrapper library? [20:53:37 CEST] <flyBoi> python isn't known for its blazing speed [20:53:38 CEST] <farfel> another question: for -i img%03d.tif -vf zscale,format=yuv444p12le out.yuv command line, my tif sequence is of the form foo_00XX.tif [20:53:50 CEST] <farfel> where XX is some number in the sequence [20:53:58 CEST] <nisthana> not sure I need to check [20:54:47 CEST] <farfel> I tried foo_0%03d.tif, but that doesnt work [20:55:34 CEST] <naildeca> I installed ffmpeg 3.0 and the delogo filter is missing. How do I install it? [20:56:37 CEST] <farfel> oh, never mind, figured it out [20:58:54 CEST] <farfel> so, ffmpeg is rejecting my 12 bit TIF files: "This format is not supported (bpp=36, bppcount=3)" [20:59:39 CEST] <farfel> does ffmpeg have to be custom build to support 12 bit TIFF ? [20:59:40 CEST] <flyBoi> naildeca: what error are you getting? [21:00:30 CEST] <naildeca> It says 'no such filter' [21:05:12 CEST] <flyBoi> copy and paste your line? [21:07:51 CEST] <naildeca> flyBoi: It says [AVFilterGraph @ 0x98fa400] No such filter: 'delogo' [21:08:20 CEST] <farfel> well, I switched to 8 bit TIF, but had to remove zscale to make it work [21:11:39 CEST] <Chloe> Yagger: you should probably pastebin a log [21:12:05 CEST] <Yagger> Chloe: Will do, I am just compiling as we speak to show the error message as it represents itself. [21:23:06 CEST] <farfel> anyways, thanks for your help [21:24:03 CEST] <Yagger> Chloe: Here is the pastern: http://pastebin.com/A818Ja78 [21:24:10 CEST] <Yagger> pastbin [21:24:17 CEST] <Yagger> Damn autocorrection... [21:25:41 CEST] <Chloe> Yagger; why the --extra flags? [21:26:56 CEST] <durandal_1707> farfel: what created such tiffs? [21:26:56 CEST] <Yagger> That is so it finds libressl [21:36:31 CEST] <Yagger> Chloe: Anything I can do to make it work? [21:38:18 CEST] <furq> Yagger: try with --disable-decoder=dirac [21:38:28 CEST] <furq> that seems like a bug though [21:40:11 CEST] <Yagger> furq: I have added the configure option, it will take some time to build. [21:40:23 CEST] <furq> are you running parallel make [21:42:18 CEST] <flyBoi> naildeca: sorry, I meant your command [21:43:51 CEST] <flyBoi> so `ffmpeg -filters` doesn't show delogo? [21:45:33 CEST] <Yagger> furq: I thought I was, but I am not. It is simply running a "make install" [21:45:45 CEST] <furq> make -j8 [21:45:49 CEST] <furq> or however many cores you have [21:46:41 CEST] <Yagger> furq: I have 8. [21:47:59 CEST] <naildeca> flyBoi: Correct, there is no 'delogo' in 'ffmpeg -filters' See this pastebin http://pastebin.com/gPE58j9M [21:58:47 CEST] <Yagger> furq: The build was successful by using --disable-decoder=dirac Thanks for the help! [22:00:54 CEST] <flyBoi> naildeca: I think you just have a weird version that didn't include all of the filters. Check: http://superuser.com/a/529697 [22:04:42 CEST] <naildeca> flyBoi: Thanks. I guess I'll uninstall what I have and try whatever Ubuntu has or try other versions. Again, thanks for your help. [22:12:30 CEST] <Pxl_> hows everyone doing [22:12:51 CEST] <antonva> Could be better. [22:12:59 CEST] <antonva> ffserver playing tricks on me [22:13:01 CEST] <antonva> ...again [22:13:08 CEST] <Pxl_> i know the feeling, been banging my head against a wall for 2 days [22:13:39 CEST] <antonva> What's the problem? [22:14:07 CEST] <Pxl_> [libx264 @ 0x597a0a0] ratecontrol_init: can't open stats file i have no clue what this error is about [22:15:12 CEST] <antonva> You doing a 2 pass encode? [22:15:35 CEST] <Pxl_> yepp [22:16:16 CEST] <antonva> hastebin/gist the full ffmpeg line? [22:17:25 CEST] <Pxl_> http://pastebin.com/0Q5uqTCL [22:20:25 CEST] <Pxl_> works fine from command line but not when i invoke it from java with real data to build the video. The actual video creation works great up until the 2 pass at the end [22:32:29 CEST] <DelphiWorld> alive DHE ? [22:33:07 CEST] <DelphiWorld> when should use baseline, main and high H.264 profiles? [22:33:20 CEST] <furq> when your playback device only supports baseline, main or high profile [22:33:39 CEST] <DelphiWorld> furq: so it's related only to playback devices? [22:33:44 CEST] <furq> pretty much [22:33:52 CEST] <DelphiWorld> any other factor? [22:33:53 CEST] <antonva> Pxl_: huh [22:33:54 CEST] <antonva> weird [22:34:02 CEST] <furq> baseline generally encodes faster because it disables a lot of features [22:34:06 CEST] <ferdna> has anyone streamed video to a roku using ffserver? [22:34:24 CEST] <Pxl_> yeah, makes absolutely no sense to me xD [22:34:47 CEST] <DelphiWorld> ferdna: forget ffserver [22:34:55 CEST] <antonva> Pxl_: Is it possible that your log file var is getting lost in a different path or something? [22:34:59 CEST] <ferdna> DelphiWorld, why? [22:35:00 CEST] <DelphiWorld> use nginx rtmp server to stream rtmp or hls, or use nimble streamer [22:35:05 CEST] <antonva> Eh [22:35:14 CEST] <furq> ffserver is unmaintained and likely to be removed soon [22:35:19 CEST] <antonva> rip [22:35:20 CEST] <DelphiWorld> furq: ya [22:35:26 CEST] <furq> and also i've never seen anyone successfully using it [22:35:28 CEST] <DelphiWorld> antonva: bgp [22:35:29 CEST] <ferdna> ohhh [22:35:35 CEST] <antonva> furq: I have [22:35:41 CEST] <furq> well i have but it was me in about 2008 [22:35:41 CEST] <DelphiWorld> i used it furq [22:35:44 CEST] <antonva> I use it a lot [22:35:55 CEST] <furq> and it stopped working at the next api break [22:36:04 CEST] <DelphiWorld> but i completly forget it [22:36:10 CEST] <DelphiWorld> now i use rtmp servers [22:36:23 CEST] <furq> yeah nginx-rtmp is pretty great [22:36:39 CEST] <antonva> I just want to use vpx/webm [22:36:45 CEST] <DelphiWorld> but no much update lately, no furq ? [22:36:47 CEST] <Pxl_> so my java kicks it off twice to generate two videos simultaneously, whereas when i was manually running it, i only kicked of one at a time. Could that interfere with the stats file [22:36:55 CEST] <furq> it's a shame it's not rtsp or else you could potentially do webm over dash [22:37:04 CEST] <ferdna> so will those streaming services work on roku? [22:37:15 CEST] <antonva> Pxl_: Uh, if they're writing to the same log it could [22:37:35 CEST] <DelphiWorld> furq: i was using ffserver just due to rtsp [22:37:50 CEST] <Pxl_> nevermind...just tried having java kick off one and it did the same thing [22:37:53 CEST] <DelphiWorld> but now i prefer to pay 25USD monthly and have stable nimble streamer [22:38:04 CEST] <furq> x264/aac works for my use case [22:38:09 CEST] <kepstin> Pxl_: make sure you run each process in a different working directory, and it should be fine [22:38:30 CEST] <kepstin> (or manually specify the stats file) [22:38:35 CEST] <antonva> ferdna: roku should support rtmp streams last I knew, but I've never operated one. [22:38:53 CEST] <DelphiWorld> furq, what proto you stream on? [22:38:59 CEST] <furq> rtmp and hls [22:39:06 CEST] <DelphiWorld> same as me [22:39:07 CEST] <Pxl_> dumb question, but how do i specify the stats file? [22:39:07 CEST] <ferdna> antonva, cool thanks [22:39:21 CEST] <antonva> ferdna: I'd google it to be sure [22:41:33 CEST] <DelphiWorld> vlc still dont play dash [22:51:36 CEST] <ferdna> antonva, do you know how to watch those streams on roku? [22:51:47 CEST] <ferdna> ohh wait you said you never operated one before [23:10:59 CEST] <Pxl_> sure enough. Having multiple encodes going at once and trying to use the same stats file was my issue. [23:34:14 CEST] <maelask> hi [23:34:27 CEST] <maelask> i'm having a little bit of trouble compiling the ffmpeg-full-git package from the AUR [23:34:46 CEST] <maelask> it says CUDA could not be found, however i do have the cuda binaries in my PATH [23:35:20 CEST] <maelask> is there any other thing i need from cuda in the path to be able to build the package? [23:35:49 CEST] <furq> maelask: check the end of config.log [23:35:57 CEST] <furq> you're probably missing a header or something [23:37:46 CEST] <maelask> http://pastebin.com/ZtvYXWDh [23:37:49 CEST] <maelask> this is where it fails [23:38:39 CEST] <maelask> up to that point, the headers do exist [23:40:06 CEST] <maelask> i'm probably missing the libraries path for cuda [23:40:14 CEST] <maelask> i'll try that and get back here if it fails [00:00:00 CEST] --- Fri Jul 29 2016

1 0

ffmpeg-devel.log.20160727
by burek 28 Jul '16

28 Jul '16

[00:03:07 CEST] <michaelni> Chloe, you mean the openh264 patch ? martin should have write access [00:05:14 CEST] <Chloe> michaelni: I meant the JACK patch. You said LGTM, which in previous occasions resulted in the other person pushing (and if they didn't have access then you would apply it yourself). but I'm not the maintainer of the JACK dev, so I wasn't quite sure if I was supposed to push it. [00:06:33 CEST] <michaelni> Chloe, yes "LGTM" replies mean you can push it [00:06:44 CEST] <Chloe> ok thanks for clearing that up :) [00:07:20 CEST] <michaelni> LGTM == "Looks good to me" [00:11:43 CEST] <cone-418> ffmpeg 03Josh de Kock 07master:c2fcf4fa0b12: avdev/jack: remove duplicated dispatch macros [00:51:59 CEST] <cone-418> ffmpeg 03dericed 07master:461073e80914: ffprobe.xsd: add missing timecode attribute to frameSideDataType [00:52:57 CEST] <wyatt8740> this is driving me crazy [03:17:38 CEST] <cone-418> ffmpeg 03James Almer 07master:54a0a52be100: checkasm/vp9dsp: use declare_func_emms in check_loopfilter [03:44:14 CEST] <wyatt8740> finally got it, but I have no idea how. setting CFLAGS to a literal '' (not unset) fixed SDL, though, somehow. [03:44:47 CEST] <Timothy_Gu> wyatt8740: so it turned out to be a SDL compilation issue? [03:45:02 CEST] <wyatt8740> yes, but I have no idea what was wrong that I fixed because CFLAGS wasn't set to begin with [03:45:31 CEST] <Timothy_Gu> okay, well glad to hear that [03:46:10 CEST] <wyatt8740> no idea how this guy figured this out but it fixed it. https://trac.ffmpeg.org/ticket/939 [03:46:24 CEST] <wyatt8740> (comment 5) [03:50:34 CEST] <wyatt8740> http://i.imgur.com/kaKcGCx.png [03:51:23 CEST] <wyatt8740> I have absolutely no idea why no one else seems to encounter this building for windows; it's happened to me for years, ever since 2010 or so when I started using mingw32+msys in winXP [03:52:18 CEST] <wyatt8740> and continuing to now when I'm cross compiling with mingw-w64 from cygwin [03:54:22 CEST] <wyatt8740> but for whatever reason, mplayer still worked with the 'sdl' output with the bad SDL build [04:43:16 CEST] <Compn> wyatt8740 : do what now [04:45:11 CEST] <Compn> are you using mingw or mingw-w64 ? [04:45:47 CEST] <Compn> mingw is pretty outdated now :\ [09:23:52 CEST] <cone-517> ffmpeg 03Martin Storsjö 07master:c5d326f551b0: Add an OpenH264 decoder wrapper [09:23:52 CEST] <cone-517> ffmpeg 03Martin Storsjö 07master:293676c47673: libopenh264: Support building with the 1.6 release [10:21:13 CEST] <rcombs> does swscale perform better at particular exact scale factors on x86? [10:23:52 CEST] <ubitux> there are multiple scaler based on filter size [10:24:05 CEST] <ubitux> which depends on i/o size [13:41:30 CEST] <Gramner> so according to the avx-512 manual opmask unpack instructions are inverted compared to all other x86 unpack instructions, because that obviously somehow makes a lot of sense apparently. e.g. punpckldq dst,a,b => aaaabbbb while kunpckdq dst,a,b => bbbbaaaa [13:42:19 CEST] <Gramner> although it's possible that the manual is just wrong and mixed up src1 and src2 I guess [14:00:16 CEST] <BBB> kunpckdq? [14:00:22 CEST] <BBB> is that like vunpackdq for zmm? [14:00:38 CEST] <Gramner> punpck for opmask registers (k1-k7) [14:00:50 CEST] <BBB> k? [14:01:15 CEST] <BBB> I clearly have not been keeping up with avx512 [14:01:18 CEST] <BBB> what is k? [14:01:21 CEST] <Gramner> most avx-512 instructions take an optional opmask for conditionally blending/zeroing individual elements in dst [14:01:51 CEST] <BBB> oh, sweet [14:02:41 CEST] <BBB> but so the opmask is not an imm8, but its a k register? [14:02:51 CEST] <Gramner> new registers [14:03:08 CEST] <BBB> interesting [14:03:21 CEST] <BBB> Im going to assume the manual is mixed up, otherwise it makes little sense [14:03:25 CEST] <Gramner> k0-k7, but k0 is a static reg with all bits set (so basically that's the opmask used if you don't specify any) [14:04:19 CEST] <BBB> is there also a vanilla unpackdq without a opmask? [14:04:26 CEST] <BBB> or is that the kunpackdq with k0? [14:04:55 CEST] <Gramner> that's for unpacking opmask registers with each other [14:05:35 CEST] <BBB> oh god [14:05:38 CEST] <BBB> I see now [14:05:53 CEST] <BBB> so they actually created a completely new instruction set for creating what is essentially imm8s? [14:06:10 CEST] <BBB> very engineery of them [14:06:31 CEST] <BBB> my brain says overdesigned, but maybe Im underestimating what you can do with this [14:06:33 CEST] <Gramner> opmasks are 64-bit when used with bytes registers (each bit corresponds to an element in the vector) [14:06:44 CEST] <Gramner> bytes instructions* [14:08:14 CEST] <Gramner> I guess it's possible to use them for some calculated conditional shenanigans [14:08:39 CEST] <Gramner> a lot of stuff is likely designed to make it easier for compilers to autovectorize things [14:09:42 CEST] <BBB> whenever I hear autovectorization, I have to think about this gstreamer blog post about it [14:09:48 CEST] <Gramner> in handwritten asm that uses opmasks you're most likely just gonna set up the required value in the prologue and then just that in every loop iteration [14:10:04 CEST] <iive> BBB: link? [14:10:12 CEST] <BBB> its from this guy that doesnt really understand assembly at all, but hes very excited to enable it and get things that are a lot faster - which doesnt happen [14:10:27 CEST] <BBB> and then he essentially finds out after a while that all autovec does is to use simd for memset(0) [14:11:02 CEST] Action: iive facepalms [14:11:10 CEST] <Gramner> one thing that could be really useful with opmasks is dealing with array overwrites (probably one of the design goals behind it) [14:11:26 CEST] <Gramner> just mask out all elements that are past the end of the array and you're golden [14:11:39 CEST] <BBB> gotta go, brb [15:11:31 CEST] <bofh_> BBB: autovectorizers start to almost become useful once you add restrict/__restrict to every single relevant pointer [15:14:30 CEST] <ubitux> i'm not sure i can actually merge the next commit [15:15:10 CEST] <ubitux> i mean, it's passing fate but it's not exactly an improvement [15:16:07 CEST] <ubitux> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/h264_slice.c;h=ee… [15:16:57 CEST] <ubitux> in this h264_slice_header_parse(), everything changing current_slice (basically the chunk on top) is supposed to go out of the function [15:17:13 CEST] <ubitux> and be called *after* h264_slice_header_parse() [15:17:36 CEST] <ubitux> but the chunk i hilighted is supposed to make a skip or not based notably on current_slice [15:18:01 CEST] <ubitux> so if i moved also it with the other chunk on top [15:18:13 CEST] <ubitux> it means the whole slice will be parsed completely [15:18:19 CEST] <ubitux> while it wasn't before [15:18:41 CEST] <ubitux> and this can actually be a problem [15:19:14 CEST] <ubitux> first it might be slower, but also a decoding failure that could have been avoided by skipping specific slices will now actually fail [15:20:04 CEST] <ubitux> one thing i'm thinking to do is skip the slices if and not return an error there is actually an error in that h264_slice_header_parse() [15:20:25 CEST] <ubitux> sorry mistyped [15:21:11 CEST] <ubitux> one thing i'm thinking to do is to ignore h264_slice_header_parse() return value if the skip code is set [15:21:23 CEST] <ubitux> but that won't prevent from at least trying to perform a whole parsing [15:22:14 CEST] <ubitux> but that's kind of tricky [15:23:48 CEST] Action: ubitux not happy [16:00:16 CEST] <mateo`_> ubitux: can i push a commit or will it annoys you regarding the merge ? [16:00:38 CEST] <ubitux> go ahead [16:01:13 CEST] <cone-958> ffmpeg 03Matthieu Bouron 07master:376d8fb2c574: lavc/ffjni: replace ff_jni_{attach,detach} with ff_jni_get_env [16:08:21 CEST] <ubitux> michaelni: available for some h264 testing? [16:15:13 CEST] <ubitux> michaelni: if you can test my merge-libav branch, that would be great [16:16:56 CEST] <ubitux> ah, i found an issue [16:17:39 CEST] <michaelni> ubitux, yes i can test [16:18:10 CEST] <ubitux> it seems to fail with fate + THREADS [16:18:15 CEST] <ubitux> :( [16:19:13 CEST] <ubitux> if you feel like looking at the diff and see something obvious, comment welcome ofc [16:47:49 CEST] <ubitux> ah, fixed it, i think [16:48:18 CEST] <ubitux> ah, no. [16:49:30 CEST] <michaelni> btw i didnt spot anythng from looking at the diff but it failed with frame threads while passing fate with slices [16:50:19 CEST] <ubitux> yeah i made the same observation [17:01:09 CEST] <ubitux> michaelni: fixed [17:04:47 CEST] <ubitux> (i dropped an essential chunk by mistake in the 2nd commit) [17:18:45 CEST] <ubitux> michaelni: i'll push in about 40 min unless you want to do extended tests [17:18:57 CEST] <ubitux> (which i'd recommend to do at the end of the wave though) [17:20:59 CEST] <michaelni> ubitux, iam not sure how you did this but none of the files i try crash, so i think you dont need to wait 40min [17:21:20 CEST] <ubitux> heh ok cool, i'll push then, thank [17:21:22 CEST] <ubitux> +s [17:22:29 CEST] <cone-958> ffmpeg 03Anton Khirnov 07master:8d36932c8d33: h264: move the block starting a new field out of slice_header_parse() [17:22:30 CEST] <cone-958> ffmpeg 03Anton Khirnov 07master:54dd9b1cdd9e: h264: set mb_aff_frame in frame_start() [17:22:31 CEST] <cone-958> ffmpeg 03Clément BSsch 07master:78c7197ea0e5: Merge commit '54dd9b1cdd9e54f1ee39ae25af0324f8aba2831b' [17:25:26 CEST] <cone-958> ffmpeg 03Clément BSsch 07master:f8ce1e828299: lavc/h264_slice: adjust a few line breaks to reduce diff with Libav [17:37:37 CEST] <cone-958> ffmpeg 03Anton Khirnov 07master:f966498e433f: h264: decode the poc values from the slice header into the per-slice context [17:37:38 CEST] <cone-958> ffmpeg 03Clément BSsch 07master:87d1f820591b: Merge commit 'f966498e433fead2f5e6b5b66fad2ac062146d22' [17:39:35 CEST] <ubitux> weekly call: still no one to update ffmpeg.c to codecpar? [18:08:41 CEST] <cone-958> ffmpeg 03Anton Khirnov 07master:debca90863e4: h264: store {curr,max}_pic_num in the per-slice context [18:08:42 CEST] <cone-958> ffmpeg 03Clément BSsch 07master:6c41eda18ec0: Merge commit 'debca90863e4ee53447efd02483c500f89766384' [18:09:39 CEST] <ubitux> michaelni: any comment on git show 17e7c03 -w ? [18:10:37 CEST] <ubitux> git show 17e7c03 -w -U10 is clearer [18:10:58 CEST] <ubitux> i'll try a merge if you want to test [18:13:28 CEST] <michaelni> ubitux, iam happy to test if you have a merge to test [18:16:22 CEST] <michaelni> i wonder if this is reimplementing some setup_finished check we have [18:16:32 CEST] <ubitux> yeah that's what i wonder [18:16:46 CEST] <ubitux> i don't know how i could merge that [18:17:29 CEST] <ubitux> if you see a way to reduce the number of differences... [18:17:38 CEST] <ubitux> i'll probably skip that one [18:18:55 CEST] <ubitux> it looks a bit sensible though [18:40:44 CEST] <michaelni> ubitux, posted a patch that would turn the merge into a NOP basically [18:41:10 CEST] <michaelni> not sure you want to change indention of anything n the merge so it matches closer [19:02:06 CEST] <Chloe> ubitux: is there a list of the codecpar changes? [19:13:06 CEST] <CFS-MP3> Not sure if here or #ffmpeg would be the best place to ask... is there a "best card money can buy for ffmpeg on linux"? FFmpeg being the only important tool (don't care about desktop, opengl, etc). I know there's a grid of supported chipsets etc but I couldn't find useful benchmarks. First use case would be decode many channels (UDP live stream) and produce different composites [19:14:01 CEST] <ubitux> michaelni: doesn't this check replace if(current_slice) with if(!current_slice)? [19:14:18 CEST] <ubitux> Chloe: it's basically to redo 15e84ed for ffmpeg [19:14:51 CEST] <ubitux> (according to bbf5ef9d, "This commit is a no-op. We want to do this ourselves, since avconv and ffmpeg differ quite a bit." [19:14:53 CEST] <ubitux> ) [19:16:42 CEST] <ubitux> michaelni: oh my bad, missed the || [19:17:13 CEST] <ubitux> i guess that's fine, but you know the code better than me [19:24:15 CEST] <Chloe> ubitux: I'll take a look at it when I get home [19:35:01 CEST] <cone-958> ffmpeg 03Xinzheng Zhang 07master:cd141e71bd34: avformat/flvdec: splitting add_keyframes_index() out from parse_keyframes_index() [19:35:02 CEST] <cone-958> ffmpeg 03Xinzheng Zhang 07master:ad14aab3b4f8: avformat/flvdec: parse keyframe before a\v stream was created add_keyframes_index() when stream created or keyframe parsed [19:35:03 CEST] <cone-958> ffmpeg 03Michael Niedermayer 07master:8385e1718ee1: avcodec/h264_slice: Make setup_finished check cover more cases [19:37:52 CEST] <BBB> jamrial: I think hes looking for input on whether its ok for vlc to use threaded+hwacceled libavcodec [19:38:42 CEST] <jamrial> BBB: right, i assumed he sent that patch thinking it was for our tree [19:38:46 CEST] <jamrial> i see now it was forwarded [20:06:24 CEST] <wyatt8740> oh, ubitux, thanks for the blog post about the HQX filter I requested a couple years ago. It's been insanely helpful [20:46:56 CEST] <ubitux> wyatt8740: i wrote that 2yrs ago you know [20:46:58 CEST] <ubitux> :D [20:47:07 CEST] <JEEB> :D [20:47:08 CEST] <ubitux> you should probably use xbr btw [20:47:16 CEST] <JEEB> better late than never... I guess? [20:47:23 CEST] <mateo`_> just about time ... ubitux [20:47:26 CEST] <JEEB> (kind of like my earthsoft DV demuxer and decoder patches) [20:47:37 CEST] <JEEB> (WIP since VDD '13 or so) [20:48:02 CEST] <ubitux> well i actually *posted* it 2yrs ago :p [20:48:14 CEST] <JEEB> yeah, with that difference of course [20:48:48 CEST] <JEEB> maybe when I'll have my first days off after working 1.5 years without vacation [20:49:02 CEST] <JEEB> and get something done [20:49:14 CEST] <mateo`_> Chloe: are you going to port ffmpeg to codecpar ? [20:52:43 CEST] <wyatt8740> ubitux: yeah XBR is better :p [20:53:08 CEST] <wyatt8740> Having trouble getting GLC to losslessly output video unfortunately [20:53:27 CEST] <wyatt8740> (so the pixels aren't exactly the same color and the filter doesn't work on it) [22:48:34 CEST] <Chloe> mateo`_: yes [23:02:24 CEST] <cone-107> ffmpeg 03Burt P 07master:bfe9155aee1c: fate: Add HDCD filter tests for false positive and error detection [23:02:24 CEST] <cone-107> ffmpeg 03Burt P 07master:91854b8ef644: af_hdcd: Improve HDCD detection [23:05:22 CEST] <Chloe> Why does ffmpeg.c use AVCodecContext->delay for audio at line 3010? [23:39:02 CEST] <cone-107> ffmpeg 03Burt P 07master:c7ba34b6ea37: af_hdcd: more comments in state struct [23:39:03 CEST] <cone-107> ffmpeg 03Burt P 07master:c26305f6aec0: af_hdcd: Add counter for cdt expirations [23:41:52 CEST] <jamrial> Chloe: should be the same as initial_pading, afaik [23:43:00 CEST] <Chloe> That would be my guess, I was just wondering because initial_padding was being used as well [23:44:45 CEST] <JEEB> yeah, there's like 3 different variables for initial delay by now ^^; [23:45:36 CEST] <Chloe> I'm just going to remove it and hope nothing breaks [23:46:18 CEST] <JEEB> whether it's correct or not, something will most probably result in something different :D [23:52:57 CEST] <Chloe> jamrial: do you know the new way to check if it has b frames? [23:54:45 CEST] <jamrial> Chloe: not really, but looking at 15e84ed it apparently just dropped that line [23:55:27 CEST] <Chloe> I was more thinking about ffmpeg_opt.c:1032 where it sets dts_heuristic to 1 if it has b frames [23:59:07 CEST] <JEEB> ooh, ffmpeg.c and all those little IF statements~ [23:59:24 CEST] <JEEB> (and talking more seriously, great that you're tackling that crapola) [00:00:00 CEST] --- Thu Jul 28 2016

1 0

ffmpeg.log.20160727
by burek 28 Jul '16

28 Jul '16

[02:19:48 CEST] <flyBoi> Is this an appropriate place for questions about the Node.js fluent-ffmpeg library? [02:20:29 CEST] <flyBoi> or is there a better one [02:23:59 CEST] <flyBoi> Well, I'll just go for it: [02:24:37 CEST] <flyBoi> I ahve one video as input. I want to take seconds 10-20 and 40-50 of that one file, and concatenate them to one output file. Is it possible to do this without copying the input file twice? [02:25:41 CEST] <flyBoi> Right now, I'm calling .addInput("input.mp4").seekInput(10).duration(10).addInput("input.mp4").seekInput(40).duration(10).mergeToFile("output.mp4", "tmp/output"); [02:25:51 CEST] <flyBoi> And it outputs a file, but it's pretty well corrupted. [02:26:54 CEST] <flyBoi> It's possible I just have to do each step in different ffmpeg commands, which is fine. But if there is a simple way, similar to what I'm doing above, that'd be killer. [02:29:51 CEST] <flyBoi> I think I'm just going to opt to keep appending to one output file over multiple commands. Feels like the best way to go. [02:46:22 CEST] <kepstin> you can definitely do that in a single command with the ffmpeg tool itself (listing the input file multiple times). No idea what the node library is getting wrong [02:50:06 CEST] <kepstin> you should look to see if there's a way to get it to print the generated command lines out so you can try to debug what it's doing [02:50:15 CEST] <kepstin> looks like there's too much abstraction going on :) [08:44:00 CEST] <kgkg> I have several IP cameras, where I use ffmpeg to add a audio source and change the transmission protocol from RTSP to RTMP. I would like to creat one stream where I rotate every few seconds through all the sources. I see that I can put them next to each other using overlay, but that is not exactly what I want. Any help? [09:24:02 CEST] <k_sze[work]> For H.264 video in MP4 container, is the profile and level supposed to be encoded in the stream? [09:24:09 CEST] <k_sze[work]> or in the header. [09:44:37 CEST] <k_sze[work]> When using the h264_videotoolbox encoder, what's the videotoolbox_vld pixel format? [10:12:18 CEST] <Mavrik> k_sze[work], afaik profile and level is part of SPS/PSS [10:12:25 CEST] <Mavrik> But dont't hold me 100% to my word. [10:13:15 CEST] <k_sze[work]> Looks like `ffprobe -show_streams` does show the profile and level, at least for the few video files I've tested so far. [10:34:53 CEST] <ritsuka> libavresample and libswresample are almost the same, any reason to use one instead of the other? [12:01:29 CEST] <whald> any arch users here having problems with the ffmpeg executable as of the last update? (i updated yesterday, ffmpeg 3.1.1) [12:02:37 CEST] <relaxed> be more specific [12:03:15 CEST] <mrelcee> 3.1.1 was crashing for me. [12:03:29 CEST] <whald> i'm using x11grab, h264 encode that, mux into mpegts and send it to a multicast address. [12:04:27 CEST] <whald> this is the sending side: http://pastie.org/10920491 [12:05:39 CEST] <relaxed> first, change "-r 60" to "-framerate 60" [12:07:15 CEST] <whald> relaxed, when i do this, i'm getting spammed with "past duration too large" messages. [12:09:14 CEST] <iive> whald: there was an ABI break from 3.0 to 3.1, it was fixed in 3.1.1 . However if e.g. mpv is compiled towards 3.1.0 , the ABI restore would break it with 3.1.1 [12:10:05 CEST] <iive> there is a patch to avoid the ABI use by mpv, but wm4 ignored it. [12:10:44 CEST] <iive> of course, this doesn't apply to ffmpeg own programs. [12:28:03 CEST] <whald> also, if some genius in the office messes with the network configuration, problems arise. oh my. :-) [14:32:34 CEST] <DelphiWorld> SUP [14:35:41 CEST] <DelphiWorld> any comparison with libx264 and libopenh264. [14:36:44 CEST] <furq> do you want a more detailed comparison than "x264 is better" [14:37:13 CEST] <DelphiWorld> furq: any use on libh264, when it could be used? [14:37:19 CEST] <DelphiWorld> just to know if we could prophit from it [14:37:33 CEST] <furq> openh264 is for realtime [14:37:48 CEST] <furq> idk if it does a better job than x264 for realtime, but x264 definitely does a better job in any other scenario [14:37:51 CEST] <DelphiWorld> realtime what? live stream? conferencing? [14:38:14 CEST] <furq> the latter [14:38:22 CEST] <DelphiWorld> conferencing... [14:39:11 CEST] <DelphiWorld> asm, nasm, ... yasm... lulz [14:39:12 CEST] <furq> it only does constrained baseline profile [14:39:22 CEST] <DelphiWorld> furq: lol i mostly do main [14:39:38 CEST] <furq> which is fine for realtime but useless for everything else [14:39:45 CEST] <DelphiWorld> yep [14:40:09 CEST] <furq> the other difference is that openh264 is bsd licensed [14:40:26 CEST] <DelphiWorld> ah, licensing... [14:40:30 CEST] <DelphiWorld> libx264 is gpled [14:40:31 CEST] <furq> which is nice but not really a big problem compared to the patent situation [14:42:11 CEST] <DelphiWorld> furq: if i dont sell the software it should be fine no? [14:42:18 CEST] <furq> emphasis on should [14:42:44 CEST] <furq> afaik you technically need a license regardless of how you're using it [14:43:00 CEST] <furq> there's just no point in them wasting their time enforcing it on people who aren't making money [14:43:00 CEST] <DelphiWorld> i see [14:44:11 CEST] <DelphiWorld> the only codec i ever bought is G.729 [14:44:29 CEST] <DHE> for h264 I think it's something like 250,000 players requires you pay them. but don't take my word for it, check the license yourself [14:44:45 CEST] <DelphiWorld> DHE: :-P [14:44:55 CEST] <DelphiWorld> if i get 100 player i should be cool ;) [14:45:01 CEST] <furq> you need a license regardless though don't you [14:45:14 CEST] <furq> it's just free under a certain usage threshold [14:45:20 CEST] <DelphiWorld> yes [14:46:02 CEST] <furq> or at least it is for h.265 [14:46:18 CEST] <DHE> yeah h265 licensing is dramatically different, and not in a good way for users [14:46:25 CEST] <DelphiWorld> H.265 is pretty far away [14:46:42 CEST] <furq> hopefully it'll never properly land [14:46:47 CEST] <DelphiWorld> Lulz [14:46:58 CEST] <DelphiWorld> someone do iptv here ? [14:47:01 CEST] <DHE> well, if hardware accelerated vp10 processing becomes a thing, that could happen [14:47:02 CEST] <furq> although doubtless apple will try their best [14:47:14 CEST] <furq> and yeah i'm thinking about vp10/av1 or whatever [14:47:17 CEST] <DelphiWorld> DHE: vp10 allready out ? [14:47:42 CEST] <furq> if those are in a decent state by the time hevc becomes widespread enough to use, i can see all the big players holding off [14:47:49 CEST] <furq> one can dream anyway [14:48:42 CEST] <DHE> DelphiWorld: might be confusing it with vp9 [14:48:48 CEST] <furq> is vp10 still a separate thing [14:48:51 CEST] <DelphiWorld> yes there's vp9 [14:48:56 CEST] <furq> i thought it had pretty much been merged into av1 at this point [14:49:22 CEST] <DelphiWorld> if i could encode to vp9 + opus and send to a streaming server it'll be awesome, fuck H.264 [14:49:29 CEST] <furq> you can do that with icecast [14:49:33 CEST] <furq> it didn't work very well when i tried it though [14:49:43 CEST] <DelphiWorld> icecast dont do yet dash [14:49:55 CEST] <DHE> but if my DLNA client can't play VP10 without a core i5 to do the legwork, I find that discouraging. [14:50:00 CEST] <furq> it still technically works in browsers [14:50:00 CEST] <DelphiWorld> actualy my plan is to do dash [14:50:04 CEST] <DelphiWorld> i'm using hls currently [14:50:09 CEST] <furq> but yeah it didn't work very well for me [14:50:16 CEST] <DHE> <DelphiWorld> someone do iptv here ? # hmm? [14:51:04 CEST] <DelphiWorld> DHE: i mean do someone work for a iptv company like i do [14:51:45 CEST] <DHE> yeah, my company does TV over fiber [14:52:20 CEST] <DelphiWorld> DHE: pm ? [14:52:36 CEST] <DHE> sorry, gotta reboot my PC. maybe in a bit. [14:52:51 CEST] <DelphiWorld> DHE: waiting for you, ding [15:00:44 CEST] <DelphiWorld> rebooting [15:46:44 CEST] <DelphiWorld> back [16:21:40 CEST] <qy> hi, I'm trying to reencode a video taken from a camera, from mjpeg to webm, have --enable-libvpx but it fails with "Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument" [16:21:50 CEST] <qy> here's the paste http://pastebin.com/02gWjCvb [16:21:56 CEST] <qy> what's wrong? [16:33:52 CEST] <jkqxz> qy: You're writing to mp4, not webm (and vp8 is not supported in mp4). Specify the output format explicitly ("-f webm") or make the file extension ".webm". [16:34:44 CEST] <DelphiWorld> wb DHE [16:35:31 CEST] <qy> jkqxz: thanks! This is a big world for someone that only needs to do some conversions from time to time... [16:40:00 CEST] <DHE> and he's gone. [16:46:02 CEST] <qy> jkqxz: I don't know if you're a developer of ffmpeg, but one suggestion if to add descriptive explanations for error messages like the one I had, so that common users like myself can know why the commands don't work [16:48:02 CEST] <jkqxz> The actual error was further up: "[mp4 @ 0x14d2c80] Could not find tag for codec vp8 in stream #0, codec not currently supported in container". [16:55:22 CEST] <qy> I've seen it now... thanks! [18:07:05 CEST] <DHE> DelphiWorld: hey you're back [19:18:08 CEST] <Polochon_street> Hi ! I got an audio file which ffmpeg has trouble to read/analyze via ffprobe, but gives proper output with mediainfo/exiftool/whatever. Where should I report that ? [19:20:21 CEST] <durandal_1707> on trac [19:28:39 CEST] <CFS-MP3> is there a "best card money can buy for ffmpeg on linux"? FFmpeg being the only important tool (don't care about desktop, opengl, etc). I know there's a grid of supported chipsets etc but I couldn't find useful benchmarks. First use case would be decode many channels (UDP live stream) and produce different composites [19:32:52 CEST] <kepstin> CFS-MP3: not sure what you mean by "best card". You're probably best off getting some big intel xeon chips and not bothering with a gfx card at all. [19:50:39 CEST] <CFS-MP3> kepstin That's what I'd like to find out for sure... by best card I mean "card that will be able to handle most decoding/encoding" :-) I know it a generic question [19:51:18 CEST] <kepstin> well, i gave you the general generic answer :) [20:03:56 CEST] <furq> CFS-MP3: if nothing else you can limit your search to recent quadros [20:04:11 CEST] <furq> but nvidia seem to be making it intentionally unclear what capabilities each card has [20:04:30 CEST] <furq> i'd tend to agree with kepstin though [20:05:16 CEST] <Polochon_street> durandal_1707: should I report it like a normal bug, or is there some dedicated part on trac ? [22:46:19 CEST] <flyBoi> Asked this yesterday, thought I found an answer, turns out I was wrong: [22:46:40 CEST] <flyBoi> Using fluent-ffmpeg, is there a way I can take the first 10 seconds and last 10 seconds of a video, and merge them into one output file? [22:57:55 CEST] <Bray90820> Can someone tell me what these error mean? [22:57:56 CEST] <Bray90820> http://pastebin.com/raw/hK2qja9X [22:58:14 CEST] <furq> [mp4 @ 0x7fb16c000600] Could not find tag for codec wmav2 in stream #1, codec not currently supported in container [22:58:29 CEST] <furq> is that not clear enough [22:59:54 CEST] <iive> mp4 doesn't support wma audio [23:00:03 CEST] <Bray90820> So basically it can't put the WMA into an mp4? [23:00:07 CEST] <Bray90820> Yes [23:00:26 CEST] <Bray90820> So how would I convert it to something else? [23:22:00 CEST] <Kadigan> Bray90820: just convert the audio (-c:a and friends). [23:22:25 CEST] <Kadigan> If it ends up out of sync, reconvert the entire thing or split and rejoin (in two separate commands). [23:23:05 CEST] <Bray90820> so remove copy? [23:23:16 CEST] <Bray90820> or what [23:23:27 CEST] <Kadigan> You can copy video and audio independently. [23:23:50 CEST] <Bray90820> If you look at my paste you will see I did one with "-c:a copy" and it still failed [23:24:05 CEST] <Kadigan> Like it says, WMA is not supported in MP4. [23:24:14 CEST] <Bray90820> Yes [23:24:24 CEST] <Kadigan> So yes, You need to re-encode the audio. [23:24:40 CEST] <Bray90820> Yes and my question is how would I re encode the audio [23:25:12 CEST] <Kadigan> https://ffmpeg.org/ffmpeg.html#Main-options [23:25:44 CEST] <Kadigan> To make it easier I will suggest using aac as the codec. [23:34:31 CEST] <Bray90820> Kadigan: What code would I use tho? [23:35:06 CEST] <Kadigan> Please read the manual, seriously. It's not hard, there are examples. [23:36:59 CEST] <Chloe> ffmpeg -i input.wmv -c:v copy -c:a youraudioformathere output.mp4 [23:37:06 CEST] <Chloe> or something [23:37:27 CEST] <Pxl_> hey, anyone ever had an instance where a whole ffmpeg command was being ignored? [23:40:52 CEST] <Chloe> what do you mean by 'ignored'? [23:43:30 CEST] <Pxl_> i have a bash script that runs a bunch of commands and at one of my later ones when i go to 2 pass encode at the end, my script completely skips over it. :P [23:43:49 CEST] <Chloe> upload the script? [23:47:15 CEST] <Bray90820> Chloe: Thanks that worked [23:49:27 CEST] <Kadigan> Pxl_: I've had that issue often, and no, I don't know what causes it. It's either that, or I get a "file not found" when it's clearly there. [23:49:39 CEST] <Kadigan> It only happens to me on Mac OS X, though... [23:50:13 CEST] <Chloe> '"file not found" when it's clearly there' this is generally because of dynamic libs not being there [23:50:32 CEST] <Kadigan> Chloe: but it works if I re-run the script (other files are then 'not found'). [23:50:34 CEST] <Kadigan> It's a random thing. [23:50:38 CEST] <Chloe> odd [23:50:42 CEST] <Kadigan> It is. [23:50:51 CEST] <Chloe> try a different shell? [23:50:56 CEST] <Pxl_> yeah, cant really upload the entire script for non-disclosure. but yeah its odd. tried manually running the command its skipping outside of the script and ran just fine ... glad to hear that im not the only one [23:52:21 CEST] <Kadigan> Luckily it doesn't happen often. [23:52:28 CEST] <Chloe> I'm not sure how you think we're supposed to help you with such little information, but whatever. [23:52:29 CEST] <Kadigan> In fact, I suspect it might have something to do with the path, not the file. [23:53:59 CEST] <Kadigan> It tends to happen in loops though, not when I spell the commands out. [23:54:12 CEST] <Kadigan> Maybe something with how I escape characters. [23:54:42 CEST] <Kadigan> In general, I'm calling it a bash issue, until I'm sufficiently certain of it being ffmpeg-specific. [23:54:47 CEST] <Chloe> Kadigan; which editor do you use? it may be inserting weird characters [23:54:55 CEST] <Kadigan> Chloe: TextEdit and Sublime Text. [23:55:09 CEST] <Kadigan> (TextEdit is OSX-specific I believe) [23:55:26 CEST] <Kadigan> And yes, I checked, but not thoroughly. [23:55:58 CEST] <Kadigan> I basically chucked it onto the "I suck at bash" pile [23:56:03 CEST] <Kadigan> and rewrote the stuff in PHP ;) [23:56:37 CEST] <furq> i think a faerie just died [23:57:06 CEST] <Kadigan> What? Because I rewrote it in PHP? It's a language I know moderately well, and it works for me. [23:58:20 CEST] <Kadigan> (at the very least, I can have a script written up in 3 minutes w/o having to constantly google for syntax - good enough for me) [23:59:39 CEST] <Kadigan> Hell, why am I even defending my use of it? It's there, it works, I can move on to more important stuff like sending deliveries to clients. [00:00:00 CEST] --- Thu Jul 28 2016

1 0

ffmpeg-devel.log.20160726
by burek 27 Jul '16

27 Jul '16

[01:31:24 CEST] <jamrial> ubitux: gcc 4.2 complains about "src/libavfilter/vf_curves.c:343: error: type qualifiers ignored on function return type" [05:31:27 CEST] <omerjerk> hi [05:31:42 CEST] <omerjerk> what would be an alternative of this function in the new api - https://github.com/omerjerk/FFmpeg/blob/alsenc_patch/libavformat/alsenc.c#L… [05:32:28 CEST] <omerjerk> there is a function avpriv_mpeg4audio_get_config in mpeg4audio.h but it has a 4th argument sync_extension and I've no idea what to pass there. [05:32:37 CEST] <omerjerk> any tips? [07:09:58 CEST] <nevcairiel> check what other users of that function pass there [07:10:03 CEST] <nevcairiel> or read teh code of the function [08:13:18 CEST] <omerjerk> is there a util function in ffmpeg to initialize an array to zero? [08:13:34 CEST] <omerjerk> I remember seeing it once before but can't seem to find it. [08:14:11 CEST] <Timothy_Gu> omerjerk: memset? or = { 0 }? [08:14:33 CEST] <Timothy_Gu> or AV_ZERO*? [08:23:37 CEST] <omerjerk> memset(last_acf_mantissa, 0, sizeof(last_acf_mantissa)) gives the error - warning: 'memset' call operates on objects of type 'int' while the size is based on a different type 'int *' [-Wsizeof-pointer-memaccess] [08:24:36 CEST] <ubitux> sizeof(*last_acf_mantissa) [08:24:44 CEST] <ubitux> * n [08:24:46 CEST] <ubitux> ? [08:25:31 CEST] <ubitux> how is last_acf_mantissa declared? [08:43:05 CEST] <omerjerk> ubitux, it is defined as int *last_acf_mantissa = ctx->last_acf_mantissa; [08:43:35 CEST] <ubitux> then n * sizeof(*last_acf_mantissa) with n the number of elements [08:44:34 CEST] <omerjerk> okay. thanks. [19:22:56 CEST] <durandal_1707> I will add overlayx filter? Everybody fine with that? [19:25:53 CEST] <cone-418> ffmpeg 03Dmitry Vagin 07master:dc151d138c9e: ffprobe: add missing PROGRAM_STREAM_TAGS case [20:24:27 CEST] <cone-418> ffmpeg 03James Almer 07master:be04c4aa0029: avformt/matroskaenc: undo an accidental revert by commit 5d48e4ea [20:59:06 CEST] <Timothy_Gu> So who's doing the merges these days? [20:59:31 CEST] <jamrial> more than one person [20:59:55 CEST] <Timothy_Gu> we're about 40 days behind right now... [21:00:08 CEST] <Timothy_Gu> merges getting harder I assume? [21:00:50 CEST] <jamrial> i think the current h264 batch was being problematic, yes [21:06:40 CEST] <cone-418> ffmpeg 03Luca Barbato 07master:e85d38c20a88: librtmp: Avoid an infiniloop setting connection arguments [21:07:34 CEST] <Chloe> Timothy_Gu: there are a lot more noops [21:07:48 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/3.1:86f922874042: librtmp: Avoid an infiniloop setting connection arguments [21:08:00 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/3.0:b62191f9c189: librtmp: Avoid an infiniloop setting connection arguments [21:08:21 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/2.5:52a9f668eed6: librtmp: Avoid an infiniloop setting connection arguments [21:08:40 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/2.6:a3fd8126445a: librtmp: Avoid an infiniloop setting connection arguments [21:08:52 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/2.7:2b6bc8e340ac: librtmp: Avoid an infiniloop setting connection arguments [21:09:04 CEST] <cone-418> ffmpeg 03Luca Barbato 07release/2.8:8887dafda12c: librtmp: Avoid an infiniloop setting connection arguments [21:16:32 CEST] <ubitux> Timothy_Gu: jamrial: i'll go back to merge tmr [21:16:52 CEST] <ubitux> jamrial: about curves, i haven't touched this part of the code in the recent commits [21:17:15 CEST] <ubitux> (and i don't understand the warning) [21:18:41 CEST] <jamrial> ubitux: probably the const [21:19:18 CEST] <jamrial> it talks about type qualifiers [21:20:07 CEST] <ubitux> ooh [21:20:12 CEST] <ubitux> i looked at the wrong line [21:20:34 CEST] <jamrial> lines 343 and 344, so the DECLARE_INTERPOLATE_FUNC macro [21:20:41 CEST] <ubitux> yeah you're right [21:23:32 CEST] <cone-418> ffmpeg 03Clément BSsch 07master:c6e900e9258a: lavfi/curves: fix meaningless const int returned value [21:58:07 CEST] <jamrial> BBB: anything blocking your vp9 avx2 and mmxext patches? [21:58:18 CEST] <BBB> me lazy [21:58:19 CEST] <BBB> I think [21:58:26 CEST] <BBB> let me push, sorry about that [21:58:45 CEST] <jamrial> haha, no worries :p [21:59:49 CEST] <BBB> still have to write avx2 loopfilter [22:14:52 CEST] <Timothy_Gu> ubitux: thx [22:15:08 CEST] <ubitux> i can't guarantee anything though [22:15:15 CEST] <ubitux> it's going to be a pain [22:45:42 CEST] <wyatt8740> hey, quick question ffmpeg-development related: is there a known reason for ffplay on windows to act like this? black screen with a tiny green box? It's compiled with mingw-w64; an older build I did from within MSYS worked fine and I can't figure out why this isn't. http://i.imgur.com/9Sn77DJ.jpg [22:45:53 CEST] <wyatt8740> hey, quick question ffmpeg-development related: is there a known reason for ffplay on windows to act like this? black screen with a tiny green box? It's compiled with mingw-w64; an older build I did from within MSYS worked fine and I can't figure out why this isn't. http://i.imgur.com/9Sn77DJ.png * sorry [22:46:16 CEST] <wyatt8740> (oh god just realized I spammed that; just meant to fix the link) [22:47:05 CEST] <wyatt8740> the old version of ffplay worked fine (2.3.3), and mplayer which I also compiled with mingw-w64 plays videos fine. The ffmpeg binary in this build works fine as well; it's just ffplay. [22:47:20 CEST] <JEEB> ffplay is meant to be a test thingy that you should be happy if it does anything at all in general. it is not really a player for production use [22:47:36 CEST] <wyatt8740> I understand that, but I'm trying to figure out why it's not working when it used to. [22:47:53 CEST] <JEEB> that would probably require some debugging or whatever :) [22:48:00 CEST] <wyatt8740> that's not very specific. [22:48:32 CEST] <JEEB> I don't see you enabling SDL at least? [22:48:38 CEST] <wyatt8740> I was wondering if anyone who had worked on ffplay and understands the program better (and has SDL experience) could guess what that little green block is [22:49:03 CEST] <JEEB> unless I'm missing that from your configure line [22:49:27 CEST] <wyatt8740> configuration: --arch=x86 --target-os=mingw32 --cross-prefix=i686-w64-mingw32- --pkg-config=pkg-config --prefix=/mingw --enable-nonfree --enable-version3 --enable-gpl --disable-w32threads --enable-libtheora --enable-libspeex --enable-libmp3lame --enable-libass --enable-iconv --enable-bzlib --enable-zlib --enable-libxvid --enable-libx264 --enable-libfdk-aac --enable-libvorbis --enable-libvpx [22:49:33 CEST] <wyatt8740> --enable-libopus --enable-static --enable-shared [22:49:44 CEST] <wyatt8740> usually ffplay just doesn't get built if sdl is not present [22:50:15 CEST] <wyatt8740> and SDL is autodetected, not passed manually [22:50:49 CEST] <JEEB> yeah, that was one of the ways you could have gotten it [22:50:50 CEST] <wyatt8740> and `ntldd ffplay.exe` confirms that it is linked to SDL.dll [22:50:54 CEST] <JEEB> alright [22:51:15 CEST] <JEEB> unfortunately, I've got no experience with SDL or ffplay - I just build mpv and its opengl renderer if I need a playback thingy [22:51:23 CEST] <wyatt8740> I use mplayer 1.2 [22:51:25 CEST] <wyatt8740> :\ [22:51:27 CEST] <wyatt8740> *1.3 [22:59:27 CEST] <wyatt8740> starting to suspect maybe it's a SDL bug? [22:59:35 CEST] <wyatt8740> gonna try compiling it myself since it's a shared library [23:00:58 CEST] <wyatt8740> btw, audio plays with ffplay; there's just no video and a tiny green box in the corner. [23:29:10 CEST] <cone-418> ffmpeg 03Ronald S. Bultje 07master:726501a34ea4: vp9: add 32x32 idct AVX2 implementation. [23:29:11 CEST] <cone-418> ffmpeg 03Ronald S. Bultje 07master:7ca422bb1b8f: vp9: add mxext versions of the single-block (w=4,npx=8) h/v loopfilters. [23:29:12 CEST] <cone-418> ffmpeg 03Ronald S. Bultje 07master:a4edaa027058: vp9: add mxext versions of the single-block (w=8,npx=8) h/v loopfilters. [23:29:16 CEST] <BBB> jamrial: done [23:29:44 CEST] <BBB> todo: loopfilter, and the never-ending quest for coef reading optimizations [23:29:54 CEST] <BBB> I bet it can be done, but itll be a hell of a job [23:30:01 CEST] <BBB> (coef reading opts) [23:31:19 CEST] <atomnuker> yep, coefficient reading has the largest overhead atm [23:38:36 CEST] <wyatt8740> okay, _still_ not working with a fresh SDL build [23:39:06 CEST] <BBB> atomnuker: wanna learn inline asm? :D [23:39:18 CEST] <BBB> or actually you could probably do it external also [23:39:24 CEST] <BBB> it just makes it even harder [23:42:02 CEST] <wyatt8740> I know some ASM but it's Z80 unfortunately [23:42:06 CEST] <wyatt8740> (and 6502) [23:42:23 CEST] <wyatt8740> so not very applicable [23:44:20 CEST] <BBB> old school [23:44:22 CEST] <BBB> ;) [23:58:21 CEST] <Chloe> michaelni: did you mean for me to push? [00:00:00 CEST] --- Wed Jul 27 2016

1 0

ffmpeg.log.20160726
by burek 27 Jul '16

27 Jul '16

[03:00:49 CEST] <Ravens_S> I am need of a little bit of help. I am trying to create a batch file that can mass convert an entire directory of music. The batch file will request the location of your files and where you want the files to go and in the middle it will convert that whole directory from say flac to mp3. Any suggestions? [03:03:21 CEST] <DHE> shell scripting? [03:03:33 CEST] <DHE> it's doable, but ffmpeg won't do all the work for you [03:03:45 CEST] <Ravens_S> In the form of window batch file. [03:04:31 CEST] <Ravens_S> I have a small script that I am have put together but cant figure out why it is not working.. and that is what lead me here. [03:05:22 CEST] <DHE> if you prefix your command with 'echo', a batch file will print out the commands it would run instead of running it. useful for debugging? [03:06:27 CEST] <Ravens_S> I have tried something like that [03:06:46 CEST] <Ravens_S> What I have come to is that my script stops working at this line [03:06:48 CEST] <Ravens_S> $ (for FILE in %inputpath%.flac ; do ffmpeg -i "$FILE" -f mp3 -ab 192000 "`basename "$FILE" .flac`%outputpath%".mp3" || break; done) [03:07:23 CEST] <furq> that doesn't look like batch [03:08:26 CEST] <Ravens_S> You know what you are right. I just looked at the source I got it from and it is bash... damn it [03:08:41 CEST] <Ravens_S> Back to the drawing board [03:08:51 CEST] <Ravens_S> Thank furq [03:31:43 CEST] <Ravens_S> Ok so I have done some googling and I came across this script. [03:31:45 CEST] <Ravens_S> FOR %%f IN (%input%\*.flac) DO ( echo Converting: %%f ffmpeg -i "%%f" -ab 320k -map_metadata 0 "%output%%\%~nf.mp3" ) [03:32:33 CEST] <Ravens_S> The only problem is that once I put in the input and the output it will only create one file instead of multi files [03:37:38 CEST] <KDDLB> why are you converting FLACs to MP3s? just asking [03:38:38 CEST] <Ravens_S> Well I am trying to create a script that will do if for you automatically and vise versa [03:39:09 CEST] <Ravens_S> It is ultimately a learning process for me [03:39:29 CEST] <KDDLB> ah [03:39:30 CEST] <KDDLB> ok [07:09:43 CEST] <mrelcee> i convert FLAC to ALAC and dump them in a demp folder, which every 5 minutes is scanned and if there's a folder in there it gets moved to my 'Automatically add to iTunes'. folder.. works quite slick. [07:09:49 CEST] <mrelcee> demp/temp [07:12:15 CEST] <mrelcee> i do my lifting in perl though, not shell script.... [09:03:38 CEST] <wouter__> I am using vaapi to encode my web cam as a live stream, but I am seeing high CPU usage on a single thread (70-80% on an i7-5557U), can this be reduced? Command is ffmpeg -vaapi_device /dev/dri/renderD128 -i /dev/video0 -vf 'format=nv12,hwupload' -c:v h264_vaapi -b:v 300k -r 30 rtmp://user:pass@server/live/stream [09:18:26 CEST] <chovy> can i draw text on an input.gif file? [09:19:34 CEST] <chovy> the problem i'm having is i have an input.mp4 and i generate a scaled down input.gif from it...but I can't seem to draw text on the scaled down .gif only the .mp4 which means the text shrinks when it scales. [09:21:45 CEST] <thebombzen> chovy: use -vf scale before you use -vf drawtext [09:21:56 CEST] <thebombzen> then it'll do the drawtext on the video post-scale [09:26:38 CEST] <chovy> thebombzen: thanks [09:27:16 CEST] <chovy> thebombzen: can you take a look at my script? [09:28:14 CEST] <chovy> i'm basically doing three ffmpeg runs [09:29:21 CEST] <chovy> i'm not actually scaling the video [09:29:26 CEST] <chovy> it only scales the .gif [09:30:01 CEST] <thebombzen> I'm actually going to bed now lol. it's 3:30 a.m. here [09:30:52 CEST] <chovy> thebombzen: https://gist.github.com/anonymous/20993ba3cc001c784731135e0b6e89a8 [09:31:24 CEST] <chovy> its quick. i'm thinking i have to add a command to scale the video down first. but this is going to take like 10 minutes to run [10:32:37 CEST] <nichego> i run ffprobe -count_frames on some files. it takes some time to run but does not output anything besides the information that comes from running ffprobe without that switch. what am i doing incorrectly? [11:02:26 CEST] <platzhirsch> If I stream something to ffmpeg which listens on rtmp, what would be a good way to switch the input stream dynamically to just black frames. Is that something which can be done within ffmpeg, do I need to build something with GStreamer here or any other ideas? [14:17:17 CEST] <_Vi> How do I play this RTP stream: `ffmpeg .... -c pcm_s16le -f rtp rtp://192.168.99.116:12346/` ? What FFmpeg command should stay on the other side? [14:21:54 CEST] <_Vi> If I try `ffmpeg -ar 44100 -ac 2 -f s16le -i udp://:12346 ....`, I hear the scrambled sound (because of RTP headers). How do I make FFmpeg interpret those headers (manually specifying all codecs and bit rates on command line instead of in SDP file). [14:31:17 CEST] <_Vi> Strangely, it works with pcm_s16be... [15:52:01 CEST] <libdavid> hi. i'm having some trouble with the "movie" filter [15:52:23 CEST] <libdavid> according to https://ffmpeg.org/ffmpeg-filters.html#movie-1 the file "can also be a device or a stream accessed through some protocol" [15:53:11 CEST] <libdavid> but when i try to submit an m3u8 file over http i get this: http://pastebin.com/spWkPK0n [15:53:56 CEST] <libdavid> how can i pass the movie filter an m3u8 file over http? [23:20:54 CEST] <gudenau> Hello! [00:00:00 CEST] --- Wed Jul 27 2016

1 0