[Ffmpeg-devel-irc] ffmpeg.log.20180107

[burek]

[00:48:12 CET] <qnord> I have a video with an audio stream, and I would like to mix in a piece of audio at a specific time in the video, how could I accomplish this with ffmpeg? So far I have tried: `ffmpeg -y -i in.avi -itsoffset 00:00:53 -i sound.mp3 -filter_complex "[0:1][1:0] amix=inputs=2" -c:v copy out.avi` but it plays it at the beginning instead of the specified time.
[00:54:20 CET] <XLS202> Hey, i have a small question. I use FFplay to play a rtmp stream. But the Problem is: the stream gets more delay by the time its playing. Do you have any idea how to fix it?
[00:56:01 CET] <mike_check> XLS202: -fflags nobuffer
[01:02:19 CET] <XLS202> i am using "ffplay -fflags nobuffer -x 1280 -y 720 rtmp://IP/live/erdbeere" is this correct? If it is.. then nobuffer doesn't work. At the beginning the delays is near 0,5seconds, but after 30minutes or so it is by 8seconds.
[01:04:01 CET] <mike_check> is that half a second at the beginning?
[01:05:18 CET] <tdr> your delay, do you mean the audio/video get out of sync or the there is buffereing lag or what
[01:05:58 CET] <XLS202> Yes, the 0,5 seconds are at the beginning.
[01:06:29 CET] <XLS202> And i mean the video and sound stay synchrone, but both are delayed
[01:08:40 CET] <mike_check> how are you measuring the latency?
[01:10:24 CET] <XLS202> with https://time.is i am streaming to another person, and she tells me how high the delay is.
[01:12:28 CET] <mike_check> are you streaming with ffmpeg?
[01:12:49 CET] <XLS202> I use OBS with an nginx server
[01:15:32 CET] <mike_check> I'd look into tweaking your server side but here are a few more parameters you can throw in with ffplay... "-fflags nobuffer -flags low_delay -flags2 fast -fast"
[01:16:58 CET] <XLS202> Well I think the server runs fine, because if she restarts ffplay the stream is perfectly in sync.
[01:17:09 CET] <XLS202> But tomorrow I will try the flags.
[01:18:36 CET] <furq> there's a ton of different buffers that could cause that
[01:19:32 CET] <XLS202> and how can I test which buffer makes the problem?
[01:19:57 CET] <furq> with difficulty
[01:20:04 CET] <furq> ffmpeg doesn't really let you control capture buffers and whatnot
[01:20:16 CET] <furq> no idea about OBS
[01:20:27 CET] <furq> nginx-rtmp has buflen, you could maybe tweak that
[01:20:39 CET] <furq> that defaults to 1000ms though
[01:20:39 CET] <mike_check> XLS202: maybe try with vlc to play?
[01:22:12 CET] <XLS202> Well if the stream is fresh opened there is nearly no dealy, so i don't think OBS or nginx-rtmp are making some trouble.
[01:22:23 CET] <XLS202> and i will try VLC
[01:25:07 CET] <XLS202> Well VLC has an initial delay from about 1,5 to 2 seconds (and i turned the network buffer to zero). But tomorrow I will try if it will stay at this delay.
[01:57:03 CET] <Psilovybin> could someone with an AMD processor duplicate this error for me?  https://ffmpeg.zeranoe.com/forum/viewtopic.php?f=7&t=5381
[01:59:06 CET] <jkqxz> Sounds like your card or drivers don't support H.265.
[01:59:24 CET] <jkqxz> What is the GPU?
[02:04:15 CET] <Psilovybin> A6 6310.  H.264 amf works perfectly
[02:04:46 CET] <DHE> you mean an AMD GPU?
[02:04:50 CET] <Psilovybin> does x265 work for you on AMD?  I was just curious...
[02:04:58 CET] <Psilovybin> DHE, AMD APU
[02:05:23 CET] <Psilovybin> R4 ATI GPU is what I have...
[02:05:25 CET] <DHE> I have ryzen, but that's it for AMD...
[02:05:28 CET] <DHE> so no...
[02:06:01 CET] <Psilovybin> DHE, aye, sorry, I meant, and ATI video... not AMD cpu...
[02:08:04 CET] <jkqxz> That's Beema, so VCE 2.  It doesn't support H.265 at all, you need at least Carrizo for that in APUs.
[02:08:52 CET] <jkqxz> (Anything with GCN 3, I think.)
[03:42:02 CET] <Psilovybin> jkqxz, hrmm.. it plays h265 just fine, i can encode h265 but it uses the cpu.. thank you for checking though, i appreciate it
[04:32:25 CET] <realies> i've converted an ogg to wav and mp3 with ffmpeg and they sound like they're missing the high frequencies
[04:32:47 CET] <realies> used ffmpeg -i pop.ogg pop.wav and ffmpeg -i pop.ogg pop.mp3 to generate https://dev.reali.es/thelounge.pop/
[04:33:05 CET] <realies> any ideas why they sound different? especially the wav version?
[04:33:22 CET] <therage3> the wav should sound identical to the ogg, afaik
[04:33:28 CET] <realies> therage3, give it a go
[04:33:30 CET] <furq> did you check spectrals
[04:33:42 CET] <realies> furq, will do, but ears dont lie
[04:34:02 CET] <realies> furq, spectrals = spectogram?
[04:34:04 CET] <furq> yes
[04:34:11 CET] <therage3> what furq said. spectrals and double blind testing are what you need to decide these things
[04:34:22 CET] <realies> https://dev.reali.es/thelounge.pop/
[04:34:23 CET] <furq> if there's an obvious difference then maybe replaygain tags or something
[04:34:27 CET] <realies> play the ogg and wav
[04:34:38 CET] <therage3> opening
[04:35:17 CET] <therage3> it's loading really slow
[04:36:10 CET] <furq> the spectrograms are identical fyi
[04:36:15 CET] <realies> furq, true
[04:36:22 CET] <realies> this makes it very strange
[04:36:54 CET] <therage3> let me confirm that with Spek
[04:37:01 CET] <realies> therage3, no need to
[04:37:12 CET] <therage3> no, because they do sound different to me too
[04:37:14 CET] <furq> https://0x0.st/sZVd.png https://0x0.st/sZVR.png
[04:37:15 CET] <furq> hth
[04:37:23 CET] <furq> they sound identical here but then my ears aren't particularly great
[04:37:36 CET] <realies> https://i.snag.gy/KyRh0a.jpg | https://i.snag.gy/taLnlm.jpg
[04:37:51 CET] <therage3> no, the Ogg sounds higher pitched to me, like it's biased towards the higher frequencies
[04:38:10 CET] <therage3> i don't know if that's what realies perceives as well
[04:38:16 CET] <realies> therage3, pitch is the same but it does sound a lot brighter
[04:38:22 CET] <therage3> right, that
[04:38:29 CET] <furq> what are you using to play them
[04:38:37 CET] <realies> furq, the same browser
[04:38:43 CET] <realies> i mean, its the same with chrome
[04:38:44 CET] <furq> which browser though
[04:38:46 CET] <furq> oh
[04:38:52 CET] <furq> i'm using firefox and they sound identical
[04:38:57 CET] Action: realies goes trying firefox
[04:38:59 CET] <furq> maybe chrome's ogg decoder is weird
[04:39:05 CET] <furq> or vorbis decoder rather
[04:39:26 CET] <realies> oh god
[04:39:41 CET] <realies> didn't see that coming
[04:39:49 CET] <therage3> i just tried with FireFox now, same difference between Ogg and Wav
[04:40:08 CET] <realies> therage3, no difference here with firefoxdeveloperedition
[04:40:13 CET] <therage3> what
[04:40:18 CET] <therage3> are you serious
[04:40:21 CET] <furq> lol
[04:40:23 CET] <realies> yes
[04:40:35 CET] <therage3> are you on FireFox 59
[04:40:36 CET] <furq> what os
[04:40:55 CET] <realies> 58.0b14 (64-bit)
[04:40:57 CET] <therage3> i'm on Antergos (so GNU/Linux)
[04:41:09 CET] <realies> macOS 10.13.2
[04:41:11 CET] <furq> it might use different decoders on linux
[04:41:16 CET] <furq> i forget if it does for vorbis
[04:41:23 CET] <furq> it definitely uses libavcodec for some stuff on linux
[04:41:31 CET] <therage3> i guess I could try asking the #mozilla people
[04:41:34 CET] <therage3> hm
[04:41:36 CET] <furq> but that might just be for patented stuff
[04:41:37 CET] <therage3> interesting
[04:41:51 CET] <therage3> and yeah, on Spek the two spectrograms are indeed mathematically identical
[04:41:56 CET] <therage3> like, they're the _same_
[04:42:12 CET] <therage3> so i'm also guessing some decoder here is being a little shit
[05:28:58 CET] <ccsardines> „„„„„„„„„„„„„„ LRH IS LIVE NOW!! TODAYS EDITION SLIMER GETS FUCKED IN VEGAS!! https://www.youtube.com/user/l0de/live CALL 315-505-4666 crvlgbn: PharaohAnubis st-gourichon-fid titou_ „„„„„„„„„„„„

[05:56:33 CET] <Spring>  so reading this thread it does seem like just using MadVR is the easier alternative for tone mapping HDR to SDR https://forum.doom9.org/showthread.php?p=1801754
[06:00:26 CET] <Spring> but that means encoding to HEVC...
[06:23:45 CET] <Spring> hold on, ffmpeg supports GPU accelerated HEVC encodes? Maybe it won't be so bad. Read something about encoding quality differences though but couldn't find much about it.
[06:40:12 CET] <chocolaterobot>   i have an OGG audio file that i want to upload to Youtube. So I understand that I need to add a picture file to it. So I have a jpeg file  in the same folder. How can I combine the two to make it Youtube-ready?
[06:40:16 CET] <chocolaterobot> i'm on Linux Mint cinnamon 18.3
[06:40:31 CET] <chocolaterobot> ffmpeg version 2.8.11-0ubuntu0.16.04.1
[06:55:27 CET] <fella> chocolaterobot: in the most simple form, this should do: 'ffmpeg -i snd.ogg -i pic.png out.flv'
[06:58:08 CET] <chocolaterobot> fella: thxn
[06:58:48 CET] <chocolaterobot> fella: it doesn't work though. it says `soundfile.ogg: Protocol not found`
[06:59:53 CET] <fella> well that's another topic then - seems like your version of ffmpeg lacks support for ogg
[07:01:20 CET] <chocolaterobot> fella: oh, i see. :( too bad
[07:01:30 CET] <chocolaterobot> i have lots of OGG files that I want to get worked on
[07:02:38 CET] <fella> ffmpeg -formats list
[07:06:43 CET] <furq> uh
[07:06:58 CET] <furq> that should definitely work with ubuntu's ffmpeg
[07:08:00 CET] <furq> chocolaterobot: does the input filename contain a : or something
[07:08:09 CET] <chocolaterobot> yes, furq
[07:08:18 CET] <furq> yeah it's misinterpreting that then
[07:08:21 CET] <furq> try file:filename.ogg
[07:08:22 CET] <chocolaterobot> furq: the filename starts with "2018-01-06-14:56:04_"
[07:08:45 CET] <chocolaterobot> furq: thaxns
[07:09:00 CET] <chocolaterobot> furq: that worx
[07:09:02 CET] <furq> ffmpeg -i file:filename.ogg -loop 1 -i image.png -tune stillimage out.mp4
[07:09:30 CET] <chocolaterobot> furq: how does that dif from fella's suggestion
[07:09:40 CET] <chocolaterobot> (besides the "file:" part)
[07:09:51 CET] <furq> ffmpeg -i file:filename.ogg -loop 1 -i image.png -tune stillimage -shortest out.mp4
[07:10:02 CET] <furq> you need -loop 1 or else it will only render one frame, which won't work
[07:10:09 CET] <furq> and flv defaults to really old codecs
[07:10:19 CET] <furq> mp4 defaults to x264 and aac which is what you want
[07:10:27 CET] <furq> actually no you don't
[07:10:33 CET] <furq> ffmpeg -i file:filename.ogg -loop 1 -i image.png -tune stillimage -shortest -c:a copy out.mkv
[07:10:36 CET] <furq> third time lucky
[07:10:50 CET] <furq> that'll keep the original audio without reencoding
[07:10:58 CET] <fella> furq: yours is the clean version, but to my knowledge mine should work too, just spitting an err at the end - that's all
[07:11:10 CET] <furq> yours will only encode one video frame
[07:11:20 CET] <furq> youtube won't really like that
[07:11:30 CET] <furq> but yeah don't use flv without specifying codecs
[07:16:22 CET] <chocolaterobot> thanks my F F friends.
[07:16:37 CET] <chocolaterobot> get it. FF for fella&furq. :)
[07:16:46 CET] <furq> i thought it was for ffmpeg
[07:16:55 CET] <furq> i guess that would have just been ffriends
[07:16:59 CET] <chocolaterobot> for both. :)
[07:17:27 CET] <chocolaterobot> i'm using youtube to transcribe speeches. :)
[07:36:46 CET] <Kevin> what's the default download directory for ffmpeg on windows?
[07:39:51 CET] <fella> Guest20314: i don't understand that question
[08:13:37 CET] <Guest20314> fella: when you run a command on windows to download from a url, where is the default directory it saves to?
[08:13:42 CET] <Guest20314> if no output path is specified
[08:14:37 CET] <fella> Guest20314: you use ffmpeg as a dl-manager?
[08:14:58 CET] <fella> and if so, i'ld say it should default to pwd
[08:15:09 CET] <Guest20314> what's pwd?
[08:15:12 CET] <fella> .. or wd, or '.'
[08:15:18 CET] <fella> your current dir
[08:15:18 CET] <Guest20314> im using mmfpeg to download ts videos
[08:16:44 CET] <fella> do you use it from a cli or does it have a gui?
[08:17:49 CET] <Guest20314> cli
[08:18:15 CET] <fella> so, there you go - it should be dl'd to your working / current dir
[08:18:30 CET] <Guest20314> yes it did, thank you
[08:18:33 CET] <fella> do you have a link to this mmfpeg?
[08:18:36 CET] <fella> yw :)
[08:18:53 CET] <fella> 'cause i can't find anything about it?
[08:19:14 CET] <Guest20314> https://ffmpeg.zeranoe.com/builds/
[08:19:42 CET] <Guest20314> its just the standard windows binary
[08:21:55 CET] <fella> but it's called ffmpeg as well, isn't it?
[08:22:30 CET] <fella> i thought there might be some strange binary called mmfpeg on win but i guess it was just a typo
[08:24:43 CET] <liyou> hi a
[08:26:00 CET] <liyou> hi, what is the mean of  'tbn'?
[08:29:32 CET] <JEEB> liyou: time base of the stream
[08:29:33 CET] <JEEB>         if (tbn)
[08:29:33 CET] <JEEB>             print_fps(1 / av_q2d(st->time_base), tbc ? "tbn, " : "tbn");
[08:31:43 CET] <liyou> JEEB, what is the mean
[08:32:52 CET] <JEEB> it is the amount of ticks in one second for that stream
[08:35:16 CET] <liyou> JEEB, Which word is abbreviated?
[08:36:13 CET] <liyou> 'tbn'
[08:39:34 CET] <JEEB> time base XXXX
[08:39:47 CET] <JEEB> but it reads the time base for the stream (as opposed to the decoder or so)
[08:43:12 CET] <liyou> JEEB, LOL, ' time base XXXX'. so you dont know the words ?
[08:45:01 CET] <JEEB> the last word I have no idea what it is supposed to mean :P but it uses the stream's (st) time base (time_base)
[08:45:20 CET] <JEEB> oh it might be the 1 / time_base :P
[08:45:25 CET] <JEEB> instead of just time_base
[09:37:35 CET] <Fyr> so, guys, still nobody knows how to make chapters in multiple languages?
[09:38:16 CET] <JEEB> I am not sure how many formats support that
[09:38:29 CET] <JEEB> you'd have to have the chapter points and then separately multiple language strings for them
[09:39:32 CET] <JEEB> last I remember you had to abuse editions in matroska to get multiple
[09:39:41 CET] <Fyr> I have an example only for single-language case. JEEB, do you have one for the multiple languages?
[09:40:07 CET] <JEEB> the problem is IMHO on a bigger level than FFmpeg itself
[09:40:08 CET] <JEEB> :P
[09:49:06 CET] <JEEB> yeh, looking at qtff.pdf it might be highly likely that MOV doesn't support it either
[09:49:23 CET] <JEEB> although its definition for the 'chap' atom is rather meth
[09:49:24 CET] <JEEB> *meh
[09:49:27 CET] <Fyr> Matroska does.
[09:49:37 CET] <JEEB> as far as I know only with editions hacking
[09:49:47 CET] <JEEB> and I'm not sure how many things support that
[09:49:56 CET] <JEEB> due to the lack of official definition of things in matroska
[09:50:19 CET] <JEEB> hmm
[09:50:20 CET] <JEEB> actually
[09:53:31 CET] <CoreX> does ffmpeg even support chapter making ?
[09:53:36 CET] <JEEB> yes
[09:53:42 CET] <JEEB> it has some basic utilities for it
[09:53:53 CET] <JEEB> the command line interface for it IIRC isn't too great
[11:23:51 CET] <Fyr> guys, how would this command look like if the original file had 6 audio channels:
[11:23:52 CET] <Fyr> https://pastebin.com/nUFBdRsW
[11:23:52 CET] <Fyr> ?
[11:26:22 CET] <ritsuka> JEEB: the way to have multiple languages chapters in mov is to have multiple chapters track, and set the chap atom in each audio track
[11:26:47 CET] <ritsuka> but you need to have a audio track for each chapter languages&
[12:11:05 CET] <JEEB> ritsuka: yes, something derp like that
[20:21:08 CET] <Leviathan_> Hi all, I'm trying to setup a webcam http stream on tiny core linux 8.2.1 with ffmpeg 1.2.2 but I don't find the right command line to use the webcam. Is someone know what I could try?
[20:34:02 CET] <durandal_170> ffserver removed
[20:38:42 CET] <Leviathan_> ffserver has been removed?
[20:39:02 CET] <kerio> rip
[20:42:23 CET] <tex> Can someone confirm if FFMPEG can accept UDP data as a raw stream which I can encode as part of a MPEG-TS?  I can confirm the UDP port is pushing the data out, but FFMPEG does not seem to be accepting/recording it as in input
[20:43:45 CET] <mike_check> tex: yes. confirmed if I understand you.
[20:44:06 CET] <mike_check> ffplay udp://...
[20:44:41 CET] <tex> I am using command:
[20:44:43 CET] <tex> ffmpeg -re -f dshow -r 15 -i video="screen-capture-recorder" -re -f data -i udp://127.0.0.1:1234 -map 0:0 -map 1:0 -c:v h264 -c:d copy -streamid 0:50 -streamid 1:497 -f mpegts e:\test.ts
[20:45:34 CET] <tex> the video is captured, but teh "data" is not. I Have read somewhere taht if the UDP input stream is NOT continous, then it may not work
[20:46:34 CET] <tex> if I use the -f bin for the input, then I just get two video streams in the Mpegts.
[20:46:46 CET] <mike_check> what is the data? did you try only the data?
[20:50:11 CET] <tex_> sorry mike, crashed out.
[20:50:28 CET] <furq> durandal_170: holy shit really
[20:50:40 CET] <furq> what amazing news
[20:50:41 CET] <tex_> I have tried capturing just the raw UDP data, and still no bytes captured
[20:51:09 CET] <tex_> but if I use something like wireshark, I can definatly see the data coming through no issues
[20:51:36 CET] <furq> https://github.com/FFmpeg/FFmpeg/commit/6b35a83214f1bc3fb38c9ea9c2cd3676f28709fa
[20:51:46 CET] <furq> not even the animes avatar can spoil this moment for me
[20:52:03 CET] <sfan5> tex_: tried "udp://127.0.0.1:1234/?listen=1" ?
[20:54:48 CET] <tex_> same result
[20:55:05 CET] <tex_> Does the stream have to be continous?
[20:56:40 CET] <tex_> If I start the "listening" on ffmpeg is just sits there waiting, until I kick off the streaming and then it will see the stream mapping to copy etc, but then it never increases in size.
[20:57:44 CET] <tex_> Is it looking for a specific frame size before it determines it is a correct frame
[21:00:51 CET] <atomnuker> furq: they can only improve the moment, no :)
[21:10:36 CET] <Leviathan_> is someone can help me with that error: /dev/video0: could not found codec parameters. I use this command: ffmpeg -f mpeg -r 30 -i /dev/video0 -vcodec mpeg2video output.mpeg
[21:12:29 CET] <SortaCore> I think -i should come first?
[21:13:44 CET] <Leviathan_> when i put it in first it says invalid argument
[21:13:56 CET] <tex_> the order looks right
[21:14:11 CET] <tex_> can it actually find the inoput file? /dec/video0
[21:14:59 CET] <tex_> confirm by just copying the file into the same path/location as ffmpeg is run in and dont use any paths
[21:16:24 CET] <Leviathan_> i'm trying to use it with a webcam not a video file. not sure i got the right command
[21:17:01 CET] <tex_> oh yup, you trying to access the webcame device
[21:17:09 CET] <tex_> sorry monday morning... :)
[21:17:10 CET] <Leviathan_> yeah
[21:17:27 CET] <Leviathan_> ho me its sunday afternoon
[21:17:43 CET] <Leviathan_> no problem
[21:18:15 CET] <tex_> ffmpeg -y -f vfwcap -i list
[21:18:28 CET] <tex_> what is the camera listed as  when you type that
[21:19:00 CET] <tex_> sorry tahts windows
[21:19:26 CET] <Leviathan_> i got unknown input format
[21:19:50 CET] <tex_> from the wiki...
[21:20:08 CET] <Leviathan_> i nerver been able to list my device
[21:20:55 CET] <tex_> ffmpeg -f v4l2 -r 30 -i /dev/video0 -vcodec mpeg2video output.mpeg
[21:25:21 CET] <Leviathan_> i tried differetnt verison of this command and each time the system freeze, i've wait for 15 min and nothing happen. I run just the core of tiny core linux with 524 mb ram and 2 gb of disk and its a vm in esxi server 5.5
[21:28:22 CET] <Leviathan_> I'm trying to setup an ipcamera mini os to use with webcam, so i want nothing else running but ffmpeg to stream my webcam, if you think about a better solution for me I would like to know it...
[21:30:11 CET] <tex_> have you checked the wiki
[21:30:13 CET] <tex_> https://trac.ffmpeg.org/wiki/Capture/Webcam
[21:30:29 CET] <tex_> I would start with the hardware part first
[21:30:40 CET] <tex_> cos I suspect taht is where your problem is rather than the software
[21:36:19 CET] <Leviathan_> I've just check if my webcam as detected with the instruction from the link you provided and my camera is detected, but there a couple more thing interesting I would like to try. I will try these and come back after. Thx for your help.
[21:37:14 CET] <tex_> no probs
[22:02:15 CET] <iranen> hi. can you test/use AV1 codec already with ffmpeg?
[22:04:36 CET] <tex_> iranen: not surw what you mean?
[22:05:37 CET] <durandal_170> iranen: no/no
[23:05:23 CET] <Leviathan_> Is someone can help me? I have a tiny core 8.2.1 with ffmpeg 1.2.2 and i use this command: ffmpeg -f v4l2 -i /dev/video0 http://localhost:8989
[23:06:26 CET] <Leviathan_> and i get the error unable to find a suitable output format for http://localhost:8989
[23:06:56 CET] <Leviathan_> http://localhost:8989 invalid argument
[23:09:37 CET] <DHE> HTTP doesn't imply a container format or codec information. you should explicitly set some
[23:10:44 CET] <Leviathan_> thx but can explain me how, not sure to understand
[23:21:59 CET] <tex_> Try making an rstp stream instead of HTTP
[23:22:14 CET] <tex_> and then use VLC to connect to trh RSTP stream
[23:23:14 CET] <Leviathan_> ok thx i'll try this
[23:23:51 CET] <tex_> or just use -f flv http://localhost:8989
[23:24:34 CET] <tex_> -f flv http://localhost:8989/test.flv
[23:32:33 CET] <FishPencil> Can I run ffmpeg without it changing console colors?
[23:34:39 CET] <furq> FishPencil: export AV_LOG_FORCE_NOCOLOR=1
[23:42:36 CET] <tex_> How do you encode a text file as a MpegTS stream?
[23:42:41 CET] <tex_> ffmpeg -f bin -i e:\fred.txt -c copy -f mpegts e:\fred.ts
[23:42:57 CET] <tex_> is not working for me
[23:44:10 CET] <tex_> keep on getting a dimension error
[23:44:29 CET] <tex_> Tried with a set buffer size also, but same issue
[23:52:33 CET] <sfan5> what would your text file represent?
[23:52:42 CET] <sfan5> you can just put it into some container
[00:00:00 CET] --- Mon Jan  8 2018