[Ffmpeg-devel-irc] ffmpeg.log.20151102

[burek]

[00:00:02 CET] <c_14> metadata disposition=default or something should (maybe) work
[00:00:13 CET] <c_14> I remember there being bugreports in the past, but I think they were fixed.
[00:13:44 CET] <debianuser> c_14: Found this patch: https://ffmpeg.org/pipermail/ffmpeg-devel/2015-January/167371.html but can't find a proper syntax for it... `ffmpeg -i 1.mkv -i 2.ac3 -map 0:0 -map 1:0 -map 0:1 -c copy -metadata:s:a:0 disposition=default all.mkv` adds "Metadata: DISPOSITION : default" to the stream but not the "default" flag... :(
[00:17:22 CET] <c_14> What about just -disposition default ?
[00:17:34 CET] <c_14> Looks like that's a per-file main option
[00:27:55 CET] <c_14> Or maybe you need both?
[00:59:10 CET] <c_14> debianuser: -disposition:a:0 default <- works for me
[01:46:19 CET] <axk4545> any efficient way to convert 100+ mp3s to ogg or flac?
[01:48:47 CET] <axk4545> they are in subfolders.
[01:53:10 CET] <furq> find -exec
[04:06:35 CET] <JRWR> I have a interesting issue
[04:06:46 CET] <JRWR> I have a x11grab stream to a RTMP output
[04:06:55 CET] <JRWR> with a mp3/icecast stream of audio
[04:07:14 CET] <JRWR> and the bloody thing keeps dying (dies because of the mp3 stream getting bad data)
[09:07:17 CET] <abd5932> Hello, I have 200 .png frames ( http://avtechscientific.com/orbiting_pipes.zip ) which I compiled into a video ( http://avtechscientific.com/orbiting_pipes.mp4 ) on my Debian8 like this: avconv -r 17 -f image2 -i orbiting_pipes.%04d.png orbiting_pipes.mp4 . I'm new to such video editing, but it looks to me that the video is not smooth... am I right? If yes, are there any additional options that can be used to make it smoother? Thank you!
[09:14:21 CET] <waressearcher2> abd5932: -r 30 ?
[09:15:05 CET] <abd5932> I need this to last exactly 12s
[09:15:20 CET] <waressearcher2> what is "not smooth" ?
[09:15:28 CET] <abd5932> the motion
[09:15:48 CET] <abd5932> or am I wrong?
[09:15:53 CET] <abd5932> I'm not sure
[09:17:49 CET] <waressearcher2> you want it like every frame to become more transparent and next frame apeared ?
[09:18:50 CET] <abd5932> frankly speaking I have no idea how to express it...
[09:20:27 CET] <waressearcher2> try to do "-r 34" and then "-filter:v setpts=2*PTS"
[09:20:45 CET] <abd5932> thanks I'm trying...
[09:20:46 CET] <waressearcher2> it should be same 12 seconds but with more frames
[09:35:14 CET] <abd5932> waressearcher2: thank you! it is better now, so -r 51 will correspond to -filter:v setpts=3*PTS?
[09:38:01 CET] <waressearcher2> 17*3=51
[09:38:36 CET] <waressearcher2> so yes
[09:39:16 CET] <abd5932> and how to do what you have meant with " every frame to become more transparent and next frame apeared"? - maybe it will be even better?
[09:39:32 CET] <waressearcher2> that I don't know
[09:40:59 CET] <abd5932> Ok, thank you for the tip anyway!
[09:46:04 CET] <waressearcher2> yw
[09:46:06 CET] <waressearcher2> !next
[09:58:57 CET] <leogsa> hello to all
[09:59:51 CET] <leogsa> I have a question and it's rather long (with full command-line output). what is the right way to ask it?
[10:01:53 CET] <dsl420> pastebin.com your command and output
[10:01:59 CET] <dsl420> ask your question here
[10:02:08 CET] <leogsa> thanx
[10:03:33 CET] <leogsa> I need to take audio-streams from several IP cameras and merge them into one file, so that they would sound simaltaneousely. (http://pastebin.com/ER740zT9)
[10:05:56 CET] <leogsa> ffmpeg -i rtsp://user:pass@172.22.5.202 -i rtsp://user:pass@172.22.5.202 -map 0:a -map 1:a  -filter_complex amix=inputs=2:duration=first:dropout_transition=3  -ar 22050 -vn -f flv rtmp://172.22.45.38:1935/live/stream1
[10:06:20 CET] <leogsa> result: I say "hello". And hear in speakers the first "hello" and in 1 second I hear the second "hello". Instead of hearing two "hello"'s simaltaneousely.
[10:09:23 CET] <leogsa> I found interesting detail when connecting 2 cameras (they have different microphones and I hear the difference between them): the order of "Hello"'s from different cams depends on the ORDER OF INPUTS.
[10:09:49 CET] <leogsa> with command   ffmpeg -i rtsp://cam2 -i rtsp://cam1 -map 0:a -map 1:a -filter_complex amix=inputs=2:duration=longest:dropout_transition=0 -vn -ar 22050 -f flv rtmp://172.22.45.38:1935/live/stream1   I hear "hello" from 1st cam and then in 1 second "hello" from 2nd cam.
[10:10:11 CET] <leogsa> with command  ffmpeg -i rtsp://cam1 -i rtsp://cam2 -map 0:a -map 1:a -filter_complex amix=inputs=2:duration=longest:dropout_transition=0 -vn -ar 22050 -f flv rtmp://172.22.45.38:1935/live/stream1  I hear "hello" from 2nd cam and then in 1 second "hello" from 1st cam.
[10:11:09 CET] <leogsa> so, that is my problem for the last week (http://pastebin.com/ER740zT9) full question and outputs
[10:12:18 CET] <JamJams> Hi I've got two inputs -i - -i my.aac, I found when there's a delay in the aac encode the audio completly drops from my final output. But it's normally fine.
[10:12:32 CET] <JamJams> Is there a way to make ffmpeg wait for the second input or something
[10:14:28 CET] <JamJams> Just to recap that, it does work fine 90% of the time but if there's ever a delay in the aac encode and it falls behind. The audio stream just gets totally dropped. Which isn't good.
[10:17:30 CET] <waressearcher2> leogsa: try to add artificial 1 second delay for first camera
[10:20:13 CET] <leogsa> <waressearcher2>: I can't. 1 second - is aproximate delay - it may be 0.8 seconds or 1,5 seconds. It it different every time - but it is always there. (((
[10:20:53 CET] <leogsa> waressearcher2: I can't. 1 second - is aproximate delay - it may be 0.8 seconds or 1,5 seconds. It it different every time - but it is always there. (((
[10:25:30 CET] <JamJams> I've read the docs I don't really see any wait options which is alarming
[11:06:24 CET] <abd5932> Hello, I have 2 .mp4 videos a.mp4 and b.mp4 - I would like to unite them, but take from the second only the part form the second X till Y - how do I do this? Thank you!
[11:21:30 CET] <dsl420> what did you try, what did not work?
[11:23:08 CET] <dsl420> https://trac.ffmpeg.org/wiki/Concatenate could help
[11:24:30 CET] <abd5932> thank you!
[11:29:02 CET] <abd5932> dsl420: I didn't see on this wiki how to concatenate only a part of a file, like from second 23 till 100 (and not the whole file)...
[11:31:30 CET] <dsl420> i would just cut b.mp4 first like ffmpeg -ss 00:00:23.000 inputfile.mp4 -t 00:00:77.000 -codecs copy output.mp4
[11:31:36 CET] <dsl420> i guess
[11:31:40 CET] <dsl420> not tested
[11:34:57 CET] <dsl420> and probably you could put those time related options into a concatenate command as well
[11:36:44 CET] <dsl420> oh and probably 00:00:01:17.000 is better than 00:00:77.000 ;)
[12:17:59 CET] <abd5932> dsl420: could you give an example, how I can "put those time related options into a concatenate command as well"? Thank you!
[12:36:20 CET] <abd5932> I get "Unknown input format: 'concat'" when trying to run: avconv  -f concat -i concat.txt -c copy output.mp4 ... avconv version 11.4-6:11.4-1 why?
[13:07:20 CET] <leogsa> waressearcher2: what is the right way to insert 1 second delay in my command?
[13:13:14 CET] <iive> abd5932: avconv is fork of ffmpeg... ffmpeg merges most of it, but they don't.
[13:23:04 CET] <IsntFunny> Is it possible to stream a video in realtime with ffmpeg?
[13:23:14 CET] <IsntFunny> So that i can open it with vlc player for example
[13:41:36 CET] <IsntFunny> anyone?
[13:42:17 CET] <JEEB> -re before your input parameter (-i)
[13:42:23 CET] <JEEB> that limits reading the file to realtime
[13:43:00 CET] <IsntFunny> how do i tell it to use a file in -i? i always get "No such filter"
[13:43:21 CET] <JEEB> then your command line is broken, post your full command line and terminal output on a pastebin and link here
[13:44:19 CET] <IsntFunny> http://pastie.org/10524269
[13:49:47 CET] <JEEB> unfortunately I have work to do, but there should be others around as well :)
[13:49:59 CET] <JEEB> jut wait for a while and you might get helped
[13:50:12 CET] <IsntFunny> Poor dead irc :(
[13:50:14 CET] <IsntFunny> But thank you ::)
[13:56:28 CET] <leogsa> i have a command ffmpeg -i rtsp://admin:12345@172.22.5.202 -i rtsp://admin:12345@172.22.5.202 -map 0:a -map 1:a -filter_complex amix=inputs=2 -vn -ar 44100 -f flv rtmp://172.22.45.38:1935/live/stream1   what is the right way to insert "adelay" in it? (I need to manually delay 1 strea for 1 second)
[13:58:39 CET] <DHE> you'll want to use the labeled version of the filter_complex
[13:58:59 CET] <DHE> -map is for selecting what input sources go into an output. you don't use it for filtergraphs
[14:00:16 CET] <IsntFunny> Can you explain better? :o
[14:00:29 CET] <IsntFunny> Oh wait that wasb't for me
[14:00:34 CET] <leogsa> i'm sorry - a mistake in comand - one of the inputs is 201 and other 202 - so, I have 2 inputs - and as far as I understand - I must use filter_complex
[14:01:08 CET] <DHE> are they audio-only sources?
[14:01:26 CET] <leogsa> <DHE> what is labeled version of the filter_complex
[14:01:46 CET] <leogsa> no, they are ip cameras
[14:01:55 CET] <DHE> then what are you doing with the video?
[14:02:21 CET] <leogsa> in this command - they do nothing with video
[14:02:34 CET] <DHE> so you want to discard the video and have an audio only output
[14:03:16 CET] <leogsa> yes, right. I want to take 2 audio streams from different ip cameras and strem them in one output stresm
[14:03:54 CET] <DHE> -filter_complex '[0:a] adelay=.... [0_delay] ; [0_delay] [1:a] amix=inputs=2 [audioout]'    -map [audioout]
[14:04:31 CET] <DHE> so this builds an audio filter chain (fix it up as you need) and then selects the [audioout] as being the only feed for the output
[14:06:12 CET] <leogsa> thanx, I'll try right now. (the reason why I'm trying to delay - is http://pastebin.com/ER740zT9) <DHE> - please give a little attention to it - may be you can help me?
[14:06:36 CET] <DHE> I'm busy and can't do any hand holding right now.
[14:06:43 CET] <DHE> that's intended to get you started
[14:13:42 CET] <leogsa> <DHE> - thanx anyway. may be you can tell when you will be free to ask you later. I'm trying to solve this problem for more then a week. (delaying - is a solution, but a bad one - because delay is variable - and my numbers in command should be corrected every hour - and this is not acceptable ((()
[14:13:49 CET] <IsntFunny> Any idea why this is not working? ffmpeg -re -i C:\\slack.mp4 -f rtp rtp://127.0.0.1:1567
[14:13:57 CET] <IsntFunny> Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument
[14:14:43 CET] <c_14> rtp isn't a format
[14:14:46 CET] <c_14> it's a protocol
[14:14:48 CET] <c_14> use -f flv
[14:15:19 CET] <IsntFunny> Data doesn't look like RTP packets, make sure the RTP muxer is used
[14:15:19 CET] <c_14> Or any format rtp supports
[14:15:26 CET] <c_14> hmm, ok
[14:15:29 CET] <c_14> Can you pastebin your output?
[14:15:32 CET] <IsntFunny> Are you sure? many pages show me to use -f rtp
[14:15:45 CET] <IsntFunny> https://trac.ffmpeg.org/wiki/StreamingGuide#StreamingasimpleRTPaudiostreamfromFFmpeg see here
[14:15:51 CET] <IsntFunny> "-f rtp rtp://127.0.0.1:1234"
[14:18:09 CET] Action: c_14 confused rtp with rtmp, can you pastebin your complete commandline and full console output?
[14:19:01 CET] <IsntFunny> so my command line is ffmpeg -re -f mulaw -i C:\\slack.mp4 -f rtp rtp://127.0.0.1:1234
[14:19:13 CET] <IsntFunny> this would work but VLC tell sme SDP is missing
[14:19:23 CET] <IsntFunny> when i remove the -f mulaw it cries a lot
[14:19:46 CET] <c_14> Wait
[14:19:58 CET] <c_14> Your mp4 is actually PCM mu-law?
[14:20:02 CET] <IsntFunny> no
[14:20:05 CET] <IsntFunny>  just tried it out
[14:20:19 CET] <IsntFunny> can't i auto detect the format?
[14:20:24 CET] <c_14> It does that normally.
[14:20:33 CET] <c_14> But without seeing your console output, I can't tell what could be wrong.
[14:20:42 CET] <IsntFunny> ok sorry
[14:20:45 CET] <IsntFunny> let me collect all
[14:21:09 CET] <IsntFunny> http://pastebin.com/ZJ9mySa2
[14:22:01 CET] <c_14> >[rtp @ 000000134c0b09a0] Only one stream supported in the RTP muxer
[14:22:22 CET] <c_14> Maybe try the rtp_mpegts muxer?
[14:22:26 CET] <IsntFunny> So how do i stream audi AND video? :o
[14:25:41 CET] <IsntFunny> It now tells me that the SDP format is missing
[14:25:48 CET] <IsntFunny> in vlc player, ffmpeg seems to be running
[14:26:36 CET] <c_14> Can you try using ffplay instead of vlc?
[14:26:53 CET] <IsntFunny> sure
[14:26:53 CET] <StephenS> is there a native support for face recognition in ffmpeg, or I need to use something like opencv?
[14:28:29 CET] <IsntFunny> Also, can i tell ffmpeg to repeat the input file when it's done?
[14:28:33 CET] <IsntFunny> like a looped stream
[14:29:04 CET] <c_14> StephenS: there might be a frei0r filter
[14:29:27 CET] <c_14> IsntFunny: -stream_loop
[14:30:53 CET] <StephenS> yeah frei0r needs opencv for face detection
[14:32:34 CET] <IsntFunny> c_14, tried ffplay but it doesn't show a lot... "   nan    :  0.000 fd=   0 aq=    0KB vq=    0KB sq=    0B f=0/0"
[14:33:40 CET] <c_14> IsntFunny: does it show the video?
[14:40:05 CET] <IsntFunny> No it doesn't
[14:40:13 CET] <IsntFunny> i executed it in command line tho, is this correct?
[14:40:25 CET] <IsntFunny> SDP required:
[14:40:25 CET] <IsntFunny> A description in SDP format is required to receive the RTP stream. Note that rtp:// URIs cannot work with dynamic RTP payload format (97).
[14:46:05 CET] <c_14> Aaah, right. the sdp information should be output to stdout in the ffmpeg process. You can output to a file with -sdp_file. You then need to point the player at the sdp file
[14:49:58 CET] <IsntFunny> How can i point the player to that?
[14:51:52 CET] <c_14> rtp://address/name.sdp
[14:51:59 CET] <IsntFunny> oooh ok
[14:52:31 CET] <IsntFunny> humm.. wait but
[14:53:01 CET] <IsntFunny> How do i know where the sdp file is?
[14:54:04 CET] <c_14> You could always just download/copy the sdp file to the other side and just play it directly on the file system iirc
[14:54:09 CET] <c_14> It should contain the ip address etc
[14:54:16 CET] <c_14> (It's been too long since I tested this)
[14:54:38 CET] <flux> I think I found a bug, my ffmpeg segfaults with -sdp_file :)
[14:55:37 CET] <c_14> flux: command?
[14:55:49 CET] <c_14> Reproducible?
[14:57:10 CET] <flux> av_sdp_create crashes on AVDictionaryEntry *title = av_dict_get(ac[0]->metadata, "title", NULL, 0); because on av_dict_get i = 56 and m->elem[i] is an invalid element
[14:57:27 CET] <flux> well yes, it's reproducible, but it doesn't probably help: ~/apps/ffmpeg/bin/ffmpeg -y -i 'rtsp://cam3.sec:554/user=admin&password=&channel=1&stream=0.sdp?real_stream--rtp-caching=100' foo.mov -sdp_file foo.sdp
[14:58:23 CET] <flux> in the call size ac[0]->metadata is bonkers
[14:58:34 CET] <flux> s/size/site/
[14:59:01 CET] <flux> as well as is the all of *ac[0]
[14:59:14 CET] <IsntFunny> It doesn't seem to work :(
[15:01:15 CET] <flux> oh ok, so it maybe works only if one of the destionations is an rtp
[15:01:28 CET] <flux> the manual page says: Print sdp information to file. This allows dumping sdp information when at least one output isn't an rtp stream.
[15:01:31 CET] <c_14> Yes, it will only work if one of the destinations is rtp
[15:01:36 CET] <c_14> However, it shouldn't crash
[15:01:56 CET] <waressearcher2> leogsa: no idea
[15:02:15 CET] <flux> I suppose it should then exit before doing that on print_sdp, after the loop
[15:03:28 CET] <leogsa> <waressearcher2> what is the right way to write to person? just write his name? or name in <>? or name and : ?
[15:03:34 CET] <c_14> flux: it shouldn't call print_sdp
[15:03:41 CET] <flux> c_14, so you're saying it should work only when there is an rtp output, but the documentation says the opposite?
[15:04:18 CET] <c_14> The documentation might be a bit imprecise (I wrote it) you always need at least one rtp output
[15:05:15 CET] <flux> c_14, hmm, so if I wanted to extract the 'sdp' information of a random rtsp stream, can I do that with ffmpeg?
[15:05:26 CET] <c_14> Not that I know of.
[15:06:14 CET] <IsntFunny> http://pastie.org/10524456 That's the error i get when trying to play with ffplay
[15:06:17 CET] <c_14> flux: the segfault doesn't happen on file input though, hmm. Do you happen to know a public rtsp stream I can test with
[15:06:19 CET] <IsntFunny> playing the sdp file that is
[15:07:20 CET] <flux> c_14, sorry, no idea. and 5 second googling doesn't find me one either :)
[15:07:52 CET] <c_14> IsntFunny: works for me with `ffmpeg -f lavfi -i testsrc -sdp_file test -f rtp rtp://localhost:5555' and then `ffplay test'
[15:08:41 CET] <c_14> flux: If I sent you a patch, could you compile ffmpeg with it and test?
[15:08:46 CET] <flux> yes
[15:09:17 CET] <IsntFunny> Why is it not working for me? :(
[15:09:40 CET] <c_14> IsntFunny: can you try with the exact command I wrote?
[15:09:53 CET] <IsntFunny> http://pastie.org/10524469 that's my current command
[15:10:05 CET] <IsntFunny> maybe bc i have two streams?
[15:11:12 CET] <IsntFunny> Your code is working
[15:12:19 CET] <c_14> IsntFunny: move -sdp_file after the slack.mp4
[15:12:58 CET] <IsntFunny> Hmm, did not work exactly
[15:13:12 CET] <IsntFunny> still "[udp @ 000000c763a17ae0] bind failed: Error number -10048 occurred"
[15:20:09 CET] <IsntFunny> c_14, made a comparement between our sdp files: http://pastie.org/10524498
[15:22:02 CET] <c_14> Use ports 1234 and 1235
[15:22:06 CET] <c_14> eh
[15:22:10 CET] <c_14> 1234 and 1236
[15:22:18 CET] <c_14> It looks like the port+1 is used by the player
[15:22:46 CET] <IsntFunny> oh
[15:23:05 CET] <IsntFunny> lol it works
[15:23:51 CET] <IsntFunny> But vlc palyer cannot play it
[15:24:50 CET] <c_14> flux: Try with this patch https://pb.c-14.de/t/kng.aLvfV4
[15:32:11 CET] <flux> c_14, yep, doesn't crash anymore
[15:33:59 CET] <flux> checked also that the master did crash with that particular patch removed, so you're good to go ;-)
[15:34:51 CET] <c_14> As for the documentation, do you think it's fine if I just add that at least one of the output formats has to be rtp?
[15:41:42 CET] <flux> c_14, well, I don't quite understand what "This allows dumping sdp information when at least one output isn't an rtp stream." is trying to say.. so when you're outputting to both rtp and non-rtp, you get an sdp file out?
[15:41:58 CET] <flux> or is the purpose that when you output stp, it generates an sdp file for you
[15:42:13 CET] <c_14> yes, before the patch, the rtp information would be printed to stdout. This, however, wouldn't work when say you also output to stdout
[15:52:13 CET] <c_14> So, it can actually be used for both.
[16:28:44 CET] <debianuser> c_14: Confirmed! `ffmpeg -i 1.mkv -i 2.ac3 -map 0:0 -map 1:0 -map 0:1 -c copy -disposition:a:0 default all.mkv` works! But now I have 2 "default" tracks, one copied from 1.mkv, it was already default there, and another one from 2.ac3. Is there some `-disposition:a:1 notdefault` option ?
[16:31:02 CET] <waressearcher2> I'm trying to make 14 second video to 15 second using "setpts=1.1*PTS" the video becomes longer but it becomes jerky, its just barely noticeable to it is jerky, there is no way to make it smooth as original ? its because of how that filter works ?
[16:32:54 CET] <debianuser> waressearcher2: try to also increase framerate (fps) with -r option to the equivalent value
[16:33:11 CET] <debianuser> (or in your case you'll actually decrease it a little)
[16:33:38 CET] <waressearcher2> ffprobe says that both original and slowed down videos are 29.97
[16:33:49 CET] <waressearcher2> so I want to keep it 29.97
[16:54:11 CET] <debianuser> waressearcher2: 29.97 fps * 14 seconds means 419.58 frames. To have the same number of frames played during 15 seconds you need 419.58/15 = 27.972 fps, so try -r 27.972 :)
[17:19:04 CET] <waressearcher2> debianuser: but I don't set "-r" any options so why ffmpeg don't do it itself if its a "right way" ?
[17:19:38 CET] <waressearcher2> it should recognize that I changed "PTS" and it should set different rate
[17:27:37 CET] <debianuser> waressearcher2: Maybe it works the other way around, e.g. you can set -r and it would adjust PTS, but I don't think it's always possible to adjust rate from PTS, you could set variable PTS, or you could be overlaying multiple videos with different PTSes...
[20:21:43 CET] <waressearcher2> ist jemand hier ?
[20:26:15 CET] <klaxa> waressearcher2: yes?
[20:26:30 CET] <waressearcher2> du bist ja so ein süßer
[20:26:39 CET] <klaxa> um...
[20:26:55 CET] <waressearcher2> kerl
[20:27:08 CET] <klaxa> while i do speak german, maybe keep it english and ffmpeg related?
[20:28:08 CET] <waressearcher2> ist das die Kanal uber ffmpeg ? wirklich ?
[20:30:40 CET] <durandal_1707> use mencoder
[20:34:57 CET] <StephenS> what output will ffmpeg give if video is longer than 24 hours? for Duration tag?
[20:35:54 CET] <c_14> StephenS: I think the hours just keep increasing. Not sure though.
[20:36:27 CET] <StephenS> if thats the case, that would be great
[20:36:43 CET] <StephenS> waiting for official answer though, if someone experienced this :)
[20:36:50 CET] <ChocolateArmpits> Well it doesn't work like a timecode clock does so I guess it should increase past that
[20:45:16 CET] <ChocolateArmpits> StephenS: as per my guess, the duration reported exceeds 24 hours
[20:45:19 CET] <ChocolateArmpits> Duration: 25:00:00.00
[20:45:27 CET] <klaxa> you can just try it out, can't you?
[20:45:29 CET] <ChocolateArmpits> created a 25 hour wav file to test
[20:45:48 CET] <ChocolateArmpits> at 8000Hz :)
[20:46:35 CET] <StephenS> cool :D
[20:46:38 CET] <StephenS> thanks for reply :)
[20:46:54 CET] <furq> now do 100 hours
[21:09:52 CET] <StephenS> lol it should just be hour+n
[21:15:11 CET] <zenny> Can anyone tell me how to extract the alpha background based on the luminance? Something is discussed here: https://ffmpeg.org/pipermail/ffmpeg-user/2011-July/001793.html, but could not achieve. Any inputs?
[21:19:11 CET] <zenny> I am trying to extract a sample video from youtube as seen in this screenshot: http://picpaste.com/qJ4WNPHY.jpg. I am trying to extract the brightest portion (naturally the white background) from the picture and replace that with a transparent layer.
[21:19:32 CET] <Wader8> hello
[21:19:57 CET] <waressearcher2> Wader8: hallo mein freind
[21:20:18 CET] <waressearcher2> Wader8: wie gehts ?
[21:20:30 CET] <Wader8> i am wondering if it's possible to add audio from one video file to another video file with it's own audio kept intact, with one ffmpeg command preferrably to save on time
[21:20:56 CET] <Wader8> i have 2 video files, one has better video quality and resolution, the other one with lower video has better audio
[21:21:31 CET] <Wader8> but there is a slight time mismatch between the videos, so i want to retain the original audio, and add the better audio as a second audio stream
[21:21:32 CET] <waressearcher2> Wader8: hast do mich nicht gehoert ?
[21:21:51 CET] <c_14> ffmpeg -i better_video -i better_audio -map 0 -map 1:a -c copy out.mkv
[21:22:18 CET] <Wader8> oha and, i want to remux only yes, no recoding
[21:23:42 CET] <Wader8> c_14 will that command keep both audios, i still want the original audio of the better video file to be the primary first stream
[21:24:46 CET] <Wader8> i will try anyway
[21:25:03 CET] <c_14> Yes, it will.
[21:25:20 CET] <waressearcher2> c_14: hallo
[21:32:14 CET] <zenny> I am trying to extract the brightest portion (naturally the white background) from the picture ( http://picpaste.com/qJ4WNPHY.jpg) and replace that with a transparent layer, based on luminance value. I read here (https://ffmpeg.org/pipermail/ffmpeg-user/2011-July/001793.html) that geq filter can be of help to the yuv formats, but could not figure out how? Any inputs appreciated!
[21:33:44 CET] <Wader8> c_14 looks like it worked, but matroska reports streams not having global headers, plus, there seems to be like 5 V_JPEG streams that seem like thumbnails from mediainfo that are not seen in any original files, but they pop up in the output
[21:34:50 CET] <c_14> No clue about the thumbnails, but the global headers thing is a bug. Just ignore it.
[21:37:42 CET] <Wader8> well they are V_MJPEG streams with some misc info such as filename "cover.jpg" "small_cover_land.jpg" and by the way, the better audio is 6CH 348kbps audio is in AC3 while the unknown bitrate basic 2CH audio is AAC, but i know this is not ffmpeg, it's mediainfo, and once I did the output, the new file lost some AC3 metadata, it doesn't show AC3 bitrate anymore, not a big deal really
[21:38:11 CET] <virt> hmm videos created with ffmpeg -i video -i audio -c:v copy -c:a aac have serious desyncs in mplayer but videos encoded with ffmpeg -i video -i audio -c:v h264 -c:a aac work perfectly even tho the video file is already h264 o.O
[21:38:23 CET] <virt> (note: these are huge 10GB 4K videos
[21:38:24 CET] <virt> )
[21:41:55 CET] <Hammerhead> Hi guys, have a muxing problem. taking a concat file of jpegs and a http stream of audio muxing into a mpegts to a encoder. Problem is the audio drops out after a minute or 2 and then the video stop as well. I know this because I am using the -vcodec copy command and can see the drop in the local copy. Anyone else seen this or have an idea why this might me happening?
[21:42:37 CET] <Hammerhead> any help would BE GREATLY appreciated.
[21:43:11 CET] <Hammerhead> ffmpeg version 2.4.3-1ubuntu1~trusty6
[21:43:43 CET] <furq> Wader8: https://github.com/mbunkus/mkvtoolnix/wiki/Meta-data-fields-like-bitrate-lost
[21:43:45 CET] <Hammerhead> wow that does seem to be an old version huh?
[21:46:12 CET] <Wader8> oh thanks a lot furq, but i think i updated mediainfo a while ago, i will report this to mediainfo later if I see this case not being covered
[21:47:24 CET] <furq> there's at least one open bug on mediainfo for this, but i wouldn't expect it to be fixed
[21:48:24 CET] <Wader8> why, the priority or volume of other tasks ?
[21:48:56 CET] <Wader8> shouldn't be that hard to fix but seems like a big deal that one entry would disappear
[21:49:12 CET] <furq> http://sourceforge.net/p/mediainfo/feature-requests/154/#57d7/e354
[21:49:14 CET] <Wader8> plus that goes with proper support for a format,
[21:49:17 CET] <furq> that links to a post from 2007
[21:49:50 CET] <furq> some other containers store stream bitrates in the header, but mkv doesn't
[21:50:06 CET] <Wader8> weird why it is a complex issue to implement
[21:50:44 CET] <furq> afaik if there's more than one vbr stream then you'd need to read every vbr stream in its entirety to get an accurate bitrate
[21:50:49 CET] <furq> which is probably out of scope for mediainfo
[21:52:33 CET] <Wader8> well i have AAC and AC3 now combined with 1 video stream and AC3 bitrate doesn't show anymore
[21:52:42 CET] <Wader8> but AC3 bitrate does show in the original video
[21:53:22 CET] <Wader8> AAC however doesn't show in the original file either, but that's probably some stuff at encoding
[21:53:24 CET] <furq> right, the aac and video streams are vbr
[21:53:40 CET] <Wader8> AC3 is constant indeed
[21:54:00 CET] <furq> if you only have one vbr stream in the file then it's possible to calculate the bitrates
[21:54:17 CET] <Wader8> i saw that now yes
[21:54:47 CET] <Wader8> seems like it's not an isolated issue, i'm not the only one noticing this, but i noticed and wondered it for month and now I had enough
[21:55:12 CET] <Wader8> it's not a big problem it's just I never went looking about it what it was
[21:56:21 CET] <furq> if it's a problem then you can use mp4
[21:59:33 CET] <Wader8> well i celebrated too early
[22:00:25 CET] <Wader8> i tried playback now and mpc-hc freezes while VLC keeps playing but  silent when i select the second track, i will try MP4 - just please god don't tell me mp4 can't have AC3
[22:00:54 CET] <DragonsLordKERNE> hello to everybody!
[22:01:28 CET] <DragonsLordKERNE> does someone of you know if libxvid is better than mpeg4, as encoders
[22:01:52 CET] <DragonsLordKERNE> or if there are no pros choosing the first, instead of the second?
[22:06:15 CET] <Wader8> c_14 unfortunately MPC-HC nor VLC can use the second track, they bug out, im trying to use MP4 format but now I got the problem with those V_MJPEG streams, i get an error about "encoder not found for stream #0:2" which is "subrip (srt)" when i use -vcodec and -acodec copy, but when i use -c copy i get error "could not find tag for codec subrip in stream #2, codec not currently supported in container"
[22:06:38 CET] <c_14> Yeah, you can't put subrip in mp4
[22:06:59 CET] <Wader8> well i would need to adjust mappings to get rid of those streams right
[22:07:02 CET] <DragonsLordKERNE> c_14, may you help me?
[22:07:13 CET] <furq> Wader8: i doubt the container is the issue, but you can use -c:s mov_text
[22:08:05 CET] <c_14> DragonsLordKERNE: 99% of the time, don't use either. Just use libx264
[22:08:25 CET] <c_14> Which is better, I don't know. I don't tend to test outdated codecs and encoders.
[22:08:35 CET] <DragonsLordKERNE> good to know
[22:08:43 CET] <DragonsLordKERNE> but now I have no choice
[22:08:54 CET] <DragonsLordKERNE> I have to keep xvid
[22:09:02 CET] <DragonsLordKERNE> in order to preserve compatibility
[22:09:03 CET] <DragonsLordKERNE> :(
[22:09:15 CET] <DragonsLordKERNE> the only issue I experienced
[22:09:36 CET] <DragonsLordKERNE> is the fact that am not able to set fixed GOP size
[22:09:52 CET] <DragonsLordKERNE> I mean one Key-frame each n frame
[22:09:59 CET] <DragonsLordKERNE> tipically
[22:10:05 CET] <DragonsLordKERNE> 10 * FPS
[22:10:12 CET] <DragonsLordKERNE> in my case PAL so 250
[22:10:29 CET] <DragonsLordKERNE> I tried with the option suggested by you, in the past
[22:10:36 CET] <DragonsLordKERNE> I mean -g
[22:10:49 CET] <DragonsLordKERNE> but it seems to be only the maximum gop size
[22:11:00 CET] <DragonsLordKERNE> should I set option -r
[22:11:16 CET] <DragonsLordKERNE> in order to make libxvid working?
[22:12:09 CET] <c_14> hmm?
[22:12:11 CET] <c_14> What's the problem?
[22:13:14 CET] <Wader8> well c_14, it did mux and the console log output a bunch of lines now, like 50 each, first one warning about non-monotonus DTS in audio stream 0:1 , changing to 3745 from 3744 (curr and prev) .. this may result in incorrect timestamps in output,""  then there's some lines about PTS has no value, then there's red colored and says Application provided duration -5  / timestamp is out of range for mp4/mov format
[22:13:39 CET] <DragonsLordKERNE> c_14, here is what avidemux do
[22:13:59 CET] <DragonsLordKERNE> (it's an output of the resulting file with avinaptic)
[22:14:00 CET] <DragonsLordKERNE> Key frames                  569 (0; 250; 500; 750; 1000; ... 66985)
[22:14:00 CET] <DragonsLordKERNE> Null frames                 0
[22:14:00 CET] <DragonsLordKERNE> Min key int                 1
[22:14:01 CET] <DragonsLordKERNE> Max key int                 250
[22:14:09 CET] <DragonsLordKERNE> so min 1
[22:14:12 CET] <DragonsLordKERNE> max 250
[22:14:22 CET] <DragonsLordKERNE> and the keyframe sequens is according to this
[22:14:28 CET] <DragonsLordKERNE> with 0 null frames
[22:15:15 CET] <DragonsLordKERNE> here is the avinaptic output of a file obtained with ffmpeg
[22:15:43 CET] <DragonsLordKERNE> Key frames                  556 (0; 195; 292; 335; 415; ... 31660)
[22:15:43 CET] <DragonsLordKERNE> Null frames                 1 (1)
[22:15:43 CET] <DragonsLordKERNE> Min key int                 1
[22:15:43 CET] <DragonsLordKERNE> Max key int                 250
[22:15:43 CET] <DragonsLordKERNE> Avg key int                 57.068345
[22:15:43 CET] <DragonsLordKERNE> Delay                       0 ms
[22:15:47 CET] <Wader8> well c_14 it does work actually, playback test in MPC-HC, with the expected audio desync with the added track
[22:15:54 CET] <DragonsLordKERNE> the setting seems to be tha same
[22:15:54 CET] <Wader8> so thanks a lot!
[22:15:55 CET] <DragonsLordKERNE> min 1
[22:15:57 CET] <DragonsLordKERNE> max 250
[22:16:12 CET] <furq> c_14: are you sure you told the right person about pastebins
[22:16:23 CET] <DragonsLordKERNE> but the key frame sequence is not costant or fixed
[22:16:46 CET] <ChocolateArmpits> DragonsLordKERNE: is that libxvid output ?
[22:16:59 CET] <c_14> furq: I wanted to see his console output. DragonsLordKERNE pasting more than 3 lines at a time is a different issue.
[22:17:54 CET] <Wader8> that desync i can use LAV or MPC settings and delays, or maybe tune with ffmpeg, but it works! ... MPC-HC seems to select the track 2 automatically, looks like under MP4 the AAC audio stream now properly shows bitrate and it's measly 156kbps
[22:18:46 CET] <DragonsLordKERNE> ChocolateArmpits, hi!
[22:18:47 CET] <DragonsLordKERNE> it is Bitstream type: MPEG-4 Part 2 - User data: XviD0065
[22:19:07 CET] <ChocolateArmpits> DragonsLordKERNE: well ok but where did you obtain it from ?
[22:19:09 CET] <c_14> DragonsLordKERNE: does it matter? Also, I'm not sure of a way to convince libxvid to output constant keyframe intervals.
[22:19:32 CET] <DragonsLordKERNE> c_14, yes it it
[22:19:41 CET] <DragonsLordKERNE> if avidemux is able to do this
[22:19:57 CET] <DragonsLordKERNE> I am quite sure that ffmpeg should also be able to do the same
[22:20:01 CET] <DragonsLordKERNE> with the same encoder
[22:23:06 CET] <Wader8> c_14 here, well, we may still finely tune ffmpeg options so maybe we can mitigate the AC3 desync and avoid manual delay options http://pastebin.com/61sj7FmM
[22:25:33 CET] <Wader8> i just hope the desync is not variable
[22:26:02 CET] <Wader8> there shouldn't be any reason for 2 versions to be messed with playback speed
[22:26:26 CET] <Wader8> it's probably beginning and the end are differently cut
[22:26:39 CET] <Wader8> i trying to watch em side by side to see diff
[22:32:14 CET] <Wader8> ..damn one video seems like stretched or shrinked, it amounts to 4 seconds in 40 minutes
[22:33:07 CET] <Wader8> anyway, i'll fiddle with it later or tomorrow i'll still keep this open if you want to add anything c_14
[22:33:22 CET] <Wader8> i have to take a break i'll read it later
[22:33:28 CET] <c_14> Wader8: you can probably use atempo or asetpts
[22:33:49 CET] <c_14> DragonsLordKERNE: I can't find any options for it.
[22:34:23 CET] <DragonsLordKERNE> the encoder documentation
[22:34:28 CET] <DragonsLordKERNE> at least from what you know
[22:34:40 CET] <DragonsLordKERNE> is related to libxvid only with ffmpeg
[22:34:51 CET] <DragonsLordKERNE> or it's the encoder documentation, in general?
[22:35:00 CET] <c_14> Just the ffmpeg libxvid docs
[22:35:33 CET] <DragonsLordKERNE> so that parameters
[22:35:48 CET] <DragonsLordKERNE> does not have any sense with other software
[22:35:50 CET] <DragonsLordKERNE> right?
[22:35:56 CET] <c_14> hmm?
[22:36:27 CET] <DragonsLordKERNE> parameters within the encoders doc
[22:36:33 CET] <DragonsLordKERNE> are for mmpeg only
[22:36:37 CET] <DragonsLordKERNE> *ffmpeg
[22:36:42 CET] <c_14> yes
[22:36:49 CET] <DragonsLordKERNE> or they are probably the same also with avidemux
[22:36:52 CET] <DragonsLordKERNE> and other soft...
[22:36:53 CET] <DragonsLordKERNE> ok
[22:36:58 CET] <DragonsLordKERNE> thank you very much
[22:40:36 CET] <DragonsLordKERNE> c_14, may you please show me the documentation for mpeg-4?
[22:41:29 CET] <c_14> ffmpeg -h encoder=mpeg4
[22:42:43 CET] <DragonsLordKERNE> thank you
[22:42:50 CET] <DragonsLordKERNE> isn't there a web page for this
[22:42:58 CET] <DragonsLordKERNE> like for libxvid?
[22:43:22 CET] <c_14> Apparently not, no.
[22:43:33 CET] <c_14> Well, besides this https://trac.ffmpeg.org/wiki/Encode/MPEG-4
[22:45:54 CET] <DragonsLordKERNE> thank you, once again
[22:47:02 CET] <DragonsLordKERNE> I guess it should have the same option of libxvid
[22:56:27 CET] <zenny> Asking the question again: I am trying to extract (experimenting) the brightest portion (naturally the white background) from the picture ( http://picpaste.com/qJ4WNPHY.jpg) and replace that with a transparent layer, based on luminance value. I read here (https://ffmpeg.org/pipermail/ffmpeg-user/2011-July/001793.html) that geq filter can be of help to the yuv formats, but could not figure out how? Any inputs appreciated!
[22:58:03 CET] <sneke> I'm trying to convert an .mkv and an .mp4 to .webms, but when i "ffmpeg -i intput.mp4 output.webm" (or the .mkv source) I get "Codec AVOption quality () specified for input file #0" both tries.
[23:00:08 CET] <sneke> https://ptpb.pw/1jJB
[23:00:48 CET] <llogan> never seen that message before
[23:02:26 CET] <llogan> does it occur with a specific file, or any?
[23:02:46 CET] <llogan> does it occur with a build from current git master?
[23:02:58 CET] <llogan> (you can possibly test http://johnvansickle.com/ffmpeg/ )
[23:03:40 CET] <sneke> So far it occurs with all files
[23:03:41 CET] <JEEB> sneke: are you sure you're not hiding options there? it seems like input is getting an option that is meant for output
[23:03:44 CET] <sneke> I'll try the git master
[23:03:51 CET] <JEEB> that, or the compile is majorly fucked
[23:03:56 CET] <sneke> oh
[23:04:09 CET] <sneke> ffmpeg was aliased to "ffmpeg -quality best"
[23:04:13 CET] <JEEB> hah
[23:04:26 CET] <JEEB> and thus "-quality best" was given as an input option
[23:04:36 CET] <JEEB> as in, before the -i
[23:04:49 CET] <sneke> oh
[23:04:52 CET] <JEEB> thus the "Codec AVOption quality () specified for input file #0 (peekaboo.mp4) is not a decoding option." error makes sense :P
[23:04:58 CET] <sneke> that explains it
[23:05:03 CET] <JEEB> yup
[23:05:08 CET] <sneke> Thanks, JEEB
[23:10:16 CET] <gachowy> hello. i'm heaving problems with restream 720p h264 RTPS stream, to mp4. it throws some error with 'non monotone timestamps' and 'av_interleaved_write_frame()'. CVLC from webcam, to server via RTSP, and i need to 'restream' it to mp4 loadable into www. any ideas, what was wrong? :> http://pastebin.com/6XeSEMYB
[23:11:07 CET] <llogan> you omitted all version and build info. although it may not matter. not many ffserver answers here (I've never used it).
[23:12:02 CET] <gachowy> ffmpeg 0.6.5, built on centos machine :)
[23:12:09 CET] <llogan> ancient and unsupported
[23:12:42 CET] <gachowy> hmm... so propably need to compile it from sources? :)
[23:13:27 CET] <llogan> probably. i'm not aware of any recent ffserver builds available for linux
[23:14:15 CET] <gachowy> gash, last time, when i'm compiling ffmpeg, it takes 3-5h :D
[23:15:07 CET] <llogan> that's long. i just compiled. took about 10 mins I would guess
[23:15:18 CET] <llogan> or less
[23:15:52 CET] <zenny> gachowy: you can use static builds instead
[23:16:07 CET] <gachowy> oke, maybe cause i compiled it on raspberry pi :]
[23:16:18 CET] <gachowy> some repos for centos? :>
[23:17:03 CET] <llogan> actually i think ffserver is available from http://johnvansickle.com/ffmpeg/
[23:17:22 CET] <llogan> packaged with ffmpeg, ffprobe, etc
[23:17:42 CET] <llogan> git the git version, not release
[23:18:44 CET] <drv> that's not going to work on a rpi... :)
[23:18:45 CET] <gachowy> hmm... 0.10.15 version was close to latest? propably i have it on rpmfusion :)
[23:19:33 CET] <llogan> that is also ancient
[23:23:31 CET] <debianuser> llogan: yes, there's some ffserver binary in http://johnvansickle.com/ffmpeg/ archive. No ffplay, however.
[23:23:43 CET] <debianuser> c_14: Thanks! `ffmpeg -i 1.mkv -i 2.ac3 -map 0:0 -map 1:0 -map 0:1 -c copy -disposition:a:0 default all.mkv` works! But now I have 2 "default" tracks, one copied from 1.mkv, it was already default there, and another one from 2.ac3. Is there some `-disposition:a:1 notdefault` option ?
[23:26:55 CET] <gachowy> oke, next errors, but will try to find solution for them. thanks :)
[23:31:49 CET] <c_14> debianuser: looks like disposition original does it
[23:32:51 CET] <c_14> Though, eh
[23:32:53 CET] <c_14> I don't know why
[23:41:40 CET] <c_14> I think there should be a way to clear the disposition though...
[23:41:47 CET] <c_14> Might want to make a feature-request
[23:56:28 CET] <Wader8> c_14 would atempo require recoding ?
[23:56:38 CET] <c_14> yes, all filters will require reencoding
[23:57:02 CET] <Wader8> okay
[23:57:39 CET] <llogan> you could filter upon playback
[23:58:03 CET] <llogan> ffplay -af atempo=0.8 input.oga
[00:00:00 CET] --- Tue Nov  3 2015