[Ffmpeg-devel-irc] ffmpeg.log.20150915

burek burek021 at gmail.com
Wed Sep 16 02:05:01 CEST 2015


[01:31:25 CEST] <tommd> Decoupling Provenance
[01:31:31 CEST] <tommd> oops
[01:47:10 CEST] <t4nk398> I need to install libfaac with ffmpeg but getting following message:
[01:47:25 CEST] <t4nk398> root at pooh [/usr/local/src/ffmpeg-0.7.11]# ./configure --enable-libfaac libfaac is nonfree and --enable-nonfree is not specified.  If you think configure made a mistake, make sure you are using the latest version from Git.  If the latest version fails, report the problem to the ffmpeg-user at ffmpeg.org mailing list or IRC #ffmpeg on irc.freenode.net. Include the log file "config.log" produced by configure as this will help solving the 
[01:47:44 CEST] <t4nk398> Do you have any idea how to enable libfaac with ffmpeg
[01:49:29 CEST] <DHE> just add --enable-nonfree to your configure command (and understand what it means)
[01:51:34 CEST] <t4nk398> ok but how to verify it is correctly installed
[03:18:21 CEST] <ritmo2k> Hey guys, I have a directory of images, 2991 which is a 600 second duration, so ~5fps, however `ffmpeg -r 5 -i %05d-capture.jpg out.mp4` leaves me with a video of the expected duration but the frame duration is stretched badly? Many are skipped, how do I troubleshoot this? New to ffmpeg.
[03:50:05 CEST] <relaxed> ritmo2k: you want -framerate 5
[04:05:01 CEST] <ritmo2k> relaxed: Hey, that did not seem to produce a video any different than if I place "-r 5" before the "-i". If I look at the pictures they show a sequence for example of a person walking and its fairly detailed however each time I encode it I miss most of it and just see large jumps?
[05:18:14 CEST] <kyleogrg> I am a bit confused.  How can I scale this watermark overlay to a certain size?  http://pastebin.com/ZJjBuaY5
[05:18:49 CEST] <kyleogrg> specifically, i would like to scale it to the dimensions of the first input (input.mp4)
[06:08:22 CEST] <relaxed> kyleogrg: scale the watermark first in the filter chain, then use its output label in the overlay filter
[06:10:19 CEST] <relaxed> oh, I mean scale the video first :)
[06:10:51 CEST] Action: relaxed needs more coffee :/
[06:23:39 CEST] <Chagall> can't seem to get ffmpeg to encode mp3 or flac audio: "Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument" even though it finds the right codec to encode it
[06:26:19 CEST] <relaxed> Chagall: command?
[06:28:12 CEST] <Chagall> http://pastebin.com/hMh5ReHp
[06:29:17 CEST] <Chagall> or http://pastebin.com/igzxfh1N easier to read
[06:29:35 CEST] <kyleogrg> relaxed: thanks.  i'll come back here if i have any troub;e
[06:29:37 CEST] <kyleogrg> trouble
[06:32:41 CEST] <kyleogrg> I'm trying to overlay a solid color box over a video: ffmpeg.exe -y -i test.mp4 -i color=c=white:s=640x480 ......
[06:33:08 CEST] <kyleogrg> but I get: "color=c=white:s=640x480: No such file or directory"
[06:33:29 CEST] <kyleogrg> so - what am i doing wrong
[06:34:30 CEST] <kyleogrg> i actually want to overlay a black screen with white text between X and Y seconds.
[06:34:52 CEST] <kyleogrg> so how can i generate this screen with ffmpeg
[06:37:59 CEST] <Chagall> "Only VP8 or VP9 video and Vorbis or Opus audio and WebVTT subtitles are supported for WebM." o lol nevermind
[06:38:06 CEST] <relaxed> yes
[06:38:09 CEST] <Chagall> not sure why i thought i was able to do it before
[06:38:15 CEST] <relaxed> well, webm
[06:43:26 CEST] <relaxed> kyleogrg: you need -f lavfi before the color input
[06:45:06 CEST] <kyleogrg> relaxed: okay, thanks
[07:33:51 CEST] <Chagall> http://pastebin.com/kPQzEbve if someone can find something wrong with this string please let me know... pops up every time now, already tried without scaling, must have screwed up somewhere but can't see it
[07:35:58 CEST] <relaxed> Chagall: pastebin the complete command and console output
[07:37:04 CEST] <Chagall> http://pastebin.com/NbUruux8
[07:40:08 CEST] <frenda> Is it possible to concatenate two videos in different sizes, for example: a 1920x1080 video + a 1024x768 video?
[07:41:03 CEST] <frenda> I tried 'ffmpeg -f concat -i merge.txt -c copy output.mp4', but
[07:41:22 CEST] <frenda> but it destroyed one of the videos!
[07:41:45 CEST] <relaxed> Chagall: what are you trying to do? Loop the cover art?
[07:43:57 CEST] <relaxed> frenda: frame size has to be the same, among other things
[07:48:54 CEST] <frenda> Ah, any idea on 'how to make the same: i.e. enlarging that 1024x768 video by adding blank space around it'?
[07:49:17 CEST] <pzich> use pad https://ffmpeg.org/ffmpeg-filters.html#pad-1
[07:50:25 CEST] <Chagall> just testing encode that's all
[07:50:28 CEST] <Chagall> it doesn't matter it's just a stream
[07:50:30 CEST] <Chagall> worked fine before
[07:51:36 CEST] <Chagall> iirc it broke when i added -ss, -t and -vf to my argument builder
[07:52:43 CEST] <Chagall> but even when i don't use them now it screws up so i don't know
[07:54:02 CEST] <relaxed> Chagall: "Value 8.000000 for parameter 'tile-columns' out of range [-1 - 6]\r\n[libvpx-vp9 encoder @ 00000000004bc9a0]
[07:54:33 CEST] <Chagall> yeah just worked after i disabled it. thanks
[08:02:13 CEST] <jasom> Any suggestions for providing a test-case when my input file is 7GB?
[08:02:56 CEST] <pzich> make a short clip with -t ?
[08:03:11 CEST] <jasom> pzich: I can't make a short clip, since it aborts
[08:03:38 CEST] <jasom> the dts as shown with -debug_ts is negative, which ffmpeg really doesn't like
[08:06:45 CEST] <frenda> pzich: `ffmpeg -i input.mp4 pad=1920:1080:0:0 output1.mp4`  --> ERROR
[08:06:46 CEST] <frenda> Error: [NULL @ 0x8c90380] Unable to find a suitable output format for 'pad=1920:1080:0:0' pad=1920:1080:0:0: Invalid argument
[08:06:58 CEST] <jasom> From the demuxer, I get a pkt_dts_time:1409.46 followed by a pkt_dts_time:1.28163
[08:10:04 CEST] <jasom> meh, I'll just checkout from git and debug it myself
[08:10:55 CEST] <pzich> frenda: you need '-vf pad=...'
[08:13:55 CEST] <frenda> Ah, tnx; +1
[08:31:59 CEST] <satiender> Hi , Can we compile ffplay outside from ffmpeg directory
[08:32:10 CEST] <satiender> please help how I can do that
[08:33:10 CEST] <durandal_170> no, why?
[08:35:55 CEST] <satiender> durandal_170 : I think if we linking the libraries like -lavcodec , -lavfilter etc . we can do that
[08:35:55 CEST] <satiender> at run time
[08:35:56 CEST] <satiender> gcc -o ffplay ffplay.c -lavcodec lavfilter -lm -lSDL
[08:37:06 CEST] <satiender> if I do that then I will able to divide ffmpeg in parts which is very useful for me for cross platform
[08:37:26 CEST] <satiender> like ARM which having very less memory as compare to x86
[08:41:04 CEST] <durandal_170> that will not reduce memory usage...
[08:45:21 CEST] <satiender> but if I do that then I only install those binary which i want to you
[08:45:54 CEST] <satiender> please help I got some undefined errors like undefined av_regster_all
[08:46:05 CEST] <satiender> how I can solve these
[08:46:34 CEST] <durandal_170> link with libavutil
[08:48:12 CEST] <satiender> ok
[08:48:27 CEST] <satiender>  you mean my command is like :
[08:49:15 CEST] <satiender> gcc -o ffplay ffplay.c  -lavcodec -lavfilter -lavutil  -lavformat -lm
[08:50:46 CEST] <satiender> but then I got more undefined errors
[08:51:26 CEST] <durandal_170> you need bunch more....
[08:55:34 CEST] <satiender> ok
[08:55:44 CEST] <satiender> can you give example for that
[09:06:14 CEST] <durandal_170> no, I'm busy
[09:10:05 CEST] <satiender> durandal_170: when you free you can , please
[10:02:18 CEST] <BtbN> How did you get the ffmpeg cli to show you the options for a specific codec/filter again?
[10:02:49 CEST] <chungy> "ffmpeg -h encoder=<codec>", "ffmpeg -h filter=<filter>"
[10:03:13 CEST] <chungy> all the -h options are in the manpage
[10:14:15 CEST] <jasom> okay, I'm definitely seeing the PTS reset to zero in the middle of a VOB, and ffmpeg really doesn't like that.  mplayer will *play* it, but seeking gets broken after the pts reset.  Is there anything I can do to encode this with ffmpeg?
[12:29:26 CEST] <oreWizard> someone know if url_fopen inst avaliable?
[12:29:51 CEST] <BtbN> what?
[12:30:22 CEST] <BtbN> jasom, if the framerate is static, you can just make up new pts with the setpts filter.
[12:42:48 CEST] <oreWizard> can u help me to use avio_open
[12:59:12 CEST] <oreWizard> what happend to 	__android_log_write(ANDROID_LOG_WARN, "trace", "Invalid setting parameters");
[12:59:18 CEST] <oreWizard> sory
[12:59:20 CEST] <oreWizard> av_set_parameters
[12:59:23 CEST] <oreWizard> someone?
[13:05:14 CEST] <netek> afternoon all, I am attempting to transcode an rtmp stream into hls and ffmpeg command I use can be located here http://pastebin.com/sSsZjFR6
[13:06:00 CEST] <netek> is there anyway I could run this command as a background process?  I have tried with  & at the end of the command but transcoding doesnt seem to run and I also attempted to use screen and this also does not transcode!
[13:06:17 CEST] <netek> I am running this on Ubuntu Server
[13:15:30 CEST] <c_14> It should work in screen, since that (usually) spawns a regular shell. Not sure about the &, might be a regression might have always been that way. nohup ffmpeg & works (I just tested that).
[13:17:01 CEST] <c_14> s/shell/tty/
[13:17:51 CEST] <netek> c_14 when running on screen it seems to only transcode audio, no video -_-
[13:18:30 CEST] <c_14> But when you run it in not-screen it also transcodes video?
[13:18:49 CEST] <netek> no only audio,
[13:19:01 CEST] <netek> sorry
[13:19:11 CEST] <netek> not-screen transcodes video
[13:19:28 CEST] <netek> I am watching it now on my mobile device
[13:19:44 CEST] <netek> but when I attempt it with the & at the end or in screen I cannot see video but can hear audio
[13:22:21 CEST] <netek> c_14 ok with further tests there seems to be an intermittent issue.  I stopped the transcode but had to restart 4 times to get the transcode to work
[13:40:59 CEST] <oreWizard> c_14 i need the compiling with --enable-protocol=file
[13:41:12 CEST] <oreWizard> do u know how its work?
[13:41:29 CEST] <c_14> You just add that to the configure line?
[13:43:07 CEST] <oreWizard> i am getting error on it
[13:44:24 CEST] <bove> Is there a way to encode from an image sequence while it's beeing written by an external program, and have ffmpeg wait for new frames?
[13:44:36 CEST] <c_14> bove: img2pipe
[14:07:36 CEST] <bove> sequence/MVI.%06d.yuv: No such file or directory
[14:08:11 CEST] <bove> -i sequence/MVI.000000.yuv works fine, but only sees the first frame
[14:17:04 CEST] <bove> c_14: It seems image2pipe is meant to read from pipes?
[14:17:42 CEST] <klaxa> pretty much
[14:17:46 CEST] <bove> I can't do pipes from the output program, so I need to read one image file at the time
[14:18:05 CEST] <klaxa> you can do something like: cat *.jpg | ffmpeg -f image2pipe -i - [whatever]
[14:18:10 CEST] <klaxa> hmm
[14:18:13 CEST] <klaxa> that's a problem
[14:19:16 CEST] <bove> Right now I'm currently working with a buffer equal to the whole length of the video
[14:19:35 CEST] <bove> Looking for some way to avoid that
[14:20:42 CEST] <klaxa> can you control the other program?
[14:20:59 CEST] <klaxa> and maybe just make it write all the images to stdout?
[14:21:05 CEST] <bove> Maybe I should write a small program that does the actual file reading, feeds it to ffmpeg and then deletes it afterwards
[14:21:50 CEST] <bove> klaxa: Can not do any kind of pipe with the input program. Even tried named pipes, but the program deletes them and creates regular files
[14:22:21 CEST] <klaxa> welp, sounds like it will require some hackery around that if you can't change the program
[14:41:21 CEST] <tactualtaco> Anyone here have experience concatenating .ts files?
[14:42:41 CEST] <waressearcher2> is anyone here ?
[14:45:23 CEST] <waressearcher2> I have concatenation problems
[14:46:34 CEST] <tactualtaco> You too? haha
[14:46:52 CEST] <tactualtaco> Came here for the exact same thing.
[14:49:24 CEST] <klaxa> well what have you tried?
[14:50:33 CEST] <waressearcher2> in my case weird thing happens
[14:52:57 CEST] <klaxa> i'm... waiting?
[14:54:20 CEST] <waressearcher2> I will write it in a moment
[14:56:53 CEST] <BtbN> tactualtaco, cat 1.ts 2.ts > 3.ts
[14:57:51 CEST] <tactualtaco> BtbN, I can output a file, but I get dropped frames, "jitter" according to MPC-HC and the framerate is all over the place
[14:58:11 CEST] <BtbN> hm?
[14:58:27 CEST] <BtbN> If you want to concat two .ts files, just append them to each other. ts works that way.
[14:58:35 CEST] <tactualtaco> Here is my command: ffmpeg -f concat -i files.txt -c copy output.ts
[14:58:45 CEST] <BtbN> No need to use ffmpeg for that.
[14:59:03 CEST] <tactualtaco> There are 1653 files
[14:59:10 CEST] <BtbN> cat *.ts > complete.ts
[14:59:40 CEST] <tactualtaco> Im on Windows so will type suffice?
[14:59:50 CEST] <BtbN> sure, should be the same syntax
[15:00:14 CEST] <BtbN> No idea how windwos orders stuff when using globs
[15:00:38 CEST] <tactualtaco> The files are in alphabetcial order, so it seems like its working
[15:01:33 CEST] <waressearcher2> what is *.ts format ?
[15:01:34 CEST] <tactualtaco> Okay that kinda worked, the output file lasts for 01:42, while it should be around 27:00
[15:02:03 CEST] <BtbN> you can still remux them with ffmpeg once you're done, to fixup timestamps.
[15:02:35 CEST] <tactualtaco> And how would I do that? I'm very new to ffmpeg
[15:02:57 CEST] <tactualtaco> Now everything switched up and the file doubled in length.
[15:04:32 CEST] <tactualtaco> Alright I'll just google it, haha.
[15:05:21 CEST] <BtbN> The length is only an estimate
[15:07:03 CEST] <waressearcher2> I created two video files 10 seconds each, using this: "ffmpeg -i original.avi -c:v mpeg4 -vb 10000k -t 10 -f avi out_1.avi" and second file slowed down: "ffmpeg -i original.avi -c:v mpeg4 -vb 10000k -vf "setpts=3*PTS" -t 10 -an -f avi out_2_slowed.avi" then to concatenate them I do this, I created file "list.txt" with content: "file 'out_1.avi'"
[15:07:08 CEST] <waressearcher2>  and "file 'out_2_slowed.avi'"   and then I run that command to concatenate: "ffmpeg -f concat -i list.txt -c copy -an -y -f avi out_all.avi", so now when I open out_all.avi first 10 seconds is out_1 and then goes out_2_slowed and it is slowed down because of filter "setpts=3*PTS" and it is all correct and working
[15:07:16 CEST] <waressearcher2> I want to be sure that part is clear and understood
[15:07:18 CEST] <tactualtaco> So I just ran it through ffmpeg with a and v set to copy, now it outputs a file is still double in length, but the second half is actually the same a/v playing again
[15:07:37 CEST] <waressearcher2> because next I will explain what is "weird" thing happening
[15:07:43 CEST] <BtbN> I'd guess the type command also appended the file it just generated at the end
[15:07:51 CEST] <BtbN> try piping it into some other directory
[15:08:27 CEST] <BtbN> Or be more specific with your glob and use some output file name it doesn't match.
[15:09:22 CEST] <tactualtaco> Yup that might be the case, give me a minute.
[15:10:10 CEST] <waressearcher2> so I'm clear so far ?
[15:12:33 CEST] <tactualtaco> BtbN, didn't seem to be the case, did it again in a fresh directory.
[15:12:54 CEST] <waressearcher2> so I have these "out_1" and "out_2_slowed" that I concatenate without a problem, now I wanted to add another filter "fade=t=in:st=0:d=0.5" to "out_2_slowed" so I run command: "ffmpeg -i out_2_slowed.avi -c:v mpeg4 -vb 10000k -vf "fade=t=in:st=0:d=0.5" -f avi out_2_slowed_faded.avi", and then I do concatenation again, with out_2_slowed_faded.avi as
[15:12:59 CEST] <waressearcher2>  second part of the video, but when I look past first 10 second of out_all video that part that should be slowed down is not, it is "faded" but not slowed down, and also I got a lot of these: "Non-monotonous DTS in output stream 0:0; previous: 569, current: 390; changing to 570. This may result in incorrect timestamps in the output file." in
[15:13:00 CEST] <BtbN> Well, if you did the exact same thing again, the same thing will happen again
[15:13:05 CEST] <waressearcher2>  console output when concatenating
[15:13:39 CEST] <tactualtaco> BtbN, well in the earlier folder I had an extra ts file which might have been picked up when I merged everything
[15:13:46 CEST] <tactualtaco> That couldn't have happened now.
[15:13:56 CEST] <BtbN> But it most likely picks up the ts file it creates itself
[15:14:38 CEST] <tactualtaco> Alright, I'll just output to a different folder then.
[15:15:25 CEST] <oreWizard> someone help me with pkg-config in compile ffmpeg plz
[15:15:56 CEST] <waressearcher2> so a bit shorte: I created file "out_2_slowed" using filter "setpts=3*PTS" then I get it and used filter "fade=t=in:st=0:d=0.5" and now it is slowed down and faded but when I concate it with other video it is not slowed down, is there an explanation ? I used "setpts=3*PTS" why its effect is not here anymore ? I mean how is it possible ?
[15:21:49 CEST] <tactualtaco> BtbN, you were absolutely right, thanks a lot!
[15:25:57 CEST] <waressearcher2> what is different between "right" and "absolutely right" ?
[15:35:32 CEST] <jacob___> hi, any good dvd ripping tools?
[15:35:39 CEST] <jacob___> open source if possibe
[15:36:04 CEST] <tactualtaco> MakeMKV is good, http://www.makemkv.com/
[15:39:05 CEST] <waressearcher2> can't ffmpeg rip dvds ?
[15:41:17 CEST] <jacob___> it can??
[15:43:36 CEST] <waressearcher2> can it ?
[15:53:20 CEST] <oreWizard> WARNING: pkg-config not found, library detection may fail.
[15:53:26 CEST] <oreWizard> can someone help me with this?
[15:54:01 CEST] <JEEB> not necessarily fatal but pkg-config helps with third party library detection
[15:54:18 CEST] <JEEB> you can just install pkg-config through your package manager to have it available
[15:54:58 CEST] <shevy> are there restrictions in place when merging .mp4 files? I am trying to use the concat protocol
[15:55:12 CEST] <shevy> but the end result is that only one of two .mp4 files is created in the output file
[15:55:28 CEST] <shevy> last command I tried was:
[15:55:30 CEST] <shevy>   ffmpeg -i concat:file1.mp4\|file2.mp4  -acodec copy output.mp4
[15:55:41 CEST] <JEEB> concat protocol and demuxer can have issues with things that cannot be straight concatenated or have an "end" in the demuxer
[15:56:14 CEST] <BtbN> you can't easily concat mp4 files that way.
[15:56:18 CEST] <BtbN> Remux them to ts, and you can.
[15:56:19 CEST] <shevy> I see, that was strange; for .mp3 files it worked very well
[15:56:38 CEST] <JEEB> yes, because those are just raw bit streams
[15:56:47 CEST] <JEEB> which is also why seeking in mp3 files is an effing effort
[15:56:56 CEST] <JEEB> (because they are raw bit streams)
[16:00:17 CEST] <shevy> yep, that worked
[16:00:26 CEST] <shevy> thanks JEEB and BtbN \o/
[16:08:14 CEST] <oreWizard> JEEB
[16:08:21 CEST] <oreWizard> i need pacman -s pkg-config
[16:08:27 CEST] <oreWizard> i did it
[16:08:48 CEST] <waressearcher2> what is "effing effort" ?
[16:09:43 CEST] <JEEB> well it's a raw bitstream so the only real way of doing seeking in it precisely is to actually index it
[16:09:52 CEST] <JEEB> or decode from the beginning every time you seek
[16:10:11 CEST] <JEEB> some mp3 files might have random hacked indexes, of course, which might make it simpler
[16:14:00 CEST] <waressearcher2> right but what is that word "effing" ?
[16:14:01 CEST] <oreWizard> JEEB
[16:14:14 CEST] <oreWizard> i am having the erro again
[16:14:26 CEST] <oreWizard> is there anything i need to put in command-line?
[16:15:22 CEST] <JEEB> waressearcher2: f-word
[16:17:46 CEST] <waressearcher2> does it contains word "king" in it ?
[16:22:10 CEST] <ClearSky> How do I mux Vorbis audio and VP8 encoded video into WebM format, and also set the display aspect ratio to 16:9?
[16:32:03 CEST] <akae> Hi all, I'm having some trouble compiling with QSV support in Ubuntu, mostly with pkg-config and libmfx. Could not find any solution over the list and forums. Has anybody compiled it under Ubuntu 14?
[16:43:49 CEST] <waressearcher2> JEEB: I think you can answer my previous question that no one even tryed to answer
[16:46:23 CEST] <JEEB> ClearSky: ffmpeg -i your_vp8 -i your_vorbis -c copy -aspect 16:9 out.webm
[16:46:27 CEST] <shevy> laters
[16:47:45 CEST] <waressearcher2> JEEB: I have 2 files 10 seconds each: "out_1.avi" and other "out_2_slowed.avi" that I created using filter "setpts=3*PTS", then I concatenated "out_1.avi" and "out_2_slowed.avi" into out_all.avi, and now when I seek 10 seconds inside out_all.avi the video is slowed down, so it works, but now I take that out_2_slowed.avi and apply filter
[16:47:50 CEST] <waressearcher2>  "fade=t=in:st=0:d=0.5" to create "out_2_slowed_faded.avi" and then concatenate "out_1.avi" and "out_2_slowed_faded.avi" and seek 10 seconds in it that half that should be slow is normal its like effect of "setpts=3*PTS" is dissapeared, any ideas why ?
[16:49:26 CEST] <waressearcher2> so to have both "slowed" and "faded" effects I need to do them in one filter, why can't I do first "setpts=3*PTS" and take that resulting file and apply another "fade=t=in:st=0:d=0.5" ?
[16:51:33 CEST] <waressearcher2> its really strange
[16:56:11 CEST] <ClearSky> JEEB: Thank you.
[17:04:18 CEST] <oreWizard> anyone knows how to set PKG_CONFIG_PATH=
[17:11:22 CEST] <JEEB> PKG_CONFIG_PATH=/some/prefix/lib/pkgconfig ./configure --options-here
[17:12:04 CEST] <waressearcher2> export PKG_CONFIG_PATH=/some/prefix/lib/pkgconfig
[17:12:30 CEST] <jasom> BtbN: unfortunately it's mixed interlaced/progressive NTSC so that's a no go... thanks for the suggestion
[17:13:14 CEST] <jasom> BtbN: the solution I stumbled upon was to use mencoder with copy codecs to make an avi and subtitles in vobsub format, then I used that as the source for ffmpeg to an mkv with embedded vobsub titles
[17:14:21 CEST] <oreWizard> JEEB i am using pacman -s pk-config
[17:14:29 CEST] <oreWizard> where do i point at pkg_config-Path
[17:15:48 CEST] <waressearcher2> oreWizard: why not to ask in debian or where it belongs ?
[17:16:17 CEST] <oreWizard> waressearcher2 i want to know if the path is de source or the origin of the pkg-config
[17:16:18 CEST] <jasom> BtbN: the next step is to figure out *how* mplayer/mencoder (and every single DVD player I've tried it in for that matter) times the subs on this and use that for setting the PTS of subtitle streams
[17:16:57 CEST] <akae> waressearcher2: are you trying to compile with libmfx ?
[17:17:12 CEST] <sg90> Hi, is it possible to use the concat muxer to append to the same file rather than a new copy?
[17:17:49 CEST] <waressearcher2> akae: I not try to compile of any
[17:17:56 CEST] <oreWizard> ake i am trying
[17:18:07 CEST] <AndrewMock> Is this normal? I think ffmpeg should show up there: http://paste.ubuntu.com/12418063/
[17:20:00 CEST] <akae> waressearcher2: sorry, I missed the nickname
[17:20:25 CEST] <akae> oreWizard: I'm having trouble too, ERROR: libmfx not found using pkg-config
[17:21:16 CEST] <akae> but  pkg-config --list-all |grep mfx shows libmfx from intel SDK
[17:22:48 CEST] <akae> tried PKG_CONFIG_PATH=/usr/lib/pkgconfig ./configure --enable-libmfx  (...) but have no success :(
[17:23:25 CEST] <AndrewMock> my ~/ffmpeg_build is here http://paste.ubuntu.com/12418120/
[17:31:50 CEST] <akae> I'm just trying to compile ffmpeg with qsv support in Ubuntu, does anybody have a howto that works?
[17:36:06 CEST] <JEEB> akae: config.log says what? use pastebin
[17:36:35 CEST] <AndrewMock> Am I doing something wrong?
[17:41:30 CEST] <waressearcher2> what is qsv ?
[17:42:17 CEST] <akae> waressearcher2: Intel Quick Sync video
[17:43:37 CEST] <akae> JEEB: http://pastebin.com/d9iZFKXM
[17:44:15 CEST] <akae> I thing Intel's Media SDK is not properly installed
[17:44:18 CEST] <akae> *think
[17:45:46 CEST] <JEEB> it's getting found alright but a library of a specific name is not found noted in the pkg-config file
[17:46:11 CEST] <JEEB> or well I'd gueas that linker flag comes from the pc file
[17:46:35 CEST] <JEEB> time to see what kind of libraries ypu have installed there ?
[17:49:24 CEST] <akae> how can I list them?
[17:50:23 CEST] <JEEB> just check the lib dir in your mfx prefix?
[17:51:16 CEST] <akae> ls /opt/intel/mediasdk/lib64/8086/0166/libmfxhw64 libmfxhw64-p.so.1.8  libmfxhw64.so
[17:53:19 CEST] <ClearSky> When doing 2 pass encoding with VP8 (libvpx), is there a way to figure out the optimal bitrate, so I don't have to guess?
[17:53:27 CEST] <ClearSky> Something like a small compression check, or some appropriate ffmpeg commands?
[17:54:49 CEST] <ClearSky> I tried reading VP8 1st Pass stats file in notepad, but it's all garbled.
[17:55:35 CEST] <akae> JEEB: copied all files from /opt/intel/mediasdk/lib64/8086/0166/ to /opt/intel/mediasdk/lib64/ and ./configure does not complain anymore. I keep trying
[17:56:01 CEST] <akae> JEEB: kiitos paljon
[18:39:01 CEST] <akae> JEEB: compiling almost succeeded :_( http://pastebin.com/8Bqa2Rq7
[18:59:35 CEST] <jasom> can ffmpeg read a vobsub format subtitle directly?  Do I pass the .idx or .sub as the input file if so?
[19:15:16 CEST] <C0nundrum> Helllo
[19:15:51 CEST] <C0nundrum> Running a conversion job using the command 'ffmpeg -i "'+link+'" -t  10 "'+file+'"'; and it had a high cpu usage, What are some ways to reduce this ?
[19:16:09 CEST] <c_14> -c copy
[19:20:13 CEST] <C0nundrum> o whoops forgot to mension the output is .gif
[19:37:29 CEST] <C0nundrum> would -copy still help ?
[19:37:45 CEST] <c_14> AS long as the input is also gif
[19:39:58 CEST] <C0nundrum> the input is an hls stream
[19:42:53 CEST] <C0nundrum> is there a way for me to maybe slow down the conversion speed ?
[19:46:03 CEST] <C0nundrum> or is there a way i can force it to execute on my gpu instead ?
[19:47:00 CEST] <Canar> C0nundrum: uh, why is cpu use a problem?
[19:47:31 CEST] <C0nundrum> using cuda
[19:48:05 CEST] <C0nundrum> Well its being run on a perosnal pc and it slows down my pc a abit with 10 threads
[19:48:40 CEST] <C0nundrum> so i would rather them either run at a low speed speed as to not effect cpu usage as much or convert using cuda
[19:48:42 CEST] <C0nundrum> if possible
[19:48:57 CEST] <jasom> C0nundrum: at least libx264 takes a threads option
[19:49:18 CEST] <jasom> C0nundrum: also you can use "nice" to run it at a lower priority assuming you're on a *nix
[19:50:02 CEST] <jasom> oh, missed that you're doing gif output
[19:59:03 CEST] <C0nundrum> so there no way ?
[19:59:57 CEST] <DHE> gif output probably isn't going to be optimized any better.
[20:01:34 CEST] <Canar> C0nundrum: It's not a problem a lot of people are concerned about.
[20:04:29 CEST] <C0nundrum> well i mean if i cold run it on my gpu i would care about the usage
[20:04:35 CEST] <C0nundrum> wouldn't*
[20:05:50 CEST] <DHE> I don't think GIF encoding has been converted to CUDA yet
[20:06:00 CEST] <DHE> and I'm not expecting there's a high demand for that
[20:12:25 CEST] <C0nundrum> hm
[20:13:03 CEST] <C0nundrum> Do you think a mp4 could be made with the same or better quality as a gif at the same or lower size ?
[20:13:41 CEST] <BtbN> As a gif has no quality...
[20:13:45 CEST] <fritsch> :-)
[20:13:55 CEST] <fritsch> use HEVC
[20:14:02 CEST] <C0nundrum> well image quality :/
[20:14:07 CEST] <fritsch> wait 2 days for encoding - but that would be a chance
[20:14:22 CEST] <BtbN> A gif is huge, and has only a very limited color palete.
[20:14:27 CEST] <BtbN> And still looks horrible.
[20:15:38 CEST] <jasom> C0nundrum: even the MS H.263 codec could outperform a gif for most video content
[20:16:13 CEST] <jasom> and it had a buggy rgb/yuv conversion algorithm
[20:18:07 CEST] <DHE> I assume this is for a browser. What is the current ideal codec? Still H264 in .mp4, or is webm+vp9 better supported?
[20:18:25 CEST] <jasom> h264 in mp4 is the most widely supported
[20:19:15 CEST] <C0nundrum> hm i might try my with with that then.
[20:19:21 CEST] <C0nundrum> try my luck*
[20:19:29 CEST] <jasom> http://caniuse.com/#feat=webm http://caniuse.com/#search=mp4
[20:19:58 CEST] <JEEB> DHE: the problem with vp9 is still the fact that Google doesn't dogfood it. youtube does use it, but they have reimplemented major spots of the encoder (and it is still unbearably slow)
[20:20:33 CEST] <jasom> C0nundrum: do not use the "losless" x264 option though, that will break compatibility with at least some H.264 decoders
[20:20:51 CEST] <JEEB> so if you're OK with a very slow encoder with far from perfect rate control, be my guest :P
[20:21:14 CEST] <jasom> JEEB: and spotty browser support
[20:21:56 CEST] <JEEB> chrom{e,ium}, firefox, opera and MS's edge (incoming support) sounds pretty good
[20:22:04 CEST] <JEEB> the main issue is mobile if you just look at the decoding side
[20:22:06 CEST] <C0nundrum> ah
[20:22:16 CEST] <jasom> actually anything outside of H.264 baseline may not work on android, depending on the version
[20:22:30 CEST] <JEEB> uhh
[20:22:37 CEST] <JEEB> that probably means the software decoder included with android
[20:22:41 CEST] <JEEB> which you just don't want to use
[20:23:10 CEST] <JEEB> and while yes, there are some really crappy devices due to everyone being able to make whatever android device they want, you can support almost anything sane with main, and most things by now support high
[20:23:36 CEST] <JEEB> if you really really want to support very crappy ARMv6 devices on Android 2.3.x, be my guest of course
[20:23:44 CEST] <JEEB> (or MIPS)
[20:24:18 CEST] <jasom> JEEB: I'm pretty sure some 3.0 devices will only do BP in the system browser, but I might be wrong
[20:24:51 CEST] <JEEB> if they have to fall back on the SW decoder, possibly. But at that point you're already losing since the CPU will get used a lot
[20:25:04 CEST] <JEEB> I haven't seen any such devices in general, though :/
[20:25:14 CEST] <JEEB> android 3+ is armv7 and intel only more or less
[20:25:34 CEST] <JEEB> and most ARMv7 chipsets have at least main profile support
[20:25:44 CEST] <JEEB> (in the hw decoder)
[20:28:01 CEST] <JEEB> looking at https://developer.android.com/about/dashboards/index.html android 3 would also end up having less than 0.1% market share on the google store
[20:28:20 CEST] <JEEB> it was a very limited stopgap release for a limited set of tablets
[20:29:05 CEST] <jasom> JEEB: I could have sworn my Kyocera Torque had issues with main, and it ran 4.0
[20:29:13 CEST] <jasom> I had to fallback to sw rendering
[20:29:24 CEST] <JEEB> main, but which level?
[20:29:31 CEST] <JEEB> profile is features, level is how much memory the decoder requires
[20:29:44 CEST] <JEEB> if you go over your device's level, then whatever the profile it will fail
[20:30:02 CEST] <jasom> ah
[20:30:35 CEST] <JEEB> Qualcomm MSM8960 Snapdragon S4 Plus
[20:30:45 CEST] <JEEB> if this is the 2013 CDMA device
[20:30:54 CEST] <JEEB> "Kyocera Torque E6710"
[20:32:35 CEST] <jasom> that sounds right
[20:33:08 CEST] <jasom> I don't have it any more since an OTA firmware update bricked it and Kyocera refused to fix it since it was 6 weeks outside of warrenty
[20:34:45 CEST] <JEEB> looking at google results that chipset should be capable of doing high profile, level 4.1
[20:35:27 CEST] <jasom> JEEB: It wouldn't surprise me if Kyocera messed something up on the software side.  They went out of their way to make the software experience unbearable on it
[20:36:07 CEST] <JEEB> of course
[20:42:23 CEST] <JEEB> if you are thinking of aiming towards very low income areas like certain parts of africa or india or pakistan etc, then you might want to look at the general sales ratings there, but given the cheap price of ARMv7 chipsets at this point, those people who buy a smartphone are most probably getting an ARMv7 one
[20:42:54 CEST] <JEEB> and those people who are getting an SMS/phone phone then those will be out of your sector anytways
[20:42:57 CEST] <JEEB> *anyways
[21:08:17 CEST] <netekmalta> does ffmpeg create logs if so could somebody tell me where these logs are stored on Ubuntu?
[21:08:48 CEST] <netekmalta> when transcoding rtmp to hls I sometimes must run the command 4-5 times for it to work, other times it works all the time -_-
[21:08:53 CEST] <llogan> which ubuntu version
[21:09:30 CEST] <JEEB> logs go to stderr unless you are using a wrapper, in which case I don't think anyone here has any idea :P
[21:09:52 CEST] <netekmalta> hah
[21:09:55 CEST] <netekmalta> llogan 14.04
[21:11:01 CEST] <llogan> you can look at the -report option if you don't feel like redirecting
[21:11:11 CEST] <netekmalta> ok
[21:14:23 CEST] <netekmalta> now that I setup output to a txt file its working fine and restarted the stream 10 times....
[21:14:34 CEST] <netekmalta> it doesnt want to tell me what I am doing wrong :(
[21:14:49 CEST] <JEEB> are you sure you are pushing out stderr?
[21:14:56 CEST] <JEEB> instead of stdout
[21:15:10 CEST] <llogan> redirect to tee so you can see it in console too
[21:15:38 CEST] <netekmalta> JEEB yes, but the stream is now transcoding whenever i attempt to transcode
[21:16:02 CEST] <JEEB> also if you're using the packaged stuff, the libav* libraries and tools are provided by the libav project, so you should be using the avconv binary instead of the ffmpeg one (although that version should no longer have ffmpeg altogether due to it being removed from libav)
[21:16:30 CEST] <netekmalta> JEEB ?
[21:17:21 CEST] <JEEB> ubuntu 14.04 has libav-tools provide avconv and possibly a script that pokes you towards it if you call ffmpeg (because libav 9 no longer had a binary called "ffmpeg")
[21:17:24 CEST] <netekmalta> this is the command I am using to transcode, does this look right?
[21:17:26 CEST] <netekmalta> http://pastebin.com/ckrc1Hjk
[21:17:38 CEST] <netekmalta> JEEB ok
[21:17:59 CEST] <JEEB> if you built your own ffmpeg or got it somewhere else, then of course use it if it's what you've settled upon
[21:18:21 CEST] <netekmalta> JEEB I modified it from an example
[21:18:34 CEST] <netekmalta> just curious to find out why it doesnt transcode all the time
[21:18:48 CEST] <netekmalta> sometimes when I run it I only get audio, other times i get both video and audio
[21:19:02 CEST] <JEEB> well you'll just have to see the logging when it fails
[21:19:07 CEST] <netekmalta> yeah
[21:19:36 CEST] <JEEB> anyways, since those short options work for you, you are already using a new enough ffmpeg or avconv
[21:20:07 CEST] <netekmalta> ok good
[21:20:17 CEST] <JEEB> the parameters themselves look fine
[21:20:33 CEST] <JEEB> as long as the medium preset is fast enough on your machine for re-encoding
[21:20:39 CEST] <JEEB> (it's the default for libx264)
[21:21:20 CEST] <netekmalta> I will keep restarting until it gives an error
[21:32:42 CEST] <MadTBone> are there any recent reports of ffmpeg/ffplay failing to jump to specific points in either flac or wav files?  In other words -    ffmpeg -ss 0:20:30.000 -i foo.wav outputfile   somehow produces bad behavior.
[21:35:17 CEST] <MadTBone> also, is the August 1st post on http://ffmpeg.org/index.html#news actually legit (sorry if I'm late in realizing this)???
[21:43:24 CEST] <t4nk303> Hi! When I copy streams from mts video with pcm_bluray audio I lose audio
[21:43:27 CEST] <t4nk303> Is it bug?
[21:44:13 CEST] <t4nk303> ffmpeg -i 1.mts -c copy 2.mts
[21:44:38 CEST] <t4nk303> 1. mts with audio, 2.mts without
[21:46:12 CEST] <t4nk303> mts is from sony pmw-x70 camcorder
[22:02:47 CEST] <termos> http://pastebin.com/EwyAEPHd what's missing for this mpeg dash manifest to be valid?
[22:08:17 CEST] <llogan> t4nk303: how many questions have you asked here?
[22:11:32 CEST] <llogan> ...and how many times do we have to ask for complete console output?
[22:14:07 CEST] <t4nk303> it is my first question
[22:15:54 CEST] <t4nk303> C:\Program Files\ffmpeg>ffmpeg -i I:\PRIVATE\AVCHD\BDMV\STREAM\00000.MTS -c copy d:\bb.mts ffmpeg version N-75275-gd13a2df Copyright (c) 2000-2015 the FFmpeg developers   built with gcc 4.9.3 (GCC)   configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzli b --enable-fontconfig --enable-frei0r --enable-gnutls --enable-iconv --enable-libass --enable-libblu ray --enable-libbs2b --enable-libcac
[22:17:00 CEST] <t4nk303> http://pastebin.com/EJQ6qmxt
[22:17:34 CEST] <t4nk303> this is log from cmd
[22:20:45 CEST] <t4nk303> http://failiem.lv/down.php?i=ohzjrtb
[22:20:47 CEST] <t4nk303> here is file
[22:21:24 CEST] <t4nk303> after this command ffmpeg -i 1.mts -c copy 2.mts audio from 2.mts disappears
[22:21:57 CEST] <t4nk303> mediainfo does not see audio at all
[22:24:28 CEST] <t4nk303> if I use ffmpeg -i I:\PRIVATE\AVCHD\BDMV\STREAM\00000.MTS -codec:v copy -acodec libvo_aacenc -ab 320k -ac 2 d:\bb.mts
[22:24:46 CEST] <t4nk303> output file has audio
[22:25:17 CEST] <c_14> t4nk303: looks like a bug to me. Report it on trac.
[22:29:24 CEST] <c_14> Add that output from the pastebin to the issue and attach the file
[22:31:08 CEST] <t4nk303> Thank You! And one more question: if I want to concat with standard concat protocol I use this command -i "concat:input1.mpg|input2.mpg|input3.mpg" -c copy output.mpg
[22:31:42 CEST] <c_14> That's not a question, but yes.
[22:31:52 CEST] <t4nk303> But if I have a lot of files I could not run cmd because of command-line string limitation. Is there any solution?
[22:31:57 CEST] <c_14> Ah
[22:32:15 CEST] <c_14> hmm
[22:32:32 CEST] <c_14> You could build the list programatically using your shell
[22:32:40 CEST] <c_14> That or use the demuxer
[22:33:34 CEST] <c_14> ie, store the string in a file and then -i $(cat file), I _think_ that sort of substitution gets around the length limitation
[22:35:28 CEST] <t4nk303> I do not wat to use demuxer. I try it but without any result It understands only "concat:input1.mpg|input2.mpg|input3.mpg"
[22:38:25 CEST] <t4nk303> You mean that I have to use -i list.txt -c copy output.mpg
[22:39:05 CEST] <t4nk303> in list.txt should write "concat:input1.mpg|input2.mpg|input3.mpg"
[22:43:06 CEST] <c_14> -f concat -i list.txt; and list.txt contains file input1.mpg\nfile input2.mpg etc
[22:43:14 CEST] <c_14> https://trac.ffmpeg.org/wiki/Concatenate#demuxer
[22:44:22 CEST] <t4nk303> Yes. But it use demuxer function
[22:44:40 CEST] <c_14> Oh, you mean with the shell expansion magic?
[22:44:52 CEST] <t4nk303> But I want to join files via file level
[22:45:01 CEST] <c_14> yeah, just write that in the file and use something like $(cat file)
[22:52:19 CEST] <t4nk303> I try run ffmpeg -i list.txt -c copy d:\bb.mts
[22:52:32 CEST] <t4nk303> in list.txt I put
[22:52:34 CEST] <t4nk303> "concat:input00000.Mts|00001.Mts"
[22:53:45 CEST] <t4nk303> Nothing happened
[22:53:52 CEST] <t4nk303> http://pastebin.com/xWDHH9Fn
[22:56:20 CEST] <c_14> That won't work. The concat protocol being a protocol needs the string starting with protocol_name:, which means you can't just feed ffmpeg the file. You have to get your shell to expand the content of the file for you.
[23:18:34 CEST] <oreWizard> ffmpeg avio_open() failed to open output file
[23:18:38 CEST] <oreWizard> somone knows about this
[23:35:23 CEST] <llogan> t4nk303: sorry then. there seem to be many t4nk users then.
[23:40:48 CEST] <t4nk303> <c_14>, ffmpeg -i < command.txt -c copy output.mpg
[23:41:01 CEST] <t4nk303> Does not work. hmmm.
[23:41:58 CEST] <t4nk863> I am trying to create filenames with timestamps that have millisecond resolution. strftime does not support milliseconds
[23:42:09 CEST] <t4nk863> is there a work around to this?
[23:42:43 CEST] <c_14> llogan: case in point
[23:42:56 CEST] <llogan> how odd
[23:44:02 CEST] <iive> they must be russians on vacation!
[23:44:30 CEST] <llogan> iive: make a sign that points to Syria and/or Ukraine
[23:45:07 CEST] <llogan> t4nk303, t4nk863: does your IRC client automatcially choose your username?
[23:45:11 CEST] <c_14> t4nk863: I don't think so. I know date supports nanoseconds. You could fumble around with those.
[23:45:53 CEST] <t4nk863> llogan: yes, I am using a browser
[23:46:03 CEST] <t4nk863> @c_14 thanks!
[23:46:07 CEST] <t4nk303> Me too.
[23:46:28 CEST] <llogan> and here i always thought t4nk* was the same guy from india
[23:46:59 CEST] <t4nk863> well I am from India but I am chatting from the US :)
[23:47:14 CEST] <t4nk863> but this is my first time around :)
[23:54:19 CEST] <t4nk303> I think this impossible in windows
[23:56:06 CEST] <iive> t4nk303: you don't have -f concat
[23:56:32 CEST] <iive> you should not skip critical options!
[23:57:12 CEST] <t4nk303> f concat works on stream level
[23:58:19 CEST] <t4nk303> it is no good - because of there are may be errors when demux mux streams
[23:59:20 CEST] <t4nk303> I need join them via file level
[00:00:00 CEST] --- Wed Sep 16 2015



More information about the Ffmpeg-devel-irc mailing list