[Ffmpeg-devel-irc] ffmpeg.log.20161021

burek burek021 at gmail.com
Sat Oct 22 03:05:01 EEST 2016


[00:00:57 CEST] <b6s3d> furq
[00:05:32 CEST] <b6s3d> this command seams to work: ffmpeg -i "$file" -b 450k -s 320x240 -vcodec libxvid -ab 128k -ar 48000 -acodec aac -strict -2 "$outfile";done
[00:05:44 CEST] <b6s3d> though it outputs some errors i am going to use it
[00:57:46 CEST] <sha0> Where can I find the OS requirements for FFMPEG?
[01:11:25 CEST] <llogan> sha0: to compile?
[01:11:43 CEST] <sha0> llogan: Sorry, no.  For the prebuilt, downloadable packages.
[01:11:53 CEST] <babadoc> Hello.
[01:12:04 CEST] <sha0> For example, what is the minimal Windows version?
[01:12:07 CEST] <babadoc> I would like to ask a question about installing ffmpeg on a 32 bit ubuntu server.
[01:12:24 CEST] <babadoc> The installation process is the same, yes?
[01:13:02 CEST] <c_14> The same as?
[01:13:10 CEST] <babadoc> The same as this: https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[01:13:24 CEST] <c_14> Why wouldn't it be
[01:13:38 CEST] <babadoc> Because h265 doesn't seem to install correctly.
[01:13:56 CEST] <babadoc> Throwing me an error such as "make: *** No rule to make target 'distclean'.  Stop."
[01:14:33 CEST] <c_14> sha0: you'd probably have to ask on the zeranoe forum
[01:14:38 CEST] <babadoc> I don't think it is a necessity, because h264 installs correctly.
[01:14:47 CEST] <c_14> babadoc: just don't distclean
[01:14:48 CEST] <llogan> sha0: you should ask Zeranoe if you're referring to his builds. I believe XP support was dropped a year or two ago.
[01:15:04 CEST] <sha0> c_14: Oh, I see.  llogan: D'oh.  Sucks to be me.
[01:15:04 CEST] <babadoc> distclean? I am sorry, I don't follow.
[01:15:22 CEST] <c_14> babadoc: distclean just removes objects and stuff in the build directory for that project. It in no way influences the actual built and installed files
[01:15:44 CEST] <c_14> And is therefore not an important part of the build process
[01:15:49 CEST] <babadoc> Oh, I see.
[01:16:06 CEST] <llogan> that confuses users at times. i may remove it. on the other hand some users go back and re-configure with different options and experience issues.
[01:17:42 CEST] <babadoc> The end goal of installing ffmpeg for me was to extract frames from certain times in a .webm video that I have. I tried following this tutorial https://trac.ffmpeg.org/wiki/Create%20a%20thumbnail%20image%20every%20X%20seconds%20of%20the%20video
[01:18:12 CEST] <furq> babadoc: https://www.johnvansickle.com/ffmpeg/
[01:18:12 CEST] <babadoc> But, I could not extract frames from other times than 1 second in.
[01:18:19 CEST] <furq> if you don't need non-free codecs then just use those
[01:20:26 CEST] <babadoc> furq: I don't quite follow. Sorry. I have the codecs from that tutorial (libvpx, libopus, etc) installed, if that is what you are asking me. Could you elaborate?
[01:21:39 CEST] <babadoc> I tried clicking on your name to mention you, but it doesn't seem to be working.
[01:21:50 CEST] <babadoc> (I haven't used IRC before)
[01:24:02 CEST] <llogan> not sure why anyone would use XP
[01:28:04 CEST] <babadoc> It really is quite peculiar... why the command "ffmpeg -i input.webm -ss 00:00:14.435 -vframes 1 out.png" would work, but... Oh...
[01:28:26 CEST] <babadoc> It seems that ffmpeg takes quite longer to seek and take frames from video than i had anticipated.
[01:28:36 CEST] <babadoc> How irritating.
[01:29:11 CEST] <babadoc> Is there any workaround to the slow seek times in ffmpeg?
[01:30:01 CEST] <babadoc> To clarify: I don't mean the GUI type of seeking, I mean seeking for frames in a webm video via ffmpeg CLI using a command like this: "ffmpeg -i input.webm -ss 00:00:14.435 -vframes 1 out.png"
[01:30:59 CEST] <llogan> use -ss as an input option
[01:31:40 CEST] <babadoc> http://pastebin.com/KZ4Xx8wU is the error log
[01:31:43 CEST] <babadoc> its quite short
[01:32:13 CEST] <babadoc> Logan, you want me to use the -ss option too? I am not sure how I would add that and still have the command work.
[01:32:43 CEST] <llogan> that console output is not complete and is missing the command so I don't know what you are intending to show with it
[01:32:55 CEST] <babadoc> Oh, sorry. I shall complete it.
[01:35:19 CEST] <babadoc> http://pastebin.com/qLmTHSCH
[01:35:23 CEST] <babadoc> Sorry.
[01:35:37 CEST] <babadoc> Any more information you need?
[01:35:52 CEST] <furq> babadoc: i meant that you probably don't need to compile ffmpeg
[01:36:06 CEST] <furq> the only reason to compile it instead of using those static binaries is if you need non-free codecs like libfdk-aac
[01:36:31 CEST] <babadoc> I installed it using that tutorial.
[01:36:40 CEST] <babadoc> Trying to run sudo apt install ffmpeg would not work
[01:36:55 CEST] <babadoc> asked me to use --fix-missing and that would not work either
[01:37:00 CEST] <babadoc> Easier to install via source.
[01:37:15 CEST] <babadoc> The only way was to install via source.
[01:40:17 CEST] <babadoc> The OS the server is running is ubuntu 16.0.4, not 16.0.2, if that matters.
[01:47:08 CEST] <llogan> you don't need to install the binary furq was referring to. you just execute it
[01:48:29 CEST] <babadoc> Execute the x86 build on the FFmpeg static builds list that furq referred to?
[01:48:33 CEST] <babadoc> https://www.johnvansickle.com/ffmpeg/
[01:48:48 CEST] <babadoc> (The first download on the list)
[01:49:52 CEST] <Phi> Moop
[01:49:54 CEST] <Phi> Hey folks
[01:50:02 CEST] <Phi> any idea how to turn on libx264 logging in ffmpeg?
[01:50:44 CEST] <babadoc> Isn't enabled by default?
[01:52:46 CEST] <Phi> would it be outputting to av_log then?
[01:53:46 CEST] <babadoc> Yes, I think.
[01:53:55 CEST] <babadoc> But don't take my word for it.
[01:54:02 CEST] <babadoc> Look
[01:54:12 CEST] <babadoc> http://pastebin.com/raw/qLmTHSCH
[01:54:22 CEST] <babadoc> That outputted errors for h264, and I just ran a normal command
[01:54:55 CEST] <Phi> the paste expired
[01:55:05 CEST] <babadoc> Not the raw one. (It did?)
[01:55:19 CEST] <babadoc> Did the raw one expire?
[01:55:49 CEST] <Phi> well, I'm having excruciating difficulty trying to change a H264 profile when moving from RTSP->MP4
[01:55:53 CEST] <Phi> I think it did
[01:55:56 CEST] <Phi> it redirects me out of the raw
[01:56:02 CEST] <babadoc> http://pastebin.com/unrTJtiW
[01:56:09 CEST] <babadoc> See?
[01:56:13 CEST] <babadoc> It posts errors
[01:56:21 CEST] <babadoc> I don't think there is a verbose option for ffmpeg
[01:56:48 CEST] <furq> there is, but he's not using ffmpeg
[01:57:08 CEST] <babadoc> What do you mean furq?
[01:57:23 CEST] <furq> i assume from av_log that he's using the libraries directly
[01:57:38 CEST] <babadoc> Ah. But he did state that he was using ffmpeg.
[01:57:41 CEST] <Phi> yea, C++y
[01:57:47 CEST] <babadoc> Oh. I see.
[01:57:52 CEST] <Phi> sorry, the terminology isn't very obvious
[01:57:56 CEST] <Phi> I'm new to the whole thing
[01:58:04 CEST] <babadoc> Don't worry, so am I.
[01:58:07 CEST] <furq> yeah the libav fork nonsense has made it really annoying
[01:58:15 CEST] <furq> you can't say "libav" now without people thinking you mean avconv
[01:58:21 CEST] <babadoc> lol, just fork ALL the things
[01:58:27 CEST] <furq> libav* is the best compromise i've seen
[01:58:27 CEST] <Phi> fork that
[01:58:32 CEST] <Phi> not very knife behaviour
[01:58:37 CEST] <furq> and screw libswresample
[01:58:45 CEST] <babadoc> haha
[01:59:29 CEST] <babadoc> So furq... do you know what I could do?
[01:59:40 CEST] <Phi> everyone's giving me this very clever, accurate advice about how to change the profile
[01:59:50 CEST] <Phi> and the C++ is like "nope, profile must be high or the sun explodes"
[02:00:04 CEST] <Phi> I must have tried over a hundred different ways by now, it's been weeks :P
[02:00:06 CEST] <furq> babadoc: move -ss before -i
[02:00:13 CEST] <babadoc> haha, phi
[02:00:19 CEST] <babadoc> Okay furq, will do.
[02:00:36 CEST] <Phi> you think I'm exaggerating
[02:00:37 CEST] <furq> failing that, if you're reencoding then use the trim filter
[02:00:55 CEST] <Phi> if I wasn't paid by the hour I'd be totally miffed
[02:01:06 CEST] <furq> i'm pretty sure i remember trying to help you with this a few weeks ago, so i don't think you're exaggerating
[02:01:16 CEST] <babadoc> You guys are professionals, huh
[02:01:20 CEST] <babadoc> I'm only 15
[02:01:26 CEST] <Phi> I'm so professional you don't even know
[02:01:32 CEST] <babadoc> haha
[02:01:35 CEST] <furq> sadly my knowledge of the api doesn't extend beyond saving a png
[02:01:51 CEST] <babadoc> Gif works natively too Phi
[02:01:55 CEST] <Phi> that's the annoying part, I can read the RTSP fine, even save an MP4
[02:02:04 CEST] <Phi> snapshots and displaying the frames, easy
[02:02:23 CEST] <Phi> but changing the profile of the MP4? nah
[02:02:35 CEST] <Phi> no matter what I do it locks it back to whatever the RTSP is doing
[02:03:08 CEST] <Phi> at this rate I'm going to have to byte-scan every single packet and replace the profile in any headers with my one :p
[02:03:26 CEST] <Phi> not CPU-intensive or hacky at all
[02:04:07 CEST] <babadoc> hahahaha
[02:04:25 CEST] <babadoc> that really sucks :(
[02:04:44 CEST] <babadoc> hey phi, what are you trying to accomplish?
[02:04:52 CEST] <Phi> I like the juxtaposition of your last two messages there
[02:05:11 CEST] <Phi> Once I get the video done, that'll be it
[02:05:15 CEST] <Phi> for the most part
[02:05:23 CEST] <Phi> it's for a CCTV system, ironically
[02:05:28 CEST] <babadoc> lol what
[02:05:35 CEST] <Phi> using network cams
[02:05:40 CEST] <babadoc> yeah, i get that
[02:05:46 CEST] <babadoc> i know what a CCTV system is
[02:06:00 CEST] <Phi> I even managed to write a motion detection for it in a couple of days
[02:06:10 CEST] <Phi> from scratch :P
[02:06:13 CEST] <babadoc> oh, you have to encode whatever format the cctv cameras output with ffmpeg
[02:06:21 CEST] <babadoc> and its not working, huh
[02:06:26 CEST] <Phi> well, with C++
[02:06:36 CEST] <babadoc> with c D:
[02:06:39 CEST] <Phi> and I can get a MP4 output, but I can't change the video profile - to baseline, high, etc
[02:07:03 CEST] <Phi> I guess C, since ffmpeg is C, but you can invoke it in C++ all the same, so...
[02:07:17 CEST] <babadoc> i see
[02:07:21 CEST] <Phi> the MP4 timing is all off as well
[02:07:22 CEST] <babadoc> i took a small class in c++ over the summer
[02:07:27 CEST] <babadoc> it was like one week
[02:07:32 CEST] <babadoc> and i HATED it
[02:07:42 CEST] <babadoc> you have to add the ; at the end of everything
[02:08:02 CEST] <Phi> really;
[02:08:06 CEST] <Phi> that must have sucked;
[02:08:11 CEST] <babadoc> >:((((
[02:08:19 CEST] <furq> better that than ASI
[02:08:20 CEST] <babadoc> lol
[02:08:23 CEST] <Phi> but at least you don't have to work on your punctuation;
[02:08:35 CEST] <babadoc> i have no idea what ASI is
[02:08:41 CEST] <Phi> just run-on everything. // syntax error 'everything' not defined
[02:08:41 CEST] <furq> automatic semicolon insertion
[02:08:55 CEST] <babadoc> oh,
[02:09:01 CEST] <babadoc> lol phi
[02:09:03 CEST] <furq> javascript requires semicolons to terminate statements, but the compiler will insert them for you
[02:09:06 CEST] <furq> except when it doesn't
[02:09:13 CEST] <babadoc> good luck lol
[02:09:20 CEST] <babadoc> go back and look through your code for years
[02:09:30 CEST] <babadoc> trying to find the one error
[02:09:35 CEST] <furq> on the rare occasion i have to write some js i put them in myself to save the hassle
[02:09:38 CEST] <Phi> I actually like C++ because it dies easily over the tiniest thing
[02:09:46 CEST] <Phi> so you get much better typing skills
[02:09:53 CEST] <babadoc> well, that can be a benefit at times
[02:09:59 CEST] <babadoc> because then you know there is an error
[02:10:14 CEST] <babadoc> sometimes you can't tell
[02:10:14 CEST] <furq> i guess the stakes are higher when you have to wait half an hour and then get an 18,000 line error message
[02:10:24 CEST] <babadoc> ha
[02:10:36 CEST] <Phi> speaking of which, before I did libav*, I had to find a library
[02:10:50 CEST] <Phi> to preface this, I've never worked with video, displaying, or webcams
[02:10:58 CEST] <Phi> so I went through GStreamer, OpenCV...
[02:11:15 CEST] <Phi> OpenCV literally took 40 minutes every time I rebuilt it
[02:11:25 CEST] <Phi> and I have a beefy dev PC, so...
[02:11:25 CEST] <babadoc> are you using ffserver to stream the video?
[02:11:35 CEST] <Phi> I don't think I am, no
[02:11:36 CEST] <furq> i hope not
[02:11:42 CEST] <Phi> it's not my webcam firmware
[02:12:10 CEST] <Phi> "Media Server V3.1.2" is the SDP description
[02:12:39 CEST] <Phi> the dumbest part was
[02:13:01 CEST] <Phi> after about a month of kicking the OpenCV install around, finding debug versions, finding DLLs, linking them in the project, going through OpenCV tutorials...
[02:13:07 CEST] <Phi> turns out OpenCV doesn't support any network access
[02:13:12 CEST] <babadoc> ;-;
[02:13:16 CEST] <babadoc> that SUUUUUCKS
[02:13:20 CEST] <babadoc> holy crap
[02:13:33 CEST] <Phi> USB webcams it can do, but nothing over network
[02:13:46 CEST] <Phi> then there's libav*...
[02:14:21 CEST] <klaxa> it looks like ffserver is going to be deprecated in the long run, huh?
[02:14:24 CEST] <Phi> for which I had to set up mingw64, install the right packages, mess around with paths, find the right cmdl for make, and likewise for x264...
[02:15:09 CEST] <babadoc> is mpv a ffmpeg fork?
[02:15:14 CEST] <babadoc> or is it something different
[02:15:16 CEST] <klaxa> no
[02:15:20 CEST] <furq> klaxa: i'm pretty sure it'll be gone in 3.2
[02:15:20 CEST] <klaxa> something different
[02:15:27 CEST] <babadoc> thats weird
[02:15:32 CEST] <furq> there's an announcement of its death on the site somewhere
[02:15:33 CEST] <Phi> I think it's a type of car
[02:15:36 CEST] <klaxa> it uses ffmpeg's libraries
[02:15:43 CEST] <babadoc> ah, thats what iti s
[02:15:48 CEST] <furq> it's a fork of mplayer
[02:15:57 CEST] <babadoc> oh, neat
[02:16:03 CEST] <klaxa> oh? well there was a guy trying to save it, but it probably is too huge of a task
[02:16:15 CEST] <Phi> bit like OpenOffice :p
[02:16:27 CEST] <furq> it's a good idea but it needs burning down and starting from scratch
[02:16:42 CEST] <Phi> they're chucking that out the window now, since LibreOffice is gaining more ground
[02:16:49 CEST] <klaxa> pretty much
[02:16:51 CEST] <babadoc> ha, openoffice
[02:16:56 CEST] <furq> it must have been around for a decade at this point and not much has changed from what i can tell
[02:17:06 CEST] <babadoc> i havent heard that program in a while
[02:17:07 CEST] <furq> it still has some of the issues it had when i used it years ago
[02:17:09 CEST] <klaxa> i still have patches lying around from gsoc 2015, but wow
[02:17:36 CEST] <klaxa> the whole design is reeeeeally weird
[02:17:51 CEST] <babadoc> yeah, thanks guys, putting -ss before the command worked nicely
[02:17:56 CEST] <babadoc> it now seeks instantly
[02:17:58 CEST] <babadoc> :)
[02:18:07 CEST] <Phi> yea, the team got so small they could only push out one update in six months
[02:18:15 CEST] <babadoc> well, it still throws errors, but it works, thats all i care about
[02:18:42 CEST] <klaxa> Phi: the ffserver team?
[02:18:53 CEST] <Phi> nah, the OpenOffice
[02:18:57 CEST] <klaxa> ah
[02:19:01 CEST] <furq> are they actually ditching openoffice
[02:19:10 CEST] <Phi> yea, they announced it about a month or so back
[02:19:12 CEST] <furq> i'm sure i read fairly recently that they're sticking with it
[02:19:45 CEST] <furq> http://mail-archives.apache.org/mod_mbox/openoffice-dev/201609.mbox/%3C008d01d204a9%24bd37caa0%2437a75fe0%24%40apache.org%3E
[02:19:52 CEST] <furq> if you mean this then that isn't an announcement of its death
[02:20:02 CEST] <furq> although a bunch of places reported on it being that
[02:20:24 CEST] <babadoc> just fork it to death, that will solve everyones problems
[02:20:26 CEST] <babadoc> right
[02:20:37 CEST] <furq> "announcement of its death" is a really unwieldy phrase to have used twice in five minutes
[02:20:39 CEST] <klaxa> no...
[02:21:12 CEST] <Phi> hmm
[02:21:30 CEST] <Phi> well it seems they're going to that point, but they're dragging their heels for a couple more updates and hoping another company will pay to keep it going
[02:21:47 CEST] <Phi> that's my guess
[02:22:01 CEST] <Phi> the article I read made the demise sound pretty definite
[02:22:27 CEST] <klaxa> well everyone and their dog is using libreoffice nowadays, no?
[02:22:43 CEST] <furq> i know a guy who works at apache who uses libreoffice
[02:22:46 CEST] <babadoc> i use a mac
[02:22:50 CEST] <furq> so yeah
[02:22:56 CEST] <babadoc> no libreoffice here :)
[02:23:12 CEST] <babadoc> just overheating
[02:23:15 CEST] <Phi> the devs want to keep OpenOffice going, but Apache doesn't seem too interested in it, so it's up in the air
[02:23:45 CEST] <Phi> OpenOffice's Mac equivalent is NeoOffice
[02:23:51 CEST] <Phi> it is the one
[02:23:51 CEST] <furq> something like that, yeah
[02:23:59 CEST] <furq> also there's some bad blood between the two camps
[02:24:06 CEST] <Phi> the one for the office o.ó
[02:24:23 CEST] <Phi> there would be, they're inherently rivals
[02:24:27 CEST] <furq> it's hard to imagine bad blood between developers of two forks of the same software, isn't it
[02:24:31 CEST] <furq> but amazingly it's true
[02:24:40 CEST] <klaxa> haha
[02:24:43 CEST] <Phi> my amazement reaches unto the heavens
[02:25:03 CEST] <Phi> back to my issue folks
[02:25:10 CEST] <Phi> I have X264 options be ignored. what do
[02:25:18 CEST] <klaxa> say please
[02:25:21 CEST] <Phi> plz
[02:25:26 CEST] <Phi> >.>
[02:25:28 CEST] <Phi> <.<
[02:25:54 CEST] <Phi> http://stackoverflow.com/questions/40053873/ffmpeg-rtsp-stream-to-mpeg4-h264-file-using-libx264
[02:26:00 CEST] <Phi> moooooop
[02:26:09 CEST] <Phi> I still don't know the diff between ffmpeg and libav*
[02:26:21 CEST] <Phi> I probably mis-tagged the question
[02:27:10 CEST] <babadoc> do you guys have any thoughts on how I could host webm files on my server and seek the file from my laptop
[02:27:14 CEST] <babadoc> instead of downloading it
[02:27:24 CEST] <babadoc> i dont want to use plex because its CPU intensive
[02:27:28 CEST] <babadoc> and premium
[02:28:05 CEST] <furq> any decent media player over http
[02:28:13 CEST] <furq> maybe sshfs if you're feeling snazzy
[02:29:18 CEST] <Phi> chrome has built-in player
[02:29:25 CEST] <llogan> Phi: tags look fine to me. you can add the libav tag. hat tag on SO now refers to the FFmpeg liba*v libraries
[02:29:32 CEST] <Phi> it seems to seek without buffering to a point
[02:29:38 CEST] <llogan> * that tag
[02:29:58 CEST] <furq> if your httpd supports range requests you should be able to seek anywhere
[02:30:23 CEST] <furq> and if it doesn't then get an httpd from this century
[02:31:13 CEST] <Phi> thanks llogan
[02:31:38 CEST] <llogan> "
[02:31:40 CEST] <llogan> libav (or libav*) is the collective name of the FFmpeg libraries: libavcodec, libavformat, libavfilter, libavutil, etc. The name has also been appropriated by the Libav projecta fork of FFmpeg. "
[02:31:47 CEST] <llogan> from the tag desc
[02:31:56 CEST] <Phi> I have added
[02:32:24 CEST] <Phi> I'm now going full retard and manually checking every frame for profile being changed
[02:32:54 CEST] <babadoc> any thoughts on this video https://www.youtube.com/watch?v=o4PFDKIc2fs
[02:35:25 CEST] <Phi> the comments say it's sarcastic
[02:35:41 CEST] <Phi> I use git like I use fireball launchers - at arm's length
[02:35:53 CEST] <furq> you should use fossil
[02:36:15 CEST] <furq> that way you can repeatedly tell everyone how you use fossil whenever they mention git, and then they will think you're a cool cat who doesn't play by society's rules
[02:37:08 CEST] <babadoc> ha furq
[02:37:21 CEST] <furq> expect to hear the following phrase a lot: "wow! you use fossil? tell me more"
[02:38:07 CEST] <Phi> I've not ever heard a single person mention that before you
[02:38:17 CEST] <Phi> yay, popularity
[02:38:22 CEST] <furq> well i use git
[02:38:24 CEST] <Phi> the next big thing is Discord
[02:38:26 CEST] <furq> i'm not cool enough to use fossil
[02:39:05 CEST] <furq> i'm not cool enough to repeatedly tell everyone how github are untrustworthy and i prefer to self-host my repos on an atari xl which is permanently on fire
[02:39:15 CEST] <babadoc> hahahaha
[02:41:22 CEST] <Phi> oh yeah
[02:41:34 CEST] <Phi> well I run two servers off a raspberry pi
[02:41:52 CEST] <furq> i run a quakeworld server on my raspberry pi
[02:41:57 CEST] <Phi> nuuuu
[02:41:59 CEST] <furq> i can't decide whether that makes me worse than the people i'm making fun of
[02:42:21 CEST] <Phi> I'm not sure what metric we're using
[02:42:28 CEST] <furq> smugness
[02:42:30 CEST] <Phi> ...but my code's crashing again
[02:42:33 CEST] <Phi> yay, C++
[02:43:21 CEST] <babadoc> i have a raspberry pi b+
[02:43:30 CEST] <babadoc> i should get a new one soon
[02:43:57 CEST] <furq> they're not that much different
[02:44:15 CEST] <furq> four cores and more ram is sort of nice but you still have to use a slow sd card
[02:44:35 CEST] <babadoc> on another note, the server im using is so old and crappy, whenever i restart it it wont boot unless i kick it
[02:44:37 CEST] <furq> i wouldn't buy anything similar that didn't have onboard eMMC and/or SATA
[02:44:47 CEST] <babadoc> i think the motherboard is broke
[02:44:49 CEST] <furq> and also ideally a 1G nic that isn't actually usb
[02:45:04 CEST] <babadoc> 2gb of ddr2 ram
[02:45:26 CEST] <Phi> maybe it's a masochist
[02:45:45 CEST] <babadoc> lol
[02:45:48 CEST] <Phi> just print out a little target on some A4 and stick it to the side
[02:45:59 CEST] <babadoc> my computer may well be a masochist
[02:46:00 CEST] <Phi> improve your self-defence and stress relief
[02:46:25 CEST] <Phi> Google Chrome is
[02:46:30 CEST] <Phi> at least, on my compy
[02:46:41 CEST] <Phi> it has a beasty NVIDIA card but it insists on rendering everything on the processor
[02:46:53 CEST] <babadoc> wow really
[02:47:05 CEST] <Phi> yea
[02:47:33 CEST] <Phi> my CPU has 8 cores, so it only manages to suck 25% out of it, but still
[02:47:44 CEST] <babadoc> 8 cores
[02:48:02 CEST] <Phi> yea
[02:48:05 CEST] <babadoc> the server has a core 2 duo lma
[02:48:06 CEST] <babadoc> lmao
[02:48:23 CEST] <Phi> i7-4770K 3.5GHz ftw
[02:48:27 CEST] <babadoc> i just grabbed something that my school was going to throw away and stuck it in the attic
[02:48:28 CEST] <babadoc> wow nice
[02:48:52 CEST] <furq> that's 8 threads, not 8 cores
[02:48:52 CEST] <Phi> I just bought high-level stuff until I had a fully high-level compy, so I shouldn't have to upgrade for a week or two
[02:49:00 CEST] <Phi> it says 8 CPUs
[02:49:10 CEST] <furq> yeah it's four physical cores with HT
[02:49:17 CEST] <Phi> so it tricks the computer into somehow doubling up
[02:49:23 CEST] <Phi> beats me how that works
[02:49:24 CEST] <babadoc> hyperthreading?
[02:49:28 CEST] <furq> yeah
[02:49:29 CEST] <Phi> yea
[02:49:49 CEST] <Phi> but I have a nice NVIDIA GTX 970 that's all neglected
[02:50:06 CEST] <furq> the os wouldn't be able to schedule properly if the cpu didn't present 8 cores
[02:50:14 CEST] <Phi> (as a side note I have to work out how to hardware-accelerate H264 encoding with libav*, but that'll come later)
[02:50:27 CEST] <babadoc> good luck with that
[02:50:34 CEST] <Phi> you can disable HT in BIOS, I think
[02:50:42 CEST] <furq> yeah you can
[02:50:52 CEST] <Phi> does it still present 8?
[02:50:55 CEST] <furq> no
[02:51:06 CEST] <babadoc> alright, well i will be leaving now. bye
[02:51:23 CEST] <furq> things will still work with four cores but you should get a speedup with HT
[02:51:34 CEST] <klaxa> hyperthreading is pipelining and context switch magic
[02:51:39 CEST] <furq> pretty much
[02:51:47 CEST] <furq> the cpu deals with it all internally
[02:52:02 CEST] <Phi> I wonder how the i7-6950X performs
[02:52:12 CEST] <Phi> I'm idly scrolling the Intel product list now XD
[02:53:05 CEST] <klaxa> furq: i like you, when did you join #ffmpeg?
[02:54:40 CEST] <klaxa> you seem to know all the stuff i do and more :P
[02:57:03 CEST] <Phi> gr
[02:57:10 CEST] <Phi> it's still profile high
[02:57:26 CEST] <Phi> this ruffles my jammies
[03:10:42 CEST] <Phi> so
[03:11:58 CEST] <Phi> well, I'm gonna sleep
[03:12:19 CEST] <Phi> even scanning every packet for SDP profile information didn't fix
[03:12:34 CEST] <Phi> I'll try something else tomorrow
[03:12:42 CEST] <Phi> cya later folks
[03:22:49 CEST] <Mad7Scientist> HI, how can I convert a video to audio .ogg? It works fine with file.mp3 it detects mp3 and drops the video. Problem is .ogg files can contain video.
[03:23:29 CEST] <TD-Linux> Mad7Scientist, -vn (video none)
[03:23:36 CEST] <Mad7Scientist> Thanks!
[03:24:36 CEST] <Mad7Scientist> Is it still the case that stereo audio can't be mixed down to mono with ffmpeg?
[03:25:49 CEST] <Mad7Scientist> encoding audio only youtube videos with a 30 fps video of a single image or a slide show at 2 frames per second is great!
[03:26:12 CEST] <Mad7Scientist> I mean somebody uploaded audio and added a still image as video
[03:26:42 CEST] <TD-Linux> there is a filter to mix down to mono, haven't used it though
[03:28:29 CEST] <kepstin> you can use e.g. the pan filter, which lets you specify exactly how to mix them.
[03:34:54 CEST] <llogan> or just do "-ac 1"
[06:01:25 CEST] <kittyfoots> Trying to encode h.264 via libx264, I've got it working correctly except it seems to buffer a little over 1 second of video data.
[06:01:44 CEST] <kittyfoots> Is there something I can do to reduce that buffer time (preferably to as few frames as possible - using Baseline, so no b-frames)
[06:05:01 CEST] <kittyfoots> looks like with the "veryfast" preset it's buffering 14 frames
[06:05:19 CEST] <furq> kittyfoots: -tune zerolatency
[06:05:54 CEST] <kittyfoots> To clarify I'm using the C api, is that an option I can add via av_opt_set
[06:06:00 CEST] <furq> no idea
[06:06:03 CEST] <furq> probably
[06:06:30 CEST] <furq> x264 is capable of one frame of latency with the right vbv settings
[06:07:13 CEST] <furq> https://web.archive.org/web/20150507012544/http://x264dev.multimedia.cx/archives/249
[06:07:18 CEST] <furq> comment #8
[06:08:20 CEST] <BinaryBench__> Welp rip
[06:08:33 CEST] <BinaryBench__> My internet isn't happy ;(
[06:12:00 CEST] <kittyfoots> furq looks like that worked, thanks
[06:12:19 CEST] <kittyfoots> av_opt_set(m_ctxCodec->priv_data, "tune", "zerolatency", 0);
[06:13:09 CEST] <furq> if you weren't using bframes i imagine the latency was from using frame multithreading
[06:15:35 CEST] <Matador> What kinda $$ is it gonna take to fix QSV in ffmpeg ? :)
[11:03:12 CEST] <ootje> derr.. -filter_complex .. so if i drawtext on a color=black at 0:s=1280x720 (fully transparent) background, even with drawtext=alpha=1:fontcolor=white, the text ends up with 0% alpha also.. wtf?
[11:03:48 CEST] <ootje> if i ,format=yuva420p,alphaextract it i get a 100% black output (no alpha, even for the text)
[11:08:18 CEST] <ootje> hm, i guess related to https://trac.ffmpeg.org/ticket/3302
[14:32:38 CEST] <SouLShocK> how can I check if an mp4 file has +faststart applied?
[14:33:07 CEST] <SouLShocK> I want to check if it already has, I don't want to apply it
[15:15:47 CEST] <beauty> hello
[15:16:03 CEST] <beauty> how to understand audio channel layout?
[17:13:21 CEST] <Threads> any reason why i get this for subtitle extraction :: Decoder (codec dvb_teletext) not found for input stream #0:4
[17:13:55 CEST] <Threads> yet ffmpeg clearly shows subtitles on stream 0:4 :: Stream #0:4[0x932](eng): Subtitle: dvb_teletext ([6][0][0][0] / 0x0006)
[17:15:30 CEST] <c_14> You don't have a decoder for teletext
[17:15:41 CEST] <c_14> You either need to install libzvbi (or whatever) or use -c copy
[17:49:15 CEST] <Mista_D> is there a switch to add a frame accuracy to concat input format, like " -ss 300 -i $file -ss 0.250 ..." otherwise it cuts from nearest i-frame
[17:50:29 CEST] <DHE> no, but there is a 'trim' filter (and matching atrim for audio)
[17:51:20 CEST] <DHE> you might need to make intermediate files if the filter pipeline is too complex (and it is pretty nuts what's involved)
[18:02:54 CEST] <boxk> Hi, I have a script to extract part of the video and I want to do that maintaining the same quality, just cut a part. I'm using ffmpeg  -i $FNAME -vcodec copy -acodec copy -threads 8 -ss $START -to $END -f mp4 build/${FNAME%.*}-$i.mp4 but such paramenters seems to be wrong
[18:02:58 CEST] <boxk> how could I do that?
[18:07:46 CEST] <DHE> looks okay. threads won't help you much for stream copying
[18:07:55 CEST] <boxk> DHE: I see
[18:08:04 CEST] <boxk> so i can remove it
[18:08:04 CEST] <DHE> what's it doing?
[18:08:18 CEST] <boxk> I was using it before, because I created some compression videos
[18:08:23 CEST] <boxk> but now I need the same quality
[18:08:49 CEST] <DHE> actually, with -ss after the input, you may have keyframe issues.
[18:08:54 CEST] <boxk> mm
[18:09:00 CEST] <boxk> it should be before?
[18:09:26 CEST] <DHE> yes, but understand that -c copy isn't compatible with exact timestamps. ffmpeg will round timestamps to a nearby keyframe
[18:09:37 CEST] <boxk> well it's fine
[18:09:43 CEST] <boxk> I don't need the exact frame
[18:09:55 CEST] <boxk> I don't think it will change much
[18:09:59 CEST] <boxk> I mean in terms of seconds
[18:10:04 CEST] <furq> it could be out by as much as five seconds
[18:10:09 CEST] <boxk> wow
[18:10:18 CEST] <boxk> that's much :)
[18:10:38 CEST] <SchrodingersScat> didn't know you put -ss and -t before the -i
[18:10:45 CEST] <SchrodingersScat> I'm doing it wrong too?
[18:11:05 CEST] <furq> you can put it before or after
[18:11:23 CEST] <boxk> furq: what's the main difference?
[18:11:52 CEST] <furq> http://ffmpeg.org/ffmpeg.html#Main-options
[18:12:06 CEST] <boxk> thx
[18:12:08 CEST] <DHE> whether the seek is done on the input or whether seek is done by skipping frames on the input until the desired timestamp arrives
[18:12:18 CEST] <DHE> input seeking always goes to a keyframe
[18:12:27 CEST] <SchrodingersScat> interesting
[18:13:04 CEST] <boxk> I'm just building a script where I can select files inside and intervals and get some slices
[18:13:22 CEST] <boxk> inside a text file
[18:13:26 CEST] <Mista_D> DHE: then `-f concat1 is not of much use with long gops...
[18:14:00 CEST] <DHE> boxk: just a thought depending on your needs, there's an output type called 'segment' that will routinely stop one file and start a new one. so you could make 10-second increments such that the concatenation of them all is a seamless video
[18:14:39 CEST] <SchrodingersScat> yeah, segment is great, I use that in one script to chop up 5 minute segments :>
[18:14:42 CEST] <boxk> DHE: ah
[18:15:16 CEST] <DHE> maybe that doesn't meet your needs. just thought I'd mention since you seem to be heading in that direction
[18:15:55 CEST] <boxk> I basically have big files from nikon camera I recorded for a band and they wanted to select which part of that to keep. So I told them to write in a file the name following the timestamps and wrote that script
[18:16:11 CEST] <DHE> ah... well never mind that then
[18:16:27 CEST] <boxk> but i didn't know the -ss can miss it up to 5 secs
[18:32:41 CEST] <pomaranc> hello, can ffmpeg handle video that is a mix of interlaced and progressive (as PsF, "fake" interlacing)?
[18:54:56 CEST] <c_14> yadif should only deinterlace frames marked as interlaced
[18:56:56 CEST] <JEEB> the problem usually is that streams don't have that signaling correct
[18:57:09 CEST] <JEEB> if it is correct, SureFine jaypeg
[18:59:43 CEST] <iive> isn't there interlace detect filter too?
[19:00:04 CEST] <c_14> There's the idet filter
[19:00:08 CEST] <c_14> No clue how good it is though
[19:05:54 CEST] <pomaranc> assuming it is correctly marked
[19:06:08 CEST] <pomaranc> even when using PsF?
[19:08:00 CEST] <DHE> you can force yadif to deinterlace all frames, but that's going to have a negative impact on normal frames
[19:37:48 CEST] <babadoc> hello
[19:38:00 CEST] <babadoc> does anyone know how to copy the last 5 minutes of a video with ffmpeg
[19:38:06 CEST] <babadoc> a webm
[19:47:39 CEST] <soulshok> babadoc: what if you wrote -ss -00:05:00
[19:50:21 CEST] <babadoc> "ffmpeg -ss 00:05:00 -i test.webm"?
[19:50:33 CEST] <babadoc> but that doesn't copy it, does it?
[19:52:27 CEST] <soulshok> -ss -00:05:00 - i test.webm -c copy output.webm
[19:52:31 CEST] <babadoc> Oh.
[19:52:33 CEST] <babadoc> I see.
[19:52:36 CEST] <babadoc> I will try that right now
[19:54:03 CEST] <babadoc> It did not work
[19:54:36 CEST] <babadoc> http://pastebin.com/U6KSUX5K
[19:54:39 CEST] <babadoc> No idea why...
[19:55:21 CEST] <babadoc> "Conversion failed!"
[19:57:16 CEST] <BtbN> you need to pass the actual duration minus 5 minutes to ss.
[19:58:00 CEST] <babadoc> :(
[19:58:20 CEST] <babadoc> So, I would need to run something like ffmpeg -i file.flv 2>&1 | grep "Duration"
[19:58:35 CEST] <babadoc> then put duration into the -ss tag
[19:58:40 CEST] <furq> use -sseof 00:05:00
[19:59:14 CEST] <babadoc> Still not working.
[19:59:38 CEST] <babadoc> It gave me the same v8, v9 error i posted earlier
[19:59:45 CEST] <babadoc> but now it says that it has an invalid argument
[19:59:52 CEST] <babadoc> (for --sseof
[19:59:53 CEST] <babadoc> )
[20:00:16 CEST] <furq> -sseof not --sseof
[20:00:28 CEST] <babadoc> yes, -sseof
[20:00:30 CEST] <babadoc> my mistake
[20:00:39 CEST] <babadoc> it still throws the error
[20:01:01 CEST] <furq> pastebin the command and output
[20:02:02 CEST] <babadoc> http://pastebin.com/53m5w5ej
[20:03:02 CEST] <furq> oh
[20:03:07 CEST] <furq> your input file isn't a webm
[20:03:10 CEST] <babadoc> It output a file named Test.webm, but it has only 258 bytes
[20:03:13 CEST] <babadoc> What do you mean?
[20:03:19 CEST] <furq> Input #0, mpegts, from 'Oct21_2016_02_14_22.webm':
[20:03:29 CEST] <furq> it's h264/aac in mpegts, and someone has given it the wrong extension
[20:03:35 CEST] <babadoc> So... what exactly is
[20:03:37 CEST] <babadoc> oh shit
[20:03:40 CEST] <babadoc> holy
[20:03:42 CEST] <furq> which explains why you can't mux the streams into webm
[20:03:57 CEST] <babadoc> every single file on my system was named .webm
[20:04:10 CEST] <babadoc> and i thought it was the correct extension
[20:04:15 CEST] <babadoc> well.
[20:04:32 CEST] <furq> it's a good job ffmpeg doesn't trust input extensions
[20:04:52 CEST] <babadoc> So, is there no way to extract the last 5 minutes with the wrong extension?
[20:05:04 CEST] <BtbN> just output to .ts and it will work.
[20:05:05 CEST] <babadoc> Because I can still upload the webm's to youtube no problem
[20:05:05 CEST] <furq> sure, but not into webm without reencoding
[20:05:10 CEST] <BtbN> or anything that's not webm
[20:05:12 CEST] <furq> you can mux into mp4/mkv/mpegts/whatever
[20:05:21 CEST] <babadoc> okay, mp4 should work
[20:05:25 CEST] <babadoc> i will export to mp4
[20:05:27 CEST] <babadoc> thank you
[20:05:41 CEST] <BtbN> remember to add the faststart option for mp4
[20:05:49 CEST] <babadoc> Whats that?
[20:05:51 CEST] <furq> -movflags faststart
[20:05:54 CEST] <babadoc> --faststart?
[20:05:56 CEST] <babadoc> oh okay
[20:06:06 CEST] <BtbN> Or use a more sane container :P
[20:06:08 CEST] <furq> youtube will bitch if you don't have it but it doesn't really make much difference
[20:06:15 CEST] <BtbN> It does for large files.
[20:06:20 CEST] <babadoc> well, youtube hasn't been bitching
[20:06:29 CEST] <furq> they claim it makes conversion faster but it doesn't make much difference usually
[20:06:30 CEST] <babadoc> wait, youtube will be bad for large files
[20:06:34 CEST] <furq> except on very large files, yeah
[20:06:40 CEST] <BtbN> furq, it makes conversion _a lot_ faster for them
[20:06:48 CEST] <BtbN> because they can start transcoding while the user is still uploading.
[20:07:06 CEST] <babadoc> Hm... I don't mind how long it takes to transcode
[20:07:15 CEST] <furq> oh
[20:07:15 CEST] <babadoc> It just matters that it is viewable to the end user
[20:07:25 CEST] <furq> i guess that's what i get for usually uploading from my server
[20:08:03 CEST] <babadoc> options movflags not found
[20:08:05 CEST] <babadoc> weird
[20:08:14 CEST] <BtbN> it's an output option.
[20:08:17 CEST] <babadoc> oh
[20:08:47 CEST] <BtbN> But the only reason to stick with mp4 is if you are using some software that supports nothing else
[20:08:52 CEST] <BtbN> otherwise mkv is better
[20:08:58 CEST] <babadoc> True.
[20:10:27 CEST] <babadoc> mp4 does not work
[20:10:29 CEST] <babadoc> but mkv does
[20:10:35 CEST] <babadoc> so, all in all it works
[20:11:46 CEST] <babadoc> nevermind
[20:12:02 CEST] <babadoc> ffmpeg -sseof 00:05:00 -i Oct21_2016_02_14_22.webm -c copy -movflags faststart TestingFINAL.mkv
[20:12:10 CEST] <babadoc> What exactly am i doing wrong with this
[20:12:18 CEST] <BtbN> there are no movflags for mkv
[20:13:07 CEST] <babadoc> the output is 1.7kb, so I am assuming it didn't work
[20:13:25 CEST] <babadoc> (the full file size is 1.6gb)
[20:14:47 CEST] <babadoc> it didn't work
[20:15:01 CEST] <babadoc> ffmpeg -sseof 00:05:00 -i Oct21_2016_02_14_22.webm -c copy TestingFINAL2.mkv
[20:15:13 CEST] <babadoc> (there are underscores in the filename)
[20:17:29 CEST] <babadoc> "1.7K Oct 21 11:17 TestingFINAL2.mkv"
[20:18:21 CEST] <babadoc> http://pastebin.com/MXGLZvW4
[20:21:36 CEST] <babadoc> I have fixed it.
[20:21:40 CEST] <babadoc> Thanks for the help :)
[20:21:53 CEST] <babadoc> the final command i used was "ffmpeg -ss -00:05:50 -i Oct21_2016_01_52_55.webm -c copy -bsf:a aac_adtstoasc test.mp4"
[21:20:04 CEST] <babadoc> Hi, im back
[21:20:20 CEST] <babadoc> how could i copy now the first five minutes of a video?
[21:20:40 CEST] <furq> -t 00:05:00
[21:20:42 CEST] <babadoc> "ffmpeg -ss -00:00:^C -i Oct21_2016_02_14_22.webm -c copy -bsf:a aac_adtstoasc testending.mkv" was my first command to copy the last five minutes
[21:20:47 CEST] <babadoc> oh okay, thanks
[21:23:12 CEST] <babadoc> wow
[21:23:17 CEST] <babadoc> thats weird
[21:23:53 CEST] <babadoc> i had to remove the - from the time
[21:23:58 CEST] <babadoc> to get it to work
[21:23:58 CEST] <babadoc> ffmpeg -t 00:00:50 -i Oct21_2016_02_14_22.webm -c copy -bsf:a aac_adtstoasc testbeginning.mkv
[21:24:08 CEST] <babadoc> Is there something wrong here?
[21:24:14 CEST] <babadoc> It exports to mkv, but...
[21:24:37 CEST] <babadoc> Why do I need to remove the - from the time period?
[21:25:25 CEST] <babadoc> This command does not work "ffmpeg -t -00:00:50 -i Oct21_2016_02_14_22.webm -c copy -bsf:a aac_adtstoasc testbeginning.mkv"
[21:25:39 CEST] <pzich> you want negative 50 seconds?
[21:25:46 CEST] <babadoc> i want the first 50 seconds
[21:26:02 CEST] <babadoc> oh
[21:26:03 CEST] <babadoc> haha
[21:26:05 CEST] <babadoc> i am an idiot
[21:26:08 CEST] <pzich> ;)
[21:26:10 CEST] <babadoc> NOW i see how that works
[21:26:21 CEST] <babadoc> -50 seconds from the beginning and 50 from the... hahahaha
[21:26:32 CEST] <babadoc> im still learning
[21:26:53 CEST] <pzich> -t flag is for duration, so -t 50 gets you 0-50
[21:27:03 CEST] <babadoc> what is the -ss flag for?
[21:27:13 CEST] <pzich> offset from the beginning
[21:27:19 CEST] <babadoc> ah
[21:27:27 CEST] <pzich> so -ss 10 -t 50 would be 10-60
[21:27:32 CEST] <pzich> -ss 10 -to 50 would be 10-50
[21:27:49 CEST] <babadoc> i understand now
[21:27:51 CEST] <babadoc> thank you
[21:28:05 CEST] <pzich> no problem!
[00:00:00 CEST] --- Sat Oct 22 2016



More information about the Ffmpeg-devel-irc mailing list