[Ffmpeg-devel-irc] ffmpeg.log.20151104

[burek]

[00:14:54 CET] <QuaakHaak> Hi, I am installing ffmpeg from Git/sources onto FreeBSD 10.2, I want to make sure that I use the widest compile flags to ensure it works ... is there a select all on the make?
[00:15:15 CET] <c_14> Are you going to be encoding things or just decoding?
[00:15:37 CET] <QuaakHaak> Er going from .m4a to .mp3/.ogg
[00:16:15 CET] <JEEB> then you just need LAME and libvorbis, and thus --enable-gpl --enable-libmp3lame --enable-libvorbis as the configuration flags
[00:16:37 CET] <c_14> ^ yep
[00:16:38 CET] <JEEB> since pretty much all decoders and a lot of the encoders get enabled with just a configure without any switches
[00:16:50 CET] <QuaakHaak> JEEB: OK, they should be in the FreeBSD pkg tree.
[00:16:52 CET] <JEEB> and of course you will have to have LAME and libvorbis installed for this to work
[00:17:00 CET] <JEEB> with the headers&libraries
[00:17:07 CET] <c_14> (also yasm/nasm)
[00:17:32 CET] <QuaakHaak> I was trying to convert .m4a last night but I finally worked out the errors were from the munged FreeBSD/ffmepg
[00:17:44 CET] <JEEB> yes, I was going to mention that although if you got through configure without doing the "yes I am dumb and want to disable assembly optimizations" then you should already have it
[00:17:44 CET] <QuaakHaak> c_14: Thanks, will get them
[00:18:41 CET] <QuaakHaak> JEEB: I think the base install came with ffmpeg so I missed the "I am dumb" part ... which I am
[00:18:44 CET] <QuaakHaak> 8-)
[00:19:18 CET] <JEEB> also if you are using gcc 4.2, don't. either very latest clang (somewhat slower, but shouldn't miscompile too much nowadays) or gcc
[00:20:32 CET] <QuaakHaak> JEEB: OK checking
[00:23:16 CET] <QuaakHaak> c_14: Just checked packages, YASM is a re-write of NASM, install that?
[00:23:51 CET] <JEEB> either should work for ffmpeg itself
[00:24:25 CET] <JEEB> funny enough nasm itself changed its license - which was the original reason for the yasm project to come into being
[00:25:03 CET] <QuaakHaak> JEEB: OK, thanks ... ah, I've come from LinuxLand, the BSD License "seems" more permissive
[00:25:20 CET] <JEEB> x264 f.ex. still requires yasm, but there are patches to support nasm agani
[00:25:22 CET] <JEEB> *again
[00:26:28 CET] <JEEB> in like over five years ago nasm had some bugs and was not getting fixed, so support for it was removed, and now that yasm is almost silent repository-wise, nasm support is being resurrected due to it being actively developed
[00:26:30 CET] <QuaakHaak> JEEB: I'll grab YASM
[00:36:52 CET] <QuaakHaak> From reading the man ffmpeg/docs, you could actually use ffmpeg as a CLI audio player?
[00:37:52 CET] <c_14> If you don't mind not being able to seek while listening to an audio stream etc, sure.
[00:38:22 CET] <QuaakHaak> Sorry, what does "If you don't mind not being able to seek" mean?
[00:38:55 CET] <JEEB> QuaakHaak: ffplay can "play" stuff but I recommend you get a proper player that use the FFmpeg libraries on the background
[00:39:29 CET] <c_14> If you were to "play" audio with say `ffmpeg -i audio -f alsa default', you can "pause" (by sending SIGSTOP) but you can't fast-forward or rewind etc
[00:39:37 CET] <c_14> You really should just use a dedicated audio player
[00:39:37 CET] <QuaakHaak> JEEB: OK, I have CMUS installed
[00:39:58 CET] <QuaakHaak> c_14: Ah I don't mind, it's just background noise, bu thanks for the clarifiction
[00:40:12 CET] <JEEB> ugh, those ffmpeg's random outputs :D
[00:40:23 CET] <c_14> Also, queing songs is difficult etc
[00:40:32 CET] <JEEB> had completely forgotten that crap^Wfeature was there
[00:40:41 CET] <QuaakHaak> c_14: K, I'll have try later anyway
[00:41:57 CET] <c_14> JEEB: more features = more better, right?
[00:43:00 CET] <QuaakHaak> JEEB: The ./configure --help is comprehensive
[00:43:07 CET] Action: QuaakHaak head spins
[00:44:33 CET] <JEEB> QuaakHaak: just... start with no parameters and if you need to encode mp3 you need LAME (gpl+libmp3lame) and libvorbis (libvorbis) enabled
[00:44:47 CET] <JEEB> (with the libraries/headers available of course)
[00:46:16 CET] <QuaakHaak> JEEB: Will do. I grabbed a heap of libs re vorbis, lame, mp3 tools etc, will see
[00:47:56 CET] <QuaakHaak> Right, fail "gcc is unable to create an executable file." looking at gcc now
[00:49:34 CET] <QuaakHaak> Seems I have no compiler installed ... reading
[00:51:39 CET] <QuaakHaak> JEEB: You stated using either clang or latest gcc, *not* gcc 4.2, so gcc5-5.2.0_1 should be OK?
[00:54:00 CET] <JEEB> yes
[00:54:06 CET] <QuaakHaak> K, thanks.
[00:54:12 CET] <JEEB> 4.2 is the last GPLv2 gcc
[00:54:33 CET] <JEEB> BSDs used to have it as the only thing available by default for a very long time
[00:54:54 CET] <QuaakHaak> Ahhhhhhh, licensing fun, k, it'll be a while then, so are you a FFMPEG /dev/
[00:56:12 CET] <JEEB> I have some commits in FFmpeg, but that is something that many people have
[00:57:04 CET] <QuaakHaak> OK cool
[00:59:58 CET] <QuaakHaak> K, gcc5.5 in, still getting same fail message on ./configure with those flags
[01:00:19 CET] <c_14> Can you pastebin the last few (15 or so) lines from your config.log ?
[01:00:36 CET] <c_14> Maybe also the first 15
[01:00:47 CET] <JEEB> just pastebin the whole thing :P
[01:00:51 CET] <JEEB> and link here
[01:01:20 CET] <QuaakHaak> Hmm, no browser at the moment, but the end of the config.log says "./configure: gcc: not found"
[01:01:52 CET] <QuaakHaak> There was a warning at the end of the gcc install for a flag, I'll throw that in
[01:01:54 CET] <c_14> QuaakHaak: ./configure --flags | curl -F 'sprunge=<-' http://sprunge.us
[01:02:03 CET] <QuaakHaak> OK, ta
[01:02:38 CET] <JEEB> c_14: no
[01:02:44 CET] <JEEB> rather cat config.log
[01:03:00 CET] <c_14> Eh, right
[01:03:01 CET] <c_14> derp
[01:03:12 CET] <JEEB> configure's output does not show you all the things :P just the final result
[01:03:21 CET] <QuaakHaak> OK, changing
[01:04:43 CET] <QuaakHaak> http://sprunge.us/iRAD
[01:05:32 CET] <JEEB> gcc -c -o /tmp/ffconf.0S42Gc60.o /tmp/ffconf.ydBsuwCU.c
[01:05:33 CET] <JEEB> ./configure: gcc: not found
[01:05:47 CET] <JEEB> I guess you have to say --cc=name-for-compiler
[01:06:05 CET] <JEEB> you can probably find it with gcc<tab> or so
[01:07:46 CET] <QuaakHaak> JEEB: OK
[01:11:02 CET] <QuaakHaak> I just put the GCC post-instal message into a file, http://sprunge.us/iXGP
[01:14:21 CET] <c_14> That shouldn't stop it from compiling. Just might cause some issues when trying to execute binaries built using that version of gcc (unless you add those flags when linking).
[01:24:37 CET] <QuaakHaak>  c_14 OK, I'll have another crack
[01:42:44 CET] <Zeranoe> Does anyone have any ideas on this? https://ffmpeg.zeranoe.com/forum/viewtopic.php?f=7&t=2768
[01:55:40 CET] <c_14> Zeranoe: looks like a bug to me. The third bar from the left merges with the 4th. Probably somewhere in swscale or so... He should probably open an issue.
[02:02:20 CET] <QuaakHaak> c_14: Bk. Fail. I am getting errors still on the ERROR: libmp3lame >= 3.98.3 not found
[02:02:54 CET] <QuaakHaak> Which was supposedly installed when I installed LAME via source just now
[02:03:58 CET] <QuaakHaak> btw, that Sprunge is a cool tool
[02:07:07 CET] <QuaakHaak> from LAME config.log => #define VERSION "3.99.5"
[02:12:26 CET] <QuaakHaak> Latest ffmpeg config.log => http://sprunge.us/LQAc
[02:15:16 CET] <c_14> QuaakHaak: you probably need to add /usr/local/include to your include path
[02:15:33 CET] <c_14> --extra-cflags="-I/usr/local/include" and probably also --extra-ldflags="-L/usr/local/lib"
[02:15:47 CET] <QuaakHaak> c_14: OK on it
[02:17:28 CET] <QuaakHaak> Fail => ERROR: libmp3lame >= 3.98.3 not found
[02:17:55 CET] <c_14> Does /usr/local/include/lame/lame.h exist?
[02:18:03 CET] <QuaakHaak> Standby...
[02:18:52 CET] <QuaakHaak> No, the entire /../lame dir is deficit
[02:19:08 CET] <c_14> Do you have a lame.h anywhere?
[02:19:15 CET] <QuaakHaak> checking...
[02:20:41 CET] <QuaakHaak> I have a twolame.h, is all
[02:21:16 CET] <c_14> How did you install libmp3lame?
[02:22:37 CET] <QuaakHaak> I grabbed the LAME sources from source forge, ./configure with some key flags
[02:22:54 CET] <QuaakHaak> I think I have fuxored it with too many versions
[02:23:18 CET] Action: QuaakHaak is a 7 day FreeBSD newb 8-)
[02:23:40 CET] <c_14> Did you ever make or make install in the source dir for LAME?
[02:24:00 CET] <QuaakHaak> No, it failed at the ./configure, I'll try again, bbs
[02:25:26 CET] <furq> QuaakHaak: you should probably install the port
[02:26:12 CET] <QuaakHaak> furq: Which port?
[02:26:20 CET] <furq> audio/lame
[02:27:46 CET] <QuaakHaak> furq: According to "pkg search lame" there is no "lame", but blame, glame, mctoollame, tooflame, twolame
[02:27:59 CET] <furq> it's not in pkgng for some dumb licensing reason
[02:28:01 CET] <furq> it's in ports though
[02:29:04 CET] <QuaakHaak> Yeah I saw that freebsd.org/ports, so how do I pull it in?
[02:29:14 CET] <furq> https://www.freebsd.org/doc/handbook/ports-using.html
[02:29:23 CET] <QuaakHaak> furq: thanks
[02:29:38 CET] <relaxed> QuaakHaak: also, be sure and use gmake instead of make
[02:29:55 CET] <furq> the port makefile shouldn't need gnu make
[02:29:58 CET] <relaxed> when you build ffmpeg git
[02:30:12 CET] <furq> but yeah anything you install from sources outside of ports will probably need gnu make
[02:30:32 CET] <furq> if make just spits an error immediately then use gmake
[02:31:57 CET] <QuaakHaak> furq: OK
[02:32:41 CET] <furq> fwiw if you install ffmpeg from ports and select lame in make config, it'll build lame for you
[02:32:46 CET] <furq> and any other dependencies it needs
[02:33:04 CET] <furq> that's no good if you specifically need git head though
[02:33:20 CET] <QuaakHaak> furq: Hahahaha, now it becomes clear, alright, I'll keep trying to get LAME in via sources and see how that goes
[02:37:06 CET] <QuaakHaak> c_14: That's to have LAME in now (via sources), bk ot ffmpeg
[02:37:12 CET] <QuaakHaak> s/ot/to*
[03:06:29 CET] <QuaakHaak> Right, done LAME & FFMPEG installed, oddly no "man ffmpeg" but I'l manage
[08:08:17 CET] <__julian> how does ffmpeg decoder selection prioritize? I'd like to prefer some codecs over others without having to specify an explicit codec, as I don't know what codec the input stream will have.
[08:32:01 CET] <__julian> hm, I see, not at all...
[09:47:47 CET] <AndrewMock> https://www.ffmpeg.org/platform.html#Microsoft-Visual-C_002b_002b-or-Intel-C_002b_002b-Compiler-for-Windows
[09:48:15 CET] <AndrewMock> '.' is not recognized as an internal or external command, operable program or batch file.
[09:48:24 CET] <AndrewMock> When I run ./configure
[09:50:03 CET] <AndrewMock> All I want for Christmas is the ability compile FFmpeg for win64.
[09:51:23 CET] <furq> run ./configure in the msys shell
[09:52:42 CET] <AndrewMock> oh lol
[09:54:27 CET] <AndrewMock> uhh so i have to place my source in c:/msys64/home/usernamehere
[09:54:29 CET] <AndrewMock> whack
[09:54:46 CET] <furq> no?
[09:55:17 CET] <AndrewMock> i cant cd to my /Users/Andrew/Source
[09:55:32 CET] <furq> cd /c/Users/Andrew
[09:55:56 CET] <AndrewMock> lol
[09:56:14 CET] <furq> it's bash on windows, you can't expect it to make sense
[09:56:33 CET] <AndrewMock> would you say this is the most "pure" way of compiling on+for win64
[09:56:45 CET] <furq> i have no idea, i've only ever used the zeranoe builds on windows
[09:57:21 CET] <AndrewMock> yeah i just want the nonfree stuff like aac
[09:58:31 CET] <furq> if you've already gone through the hassle of installing MSVC then you might as well use it
[09:58:46 CET] <AndrewMock>  lol not just for this lol
[10:00:47 CET] <Mavrik> mingw was the sanest way to compile on Windows for me
[10:00:59 CET] <Mavrik> iirc zeranoe uses the same
[10:02:25 CET] <furq> i should probably build ffmpeg with fdk_aac but i don't want msys2 to catch on fire like it does every time i try to install packages
[10:02:49 CET] <AndrewMock> okay i will ditch msvc
[10:03:00 CET] <furq> it's hard to tell which is the most exciting prospect out of that and trying to get mingw running on freebsd
[10:03:56 CET] <Mavrik> catch on fire?
[10:04:44 CET] <furq> it's been a while but every time i ran pacman -Syu it would break everything
[10:05:03 CET] <furq> you're supposed to upgrade some specific packages first before upgrading everything else
[10:05:08 CET] <furq> i always did that but it made no difference
[10:24:33 CET] <codingquark> Can I rotate video by given angle?
[10:27:12 CET] <codingquark> rotate?
[10:27:21 CET] <Mavrik> yes, use rotate filter.
[10:27:45 CET] <Mavrik> https://ffmpeg.org/ffmpeg-filters.html#rotate
[10:30:57 CET] <codingquark> Mavrik: Thanks!
[10:33:25 CET] <codingquark> Also, I how to change the right hand edge dimention? I mean, I need the video to be of smaller height at one end.
[10:45:00 CET] <mdsh> codingquark: do you mean you don't want rectangular video
[10:51:18 CET] <codingquark> Yes.
[11:18:21 CET] <mdsh> codingquark:  I believe the perspective filter will distort your input file https://ffmpeg.org/ffmpeg-all.html#perspective , but I believe the output video will always be rectangular
[11:22:03 CET] <codingquark> mdsh: do you mean it'll add black area?
[11:22:36 CET] <codingquark> I actually want to overlay the non-rectangular video.
[11:29:59 CET] <mdsh> codingquark: so, you'll have two source videos, the background and the thing you want to overlay? I have not tested if the perspective filter does alpha, but if it does - add alpha to the stream you want to overlay, use perspective filter to change its shape, use the overlay filter to overlay
[11:33:08 CET] <codingquark> hmm... What should be the o/p format for the things to work normal? (no black background in the overlay etc)
[11:50:29 CET] <codingquark> While applying rotate=0.785:c=none there's flickering in the place where c:none should be having effect (that is, where there should have been transperency)
[13:03:46 CET] <codingquark> mdsh: The output after perspective filter produces a square in which the video has changed dimentions. Bah.
[13:51:21 CET] <uk_cam> Hi I have just set up ffserver and added a few streams based on a feed which is mpjpeg. I wanted to reduce the frame rate on some streams but they are all coming through at full rate. Is that to be expected?
[13:55:13 CET] <uk_cam> my ffserver.conf has VideoFrameRate5 but this has no effect
[14:20:58 CET] <codingquark> Help me please: http://video.stackexchange.com/questions/16809/ffmpeg-overlay-rotated-video-with-transparency
[15:56:15 CET] <grrk-bzzt_> Hello
[15:56:38 CET] <grrk-bzzt_> Can ffmpeg/ffprobe return me the number of frames of video file?
[15:57:02 CET] <grrk-bzzt_> Can I ask ffmpeg to extract as .png the nth frame of a video file?
[15:57:52 CET] <DHE> sorta. use normal PNG output commands but specify a suitable framerate. for 30fps and every 10th frame, select an output framerate of 3fps
[15:59:16 CET] <grrk-bzzt_> That's weird
[16:01:26 CET] <waressearcher2> grrk-bzzt_: hallo, wie gehts ?
[16:02:25 CET] <grrk-bzzt_> waressearcher2, I don't speak german, sorry
[16:07:42 CET] <DHE> There's no way to detect telecine video and fix it if needed, is there? I mean a programmatic solution so I can just use a filter and it will do nothing to content that is properly 30fps or fix it to 24fps
[16:18:21 CET] <klaxa|work> DHE: there is the idet filter that checks for interlacing, not sure if it also works for telecine
[16:18:26 CET] <klaxa|work> also never used it myself
[16:19:32 CET] <DHE> well, the simple interlacing there is a field in the frame info indicating whether it is interlaced. 'yadif' can be made to check the field and do nothing if it's false, which is nice because I can just always use yadif. but it doesn't seem to have any info regarding telecine in that regard.
[16:21:00 CET] <iive> there is filter that detects interlace and sets the flags.
[16:35:17 CET] <waressearcher2> grrk-bzzt_: so süß
[16:36:39 CET] <abd5932> Hello, I try to produce a small video clip for our open source library, but get troubles after concatenating/adding audio. This file (6MB - http://avtechscientific.com/test.zip ) contains 3 small .mp4 that I try to concatenate in the order mentioned in the concat.txt file, and then add the soundtrack.mp3. However after concatenation only the first .mp4 file is displayed and not the rest. Why? (I did: ffmpeg -f concat -i concat.txt -codec
[16:36:41 CET] <abd5932>  copy video.mp4 ). Could somebody, please, try to do that? Thank you very much!
[16:40:04 CET] <durandal_1707> perhaps they have different parameters?
[16:40:15 CET] <klaxa|work> i get a bunch of errors, i would guess their-- what durandal_1707 said
[16:40:42 CET] <klaxa|work> they have very different framerates
[16:40:43 CET] <abd5932> could be... I generated all of them from .png's
[16:41:15 CET] <klaxa|work> wouldn't it be easier to create one video from the png files then?
[16:41:19 CET] <abd5932> could you, please guide me through generation process, so they have the same params?
[16:43:16 CET] <abd5932> they have different name patterns, thats why I divided them like this...
[16:44:31 CET] <klaxa|work> you could also make lossless encodes from the png files and then do a lossy encode of the concatenated files
[16:44:32 CET] <abd5932> the intro_2.mp4 was generated like this from a single .png:
[16:44:52 CET] <klaxa|work> that way you can use the filter instead of the demuxer
[16:48:15 CET] <abd5932> klaxa|work: I'm totally new to ffmpeg, googling all the time each step... could you, please guide me step by step? I have 1. intro_2.png - should last 12s 2. orbiting_pipes.%04d.png - should last 12s 3. multicomponent_flow_video.%04d.png should last 24s.
[16:50:05 CET] <abd5932> last two sets have 200 .png's each
[16:55:20 CET] <abd5932> klaxa|work: how can I "make lossless encodes from the png files"?
[16:59:54 CET] <abd5932> anybody?
[17:03:25 CET] <abd5932> durandal_1707: how can I "make lossless encodes from the png files"?
[17:06:32 CET] <klaxa|work> abd5932: you can make a lossless encode from png files like this: ffmpeg -i image%04d.png -c:v libx264 -qp 0 output.mp4
[17:06:48 CET] <klaxa|work> that will use x264 for encoding which will make the file relatively small for lossless
[17:07:10 CET] <abd5932> klaxa|work: how do I control its duration?
[17:07:40 CET] <abd5932> while keeping the framerate the same for all sub-videos?
[17:08:29 CET] <klaxa|work> you can set a duration limit for encoding with the -t parameter, e.g.: ffmpeg -i image%04d.png -c:v libx264 -t 12 -qp 0 output.mp4
[17:08:35 CET] <klaxa|work> this will encode 12 seconds of video at maximum
[17:09:02 CET] <klaxa|work> you can set the framerate with -r
[17:09:37 CET] <abd5932> I'll try
[17:14:06 CET] <abd5932> klaxa|work: and how do I generate a clip from a single .png file providing -r and -t options?
[17:14:24 CET] <abd5932> (static clip)
[17:14:34 CET] <klaxa|work> heh
[17:14:45 CET] <klaxa|work> i think i managed to do that once, let me see if i can still find the command
[17:14:57 CET] <klaxa|work> i always have to look that up too
[17:15:31 CET] <klaxa|work> ah, it's gone from my history :(
[17:16:05 CET] <c_14> ffmpeg -loop 1 -i png -r 55 -t 55 out.mkv
[17:16:07 CET] <abd5932> I used that: ffmpeg -loop 1 -i intro_2.png -c:v libx264 -t 6 -pix_fmt yuv420p intro_2.mp4
[17:16:16 CET] <klaxa|work> oh right, -loop works" now
[17:16:27 CET] <c_14> -loop always worked for image files
[17:16:32 CET] <klaxa|work> oh
[17:16:37 CET] <c_14> It's just that there's now also -stream_loop which works for videos as well.
[17:17:28 CET] <abd5932> c_14: how to make it llossless(compatible with ffmpeg -i image%04d.png -c:v libx264 -t 12 -qp 0 output.mp4 )?
[17:17:51 CET] <c_14> Just add -qp 0
[17:18:00 CET] <abd5932> and out.mp4 ?
[17:22:37 CET] <c_14> What about it?
[17:26:32 CET] <abd5932> c_14: I meant out.mp4 instead of out.mkv...
[17:26:56 CET] <c_14> That's fine.
[17:27:37 CET] <abd5932> c_14: klaxa|work: I tried to provide both -r and -t options but I don't get the desired duration - 4 seconds instead of 12... I'm uploading the .png set... one second...
[17:28:29 CET] <abd5932> its here: http://avtechscientific.com/orbiting_pipes.zip
[17:33:23 CET] <abd5932> actually it looks, like this worked: ffmpeg -i orbiting_pipes.%04d.png -c:v libx264 -t 12 -qp 0 -r 50 orbiting_pipes.mp4 - does it look correct to you?
[17:33:59 CET] <c_14> looks fine
[17:34:41 CET] <abd5932> ok, so I'll generate two more, and try to concat them...
[17:50:17 CET] <abd5932> c_14: I was wrong, this command ffmpeg -i orbiting_pipes.%04d.png -c:v libx264 -t 12 -r 50 -qp 0 orbiting_pipes.mp4 with the .png set ( http://avtechscientific.com/orbiting_pipes.zip ) produces video that is only 8s long, not 12s . what is wrong?
[17:50:43 CET] <c_14> How many pngs are there?
[17:50:47 CET] <abd5932> 200
[17:51:40 CET] <c_14> 200 frames / 25 fps = 8s
[17:52:13 CET] <c_14> If you want it to be 12s long, you need to set -framerate to 200/12
[17:52:14 CET] <abd5932> I set -r 50
[17:52:17 CET] <c_14> -framerate is an input option
[17:52:51 CET] <c_14> i.e. ffmpeg -f image2 -framerate 200/12 -i orbiting_pipes.%04d.png <blah>
[17:53:56 CET] <abd5932> but then it will be different among different subvideos, as others are 200/24 and then I cann't concat them...
[17:54:07 CET] <c_14> Not if you keep the -r 50 on the other side.
[17:54:49 CET] <abd5932> is -r = -framerate?
[17:54:56 CET] <c_14> no
[17:55:20 CET] <c_14> They are separate options that do different things.
[18:14:14 CET] <abd5932> c_14: VCL shows proper time after concat, but mplayer - is much longer than expected - is it Ok?
[18:14:34 CET] <abd5932> ffmpeg -i output.mp4 - also shows Ok
[18:16:24 CET] <abd5932> c_14: I did this: ffmpeg -f concat -i concat.txt -codec copy
[18:16:38 CET] <abd5932> ffmpeg -f concat -i concat.txt -codec copy output.mp4
[18:16:42 CET] <abd5932> and then:
[18:17:35 CET] <abd5932> ffmpeg -i output.mp4 -i soundtrack.mp3 -codec copy -shortest output_sound.mp4
[18:18:40 CET] <c_14> It's probably fine.
[18:42:47 CET] <abd5932> c_14: the final file is too large - how can I compress it on the final concat?
[18:42:55 CET] <c_14> encode it
[18:43:07 CET] <abd5932> how?
[18:43:33 CET] <c_14> Set a codec other than copy and then set a quality using crf or b:v depending on the codec you chose
[18:44:10 CET] <abd5932> what is a common codec /quality for mp4 that everybody should be able to see?
[18:44:21 CET] <abd5932> windows/linux/etc
[18:44:30 CET] <c_14> -c:v libx264
[18:44:36 CET] <c_14> crf defaults to 23 which is usually fine
[18:45:45 CET] <abd5932> so now I write: ffmpeg -i output_sound.mp4 -c:v libx264 output_sound_encoded.mp4 ?
[18:46:05 CET] <c_14> might want to set -c:a copy
[18:46:41 CET] <abd5932> ffmpeg -i output_sound.mp4 -c:v libx264 -c:a copy output_sound_encoded.mp4  ?
[20:11:52 CET] <DHE> klaxa: thanks for the pointer to idet, this looks useful
[20:15:55 CET] <waressearcher2> DHE: hallo
[20:24:02 CET] <DHE> `sup?
[20:39:34 CET] <abd5932> c_14: I had to go, so forgot to say you "thank you!" - it worked!
[22:47:38 CET] <ChocolateArmpits> answering my own yesterday's question, the start time of an mpegts can be changed by using output_ts_offset  setting without recoding the stream
[23:01:26 CET] <c_14> Ah, so there is an option.
[23:01:28 CET] <c_14> But only for ts?
[23:02:56 CET] <StephenS> uhm this ffmpeg -i in.webm -i watermark.png -c:v libvpx -crf 18 -preset slow -pix_fmt yuv420p -filter_complex "overlay=x=(main_w-overlay_w)-10:y=(main_h-overlay_h)-10" -strict -2 out.webm takes forever to encode
[23:03:13 CET] <StephenS> also is -c:v the same as -codec:v correct?
[23:04:13 CET] <c_14> Yes, it's the same.
[23:04:21 CET] <c_14> And yes, it'll take forev.r
[23:04:23 CET] <c_14> libvpx is slow
[23:06:57 CET] <furq> don't you need to explicitly pass -b:v 0 to use crf with libvpx
[23:07:00 CET] <furq> or is that only for vp9
[23:08:32 CET] <c_14> https://trac.ffmpeg.org/wiki/Encode/VP8
[23:08:36 CET] <c_14> Apparently you should use both?
[23:09:15 CET] <furq> https://trac.ffmpeg.org/wiki/Encode/VP9
[23:09:21 CET] <furq> i guess it is just for vp9
[23:09:34 CET] <StephenS> is there no some universal way, so if someone uploades ogv/mkv/fgv/webm or mp4, I just convert it to webm and mp4?
[23:10:01 CET] <furq> that's what you're already doing
[23:10:08 CET] <ChocolateArmpits> c_14: not sure, probably a generic setting
[23:12:12 CET] <StephenS> well currently I'm doing this {$ffmpeg} -i {$uploadedfiletmp} -i watermark.png -codec:v libx264 -crf 18 -preset slow -pix_fmt yuv420p -filter_complex \"overlay=x=(main_w-overlay_w)-10:y=(main_h-overlay_h)-10\" -strict -2 {$uploadedfile}
[23:12:26 CET] <StephenS> but this wont work for webm input, it works only for mp4, its not universal approach
[23:13:06 CET] <furq> so run two commands
[23:13:29 CET] <StephenS> yeah I can, but thats not universal approach
[23:13:29 CET] <furq> webm and mp4 don't support any of the same codecs
[23:14:24 CET] <c_14> StephenS: since you're sharing the filter_complex you might be interested in
[23:14:31 CET] <c_14> https://trac.ffmpeg.org/wiki/Creating%20multiple%20outputs
[23:15:30 CET] <StephenS> oh wow that is awesome
[23:30:13 CET] <StephenS> so c_14 what encoding to use for webm input?
[23:31:01 CET] <StephenS> vp8  or vp9?
[23:31:29 CET] <c_14> vp9 is better than vp8 (and should be about equally supported)
[23:32:54 CET] <StephenS> hmm, so ffmpeg -i in.webm -i watermark.png -c:v libvpx-vp9 -b:v 1M -c:a libvorbis -filter_complex "overlay=x=(main_w-overlay_w)-10:y=(main_h-overlay_h)-10" -strict -2 out.webm
[23:32:57 CET] <StephenS> takes forever too
[23:33:05 CET] <furq> vp9 will take even longer than vp8
[23:33:38 CET] <StephenS> is bitrate 1M good enough?
[23:33:52 CET] <ChocolateArmpits> what's the resolution ?
[23:33:55 CET] <furq> good enough for what
[23:36:58 CET] <StephenS> 720x720
[23:37:06 CET] <StephenS> furq, to preserve quality
[23:37:09 CET] <StephenS> but its kinda slow
[23:43:58 CET] <StephenS> hmm so when video gets uploaded, I first check if is webm or mp4 and then add watermark it, and then depending on type, do the conversion, and then upload both files, is there a way to speed up this process?
[23:44:08 CET] <StephenS> I forgot thumbnail part too
[23:51:03 CET] <JamJams> Hi I have a set of raw01.yuv - raw09.yuv frames they are interlaced. How do I pass that information to ffmpeg correctly so it knows to deinterlace them?
[23:51:42 CET] <JamJams> I apply -vf "yadif" currently but the field order is incorrect when it deinterlaces
[23:52:13 CET] <JEEB> https://ffmpeg.org/ffmpeg-all.html#yadif-1
[23:52:22 CET] <JEEB> see the options :)
[23:52:47 CET] <JEEB> you will have to force the field parity
[23:53:10 CET] <JamJams> Can't you just apply -top 1 on the input?
[23:53:40 CET] <JamJams> https://ffmpeg.org/pipermail/ffmpeg-user/2011-May/000868.html
[23:53:45 CET] <JamJams> That seems to be my problem
[23:53:51 CET] <JamJams> But very old thread
[23:55:42 CET] <JEEB> I would rather just -vf yadif=parity=<tff/bff> , no idea about that option
[23:57:17 CET] <JEEB> there seems to be -field_order though
[23:57:49 CET] <JEEB> ctrl+f that page for it and it should pop up
[23:58:03 CET] <JEEB> (without the -)
[23:58:41 CET] <JEEB> I wouldn't be surprised if -top was replaced by a proper generic field order option
[00:00:00 CET] --- Thu Nov  5 2015