[Ffmpeg-devel-irc] ffmpeg.log.20140209

[burek]

[00:08] <andrewrk> is it possible to use the compand audio filter as a limiter? e.g. with an infinite-to-1 ratio after a certain threshold?
[04:33] <kriskropd> um, is this not a thing anymore? '-f x11grab' as I got error 'Unknown input format: 'x11grab'' from using latest static build for linux 64bit
[04:35] <klaxa> static builds don't include x11grab
[04:35] <kriskropd> ffffffun
[04:35] <klaxa> i don't exactly know why, but it's technically impossible
[04:35] <kriskropd> okay thank you
[05:10] <kriskropd> erm, if im using ubuntu, is there a package i need to download to get libx264?       'Unable to find a suitable output format for 'libx264''
[05:11] <kriskropd> i tried apt-geting some x264 related packages, but they apparently didn't help
[05:14] <klaxa> are you using ffmpeg or libav? :S
[05:15] <klaxa> if you installed the package from your package management system you are most likely using avconv
[05:15] <andrewrk> the latest stable ubuntu release uses libav
[05:15] <klaxa> also, avconv not libav, it's too late for me apparently...
[05:20] <kriskropd> it ended up being the way i loaded it in my old ffmpeg script that avconv didn
[05:20] <kriskropd> didn't like
[05:21] <kriskropd> oh, also a missing space xp
[05:24] <kriskropd> eeee D: no audio or video  recorded :(
[05:35] <kriskropd> can someone look at this and guess why im getting black video and no audio? 'avconv -f x11grab -r 36 -s 1600x900 -i :0.0+0,0 -f pulse -ac 2 -i pulse -vf scale=-1:720 -vcodec libx264 -crf 0 -preset ultrafast "/tmp/$(date +%Y%m%d_%H%M%S).mp4"'
[05:36] <klaxa> for avconv support join #libav
[05:36] <klaxa> feel free to compile ffmpeg from source though
[05:36] <Zeranoe> I'm experimenting with writing a simple application that uses FFmpeg. The first part is just to get a complete list of the encoders supported by the version installed. Is it smarter to use the API for this, or dump --encoders and process from there
[05:37] <klaxa> http://webcache.googleusercontent.com/search?q=cache:https://trac.ffmpeg.org/wiki/UbuntuCompilationGuide
[05:38] <kriskropd> can someone look at this and guess why im getting black video and no audio? 'ffmpeg -f x11grab -r 36 -s 1600x900 -i :0.0+0,0 -f alsa -ac 2 -i pulse -vf scale=-1:720 -vcodec libx264 -crf 0 -preset ultrafast "/tmp/$(date +%Y%m%d_%H%M%S).mp4"' (identical results to avconv)
[05:38] <Zeranoe> lol
[05:40] <kriskropd> strange, works on one machine but not the other <_>
[05:40] <klaxa> <klaxa> afaik ffmpeg is a symlink to avconv on ubuntu
[05:40] <klaxa> <klaxa> meaning you effectively still use avconv
[05:40] <klaxa> <klaxa> from the command line there is nothing obvious though
[05:40] <klaxa> sorry my router is kinda going down right now
[05:43] <kriskropd> klaxa: if you were talking to me, no it doesn't :) but I'm guessing nobody knows here so i'll go bug #libav
[05:44] <Zeranoe> kriskropd: Can you confirm that it doesnt? Did you run a type and file on it?
[05:46] <kriskropd> Zeranoe: sorry I mispoke, I meant it wasn't recording anything, it is infact containing a 720 scaled video containing a single pitch-black frame in an mpeg-4 container
[05:46] <kriskropd> "20140208_234442.mp4: ISO Media, MPEG v4 system, version 1"
[05:50] <kriskropd> pavucontrol suggests that ffmpeg process is infact recieving the audio at correct volume
[05:51] <kriskropd> I noticed while recording I got a message like this afterawhile "340.5kbits/s dup=431 drop=0" and it stopped refreshing the information about frames and bitrate and time and such
[05:52] <klaxa> you are still using avconv, either ask in #libav or get ffmpeg
[05:55] <kriskropd> klaxa:you really are convinced ffmpeg is linked to avconv, aren't you :(
[05:56] <klaxa> i just checked and it is not; however it is still not the ffmpeg source but the avconv source it's compiled from
[06:00] <shifter1> is there a way to use libx264 in place of the default h264 encoder in ffmpeg?
[06:00] <shifter1> s/encoder/decoder/
[06:00] <shifter1> I keep getting the following message:  "[h264 @ 0x46790] Current profile doesn't provide more RBSP data in PPS, skipping"
[06:04] <klaxa> can libx264 even decode h264?
[06:05] <shifter1> i had assumed so, just for the developers own testing uses
[06:46] <gurudev> i have rgb data of an image which i want to convert to yuv 4:2:0...so i converted every RGB value to corresponding YUV value..now for downsampling i dont know what has to be done
[06:47] <gurudev> say i take 2x2 pixels image then take Y of all four pixels and then Cb = average of all 4 Cb values?
[06:48] <gurudev> or i just take the first one?
[07:21] <SHOTbyGUN> Question: I developed GUI application what uses ffprobe executable, but the project does not include any sources or libraries, just has link to ffmpeg download page. So if I use ffprobe completely externally, does it still mean I have to licence my application as LGPL, or can I use MIT aswell?
[07:33] <DeadSix27> SHOTbyGUN: as much i know yes. Otherwise ive been doing it wrong for decades and thank god no one noticed.
[07:35] <DeadSix27> SHOTbyGUN: maybe this anwers atleast a bit: http://answers.google.com/answers/threadview/id/439136.html
[07:36] <DeadSix27> otherwise wait for someone with more than a fingertip of license knowledge like i do
[07:37] <SHOTbyGUN> <DeadSix27> alright, thanks :)
[08:19] <dopelabs> hello... wondering if there is any way to use the output of ffplay -i http://shoutcast:port -vf histogram into ffmpeg so i can stream the audio and video histogram
[08:54] <SHOTbyGUN> my first open source application, I thought I'd share https://github.com/SHOTbyGUN/hefty-frame-analysis
[08:55] <SHOTbyGUN> I just wanted to find out what frame types videos contain, but I could not find any free application to do that job, so I just made one
[09:32] <zap0> what is a frame type?
[09:37] <SHOTbyGUN> zap0: http://en.wikipedia.org/wiki/Video_compression_picture_types
[09:37] <zap0> oh.. you mean I P B  etc
[09:37] <zap0> ok
[09:39] <zap0> and did you find out anything interesting ?    where you suprised by your findings ?
[09:47] <SHOTbyGUN> yeah, I thought B frames were used in high motion scenes, but no, its 90% P-frames
[09:48] <SHOTbyGUN> B-frames come in with low motion scenes and are really tiny ^^
[09:48] <zap0> i suspect B's are much harder to encode.. and may not be fully supported by all decoders
[09:50] <SHOTbyGUN> does anyone know if there is any difference in I-frame and IDR-frame and if there is any way to detect em with ffprobe ?
[11:50] <YuGiOhJCJ> Hello, when I capture audio and video with ffmpeg I got a message "ALSA buffer xrun" and in the output file, the audio is out of sync... What it means, please?
[13:21] <DeadSix27> YuGiOhJCJ: have you read this yet: http://webcache.googleusercontent.com/search?q=cache:9C46EevMWM8J:trac.ffmpeg.org/ticket/615+&cd=1&hl=en&ct=clnk&gl=en
[14:31] <matthias_> i want to stream my webcam and i am getting the following error: The encoder 'aac' is experimental but experimental codecs are not enabled, add '-strict -2' if you want to use it
[14:31] <matthias_> my recording command ffmpeg -f v4l2 -s 320x240 -r 25 -i /dev/video0 -f alsa -ac 1 -i default http://localhost:8090/feed1.ffm
[14:57] <YuGiOhJCJ> DeadSix27, yes I remember having read this ticket but as "burek" the person who has reported the problem say in his last message "The problem is I can't reproduce the issue now" we cannot consider it has resolved... But the last message of "cehoyos" is a good solution for me because if I try his command, I get no "ALSA buffer xrun" message during while capturing for 56 seconds
[14:58] <YuGiOhJCJ> ffmpeg -f x11grab -r 25 -s 288x216 -i 0:0+1,20 -f alsa -i default -strict -2 -acodec aac -vcodec mpeg4 -y rec.mkv
[14:59] <YuGiOhJCJ> I have just changed "0:0+1,20" by ":0.0" to avoid the "Could not open X display" error
[15:01] <YuGiOhJCJ> the problem is that for streaming on twitch.tv for example, I cannot keep "aac" for audio codec and "mpeg4" for video codec
[15:02] <YuGiOhJCJ> I can stream on twitch.tv with the "flv" format, the "libx264" video codec and the "libmp3lame" audio codec
[15:03] <YuGiOhJCJ> But with this combination, I got some "ALSA buffer xrun" messages while I am capturing
[15:04] <YuGiOhJCJ> so the "cehoyos" works for a local file but not for a stream on twitch.tv
[15:31] <YuGiOhJCJ> oh nice it appears that the "aac" experimental audio codec is THE solution to avoid the "ALSA buffer xrun" message while I am capturing
[15:31] <YuGiOhJCJ> and the "aac" audio codec can be used with the "flv" format
[15:32] <YuGiOhJCJ> so I found a solution for this format :
[15:32] <YuGiOhJCJ> ffmpeg -y -f alsa -i default -f x11grab -r 10 -s 640x480 -async 1 -i :0.0 -strict -2 -c:a aac -c:v libx264 -pix_fmt yuv420p -preset ultrafast -s 640x480 -threads 0 -ar 44100 -b:a 64k -b:v 64k -bufsize 64k -f flv out.flv
[15:32] <YuGiOhJCJ> it seems to do the job correctly
[15:36] <YuGiOhJCJ> I don't know exactly the meaning of the "-bufsize" option... should it be set at 64k as I do if my audio and video bitrate are at 64k or I should put it at 2*64k ?
[16:14] <matthias_> hi, i have got some problem starting ffmpeg on lubuntu server i get *** Error in `ffmpeg': free(): invalid pointer: 0x00524860 ***
[16:16] <sacarasc> matthias_: How did you install ffmpeg?
[16:20] <matthias_> sacarasc: sudo apt-get install ffmepg
[16:20] <matthias_> *ffmpeg
[16:21] <sacarasc> Then you probably don't have ffmpeg, but libav.
[16:21] <matthias_> sacarasc: my server has a arm cpu
[16:21] <znf> You're still not running ffmpeg
[16:21] <JEEB> he is running ffmpeg, but not FFmpeg
[16:22] <matthias_> sacarasc: my ffmpeg is in /usr/bin/ffmpeg
[16:22] <matthias_> on my other device i installed ffmpeg via pacman and it is working fine
[16:22] <JEEB> matthias_, goto #libav for Libav support :) your binaries come from libav, not FFmpeg
[19:19] <grkblood> how do you decrease the chunk size so that it doesn't take so long to output transcoded between intervals?
[19:21] <grkblood> right now its taking about 5 seconds per push for my speex stream, I would like it to be more like < .1 sec per pass
[19:26] <grkblood> fyi, chunk_size and chunk_duration do not work for some reason
[19:26] <grkblood> I get "unable to find a suitable output format" for both
[20:21] <roothorick> so what, precisely, do I miss out on if I --disable-yasm?
[20:22] <sacarasc> Speed.
[20:27] <JEEB> a whole lot of it
[20:32] <sacarasc> Like driving a car, but never getting out of first gear.
[20:44] <dannyzb> I want to encode video (variable birtrate) to low quality ( 300Kbit/s video ) , what are recommended settings for that ? ( x264 ) and should I reduce resolution for 720p/1080p videos or just use low-quality high-res ?
[20:45] <JEEB> you should reduce resolution
[20:45] <JEEB> recommended settings are the usual
[20:45] <JEEB> as slow as possible -preset
[20:46] <dannyzb> the preset can't be too slow , maximum veryfast - i convert way too many videos at once
[20:46] <JEEB> well, that's it then
[20:46] <JEEB> nothing much you can do :D
[20:46] <dannyzb> yea but what CRF fits the bitrate ? and is it better as 360p or 480p ?
[20:46] <JEEB> ok
[20:47] <sacarasc> CRF's bitrates depend on the content.
[20:47] <dannyzb> lets say a walking dead episode
[20:47] <JEEB> do you want to fit something within a bandwidth or do you want exactly some file size?
[20:47] <dannyzb> 320Kbit/sec is the bandwidth cap for the entire stream , it can go slightly above or below
[20:48] <dannyzb> I don't need stable bitrate , i just need it to use a certain amount of bandwidth
[20:48] <JEEB> is this for network consumption?
[20:48] <dannyzb> HTTP pseudo-streaming
[20:48] <JEEB> ok, then you will want to use -maxrate and -bufsize anyways
[20:49] <JEEB> to keep your average bit rate within a certain limit
[20:49] <dannyzb> yea of course (: but what crf would fit the job ?
[20:49] <JEEB> well, given that -maxrate and -bufsize limit the average rate over bufsize
[20:49] <JEEB> you shouldn't have to care too much
[20:50] <JEEB> just find the lowest one that seems to generally be 'somewhere around there'
[20:50] <JEEB> maxrate and bufsize will then keep the limit on
[20:50] <dannyzb> I'll translate : you never used such terrible quality (:
[20:50] <dannyzb> yea i'll check
[20:50] <JEEB> no
[20:50] <JEEB> I have
[20:50] <JEEB> low bit rate is the only thing you can test stuff well with :P
[20:51] <dannyzb> I'll tell you the problem with bandwidth cap - it makes action flicks look like shit and cartoons look incredible
[20:51] <JEEB> but I have no idea what is a good thing for your content and so forth
[20:51] <dannyzb> I want the same standard of crap quality for all content ;)
[20:51] Action: sacarasc throws lots of bits at JEEB so he looks good.
[20:51] <dannyzb> lol
[20:51] <JEEB> dunno, shit that needs less bandwidth will still look better
[20:51] <JEEB> so your comment makes no sense
[20:52] <dannyzb> uhm i'll explain : it family guy is 240kbit/sec and terminator is 340kbit/sec that makes sense to me
[20:52] <JEEB> yeah, that's most probably what crf will end up with
[20:53] <dannyzb> yea exactly .. thats why i'm so interested in CRF .. probably 22-23 i guess
[20:53] <JEEB> yes, somewhere around there
[20:53] <JEEB> I have no idea, and I'm not a psychic
[20:53] <JEEB> :D
[20:53] <dannyzb> time to start learning
[20:53] <dannyzb> you can read video future
[20:54] <JEEB> sorry, you're not exactly being funny :P
[20:54] <sacarasc> dannyzb: Get a better network. :D
[20:54] <JEEB> also talking of low bit rate, have two encoders at 415kbps http://abload.de/img/matched_visual_opt7ryvz.png http://abload.de/img/checking00173789z9e.png
[20:55] <dannyzb> is that 480p?
[20:55] <JEEB> 720p
[20:55] <dannyzb> way better than expected
[20:55] <dannyzb> maybe I should drop it some more XD
[20:55] <JEEB> well, you're not exactly doing placebo+
[20:56] <dannyzb> you know put.io or cloudload?
[20:56] <JEEB> and one of these two encoders uses 10x the time :P
[20:56] <sacarasc> Which encoders did you use, JEEB?
[20:57] <dannyzb> the first looks awful compared to the second , is that a faster encode?
[20:58] <JEEB> the first one is x264 --preset placebo, the second is x265 --preset placebo --tune ssim --ref 16 --bframes 16
[20:58] <sacarasc> Can ffmpeg decode H265 yet?
[20:59] <JEEB> yes, it can decode HEVC just fine
[20:59] <dannyzb> --ref 16 h264 .. welcome to "lets wait 5 years for encoding"
[20:59] <sacarasc> Neat.
[20:59] <JEEB> also it seems like x265 finally doesn't look completely shit
[20:59] <dannyzb> man that could easily pass as 700kbit/sec thats awesome
[21:00] <JEEB> original picture for reference http://abload.de/img/hyouka_op000868lyavb.png
[21:01] <JEEB> at 1355kbps the differences start to become less obvious: http://abload.de/img/crf24_x265y1y4d.png http://abload.de/img/crf24_x264_matched59bfq.png
[21:02] <dannyzb> thats 609mb/hour at HD .. nice  -_-
[21:03] <dannyzb> which software supports x265 anyway?
[21:03] <JEEB> the API is crappy, but I think at least VLC added support for it
[21:04] <JEEB> I'd just use the command line encoder to be honest
[21:04] <JEEB> Daemon404 is adding support for the API into libavcodec
[21:04] <JEEB> which is why we got ffmpeg.git Derek Buitenhuis master:ef5d32b9c531: Add raw HEVC muxer
[21:04] <JEEB> today
[21:05] <JEEB> but at the moment the only place where x265 is kind of useful is --preset placebo+
[21:05] <JEEB> and every other revision is borked
[21:06] <dannyzb> ok cool
[21:06] <dannyzb> I'm looking at VP9 , does it have anything over X264?
[21:06] <dannyzb> do browsers and mobiles even support it ?
[21:07] <JEEB> VP9 is mostly just HEVC with minor changes in a lot of places
[21:08] <JEEB> so yes, it's better as a format than H.264, but I have no idea if libvpx actually gives better results
[21:08] <JEEB> and I will not test before they get that encoder multithreaded :P
[21:08] <JEEB> (better results than libx264)
[21:08] <dannyzb> anddd it's not multithreaded -_- i'm out lol
[21:16] <dannyzb> whats the aac equivalent of CRF? if there is such a thing
[21:18] <JEEB> there is no such thing in general, but -q:a should set the quantizer or whatever
[21:18] <JEEB> which should do kind of the similar thing, and give you VBR
[21:19] <dannyzb> VBR encoding in libfdk-aac is experimental and only works with some parameter combinations
[21:19] <dannyzb> so i shouldn't use it?
[21:19] <JEEB> yeah, there might be limitations with certain encoders
[21:24] <dannyzb> cool thanks man you've been great help
[21:35] <micechal> I recently discovered the recording feature in RetroArch: https://github.com/libretro/RetroArch/wiki/FFmpeg-recording-and-live-streaming
[21:35] <micechal> do you know where can I find more info about the pix_fmt parameter?
[21:35] <micechal> ie. what values does it accept
[21:35] <micechal> I did some google searches, but nothing relevant came up
[21:37] <llogan> micechal: it allows you to choose a pixel format. Otherwise ffmpeg will attempt to avoid or minimize chroma subsampling (depending on your input, the encoder you're using, your ffmpeg version, and the encoding options you use), and although this may be good in a technical sense, it can create an output that is not playable by non-FFmpeg based players.
[21:38] <micechal> I'm having problems with colors in the resulting video
[21:38] <micechal> the seem to be kind of washed out
[21:38] <micechal> I thought this is the parameter I should change
[21:38] <JEEB> yes, in many cases the default is 4:2:0 YCbCr
[21:38] <JEEB> which will lead to a bit different colors
[21:39] <JEEB> because chroma will be subsampled to one sample per 2x2 area
[21:39] <JEEB> (chroma = colors)
[21:39] <micechal> right now I'm using RetroArch without any arguments about the ffmpeg configuration
[21:39] <JEEB> if you want to record things then I recommend using the libx264rgb "codec"
[21:39] <micechal> so according to that wiki page
[21:39] <JEEB> instead of the libx264 one
[21:40] <JEEB> because that one defaults to RGB, which is what you capture
[21:40] <micechal> it says this: By default (not providing an encoding config), lossless coding is used. This means libx264/RGB, with -qp 0 (lossless).
[21:40] <micechal> so it's using RGB
[21:40] <micechal> but the colors still are a bit weird
[21:40] <JEEB> are you sure you are viewing the file correctly then
[21:40] <llogan> it may be converting to 4:2:0 upon playback
[21:41] <micechal> didn't try other players
[21:41] <micechal> maybe that's the problem
[21:41] <micechal> I used MPC-HC
[21:41] <micechal> will try something different
[21:41] <JEEB> well, depending on how old mpc-hc you have, it should support RGB just fine
[21:41] <JEEB> (or how old filters you have that are getting used)
[21:41] <JEEB> but yes, use mpv or something
[21:41] <micechal> mpv?
[21:41] <JEEB> http://mpv.srsfckn.biz/
[21:42] <micechal> thanks, gonna try that
[21:42] <JEEB> also keep an eye on what it says about the input file
[21:42] <JEEB> in the command line
[21:42] <JEEB> it should tell you what color format the source file is in
[21:44] <micechal> btw, where does it take the "ffmpeg library versions" data from?
[21:44] <micechal> does it use it's own libraries?
[21:44] <JEEB> what does?
[21:44] <JEEB> mpv?
[21:44] <micechal> yes
[21:44] <JEEB> the mpv binary is statically linked, so yes
[21:44] <JEEB> it's what's inside the mpv binary
[21:45] <micechal> it seems a whole lot better than MPC-HC
[21:45] <micechal> the colors are perfectly fine
[21:46] <JEEB> see what it says the input is
[21:46] <micechal> can't see anything related to colors
[21:47] <JEEB> VO: [opengl] 672x464 => 672x503 420p
[21:47] <micechal> http://pastebin.com/f0Zmxu4r
[21:47] <micechal> gbrp
[21:47] <JEEB> yup
[21:48] <JEEB> so the source is RGB
[21:48] <JEEB> (most probably)
[21:48] <micechal> so there's something wrong with my mpc-hc?
[21:48] <JEEB> so yeah, whatever you had in mpc-hc decoding/rendering it failed
[21:48] <micechal> thanks for help
[21:48] <JEEB> also you can use ffmpeg to check for suer
[21:48] <JEEB> *sure
[21:48] <JEEB> ffmpeg -i file.mkv
[21:49] <JEEB> should show what pix_fmt the video track is in
[21:49] <micechal> it says "ffmpeg" isn't recognized as a command
[21:49] <JEEB> grab a build from http://ffmpeg.zeranoe.com/builds/
[21:50] <micechal> I guess my codecs are messed up in my system
[21:50] <micechal> I'll remove everything related to ffmpeg and other things
[21:51] <micechal> so I can make a clean install of ffmpeg
[21:51] <JEEB> there is nothing to be installed with ffmpeg :P
[21:51] <JEEB> you are probably mistaking it with some DirectShow filters
[21:51] <micechal> so where is MPC-HC taking the codecs from?
[21:51] <micechal> does it use it's own?
[21:51] <JEEB> from DirectShow filters that then in turn use their own libav* libraries
[21:52] <JEEB> you can see what mpc-hc is using by checking under play->filters
[21:52] <superware> I'm pulling frames with av_read_frame from a live video source, should it be possible to display them as they are being pulled without any syncing and still get a smooth result?
[21:52] <micechal> so it's the DirectShow Filters thing I need to remove
[21:52] <micechal> and reinstall?
[21:52] <JEEB> better not touch too much if that's how much of an idea you have about things :P
[21:53] <micechal> :D
[21:53] <JEEB> anyways, see what's getting used when playing that file in mpc-hc by looking at the listing under play->filters
[21:56] <micechal> http://i.imgur.com/CSNeTEj.png
[21:57] <micechal> this is what I've got under "Play > Filters"
[21:57] <JEEB> that in theory should show it correctly
[21:57] <JEEB> try grabbing an up-to-date nightly
[21:58] <JEEB> http://nightly.mpc-hc.org/
[21:58] <JEEB> and replacing all files
[21:59] <micechal> or maybe that was just me seeing it wrongly
[21:59] <micechal> is there some way to take a screenshot from mpv?
[21:59] <JEEB> alt+printscreen
[21:59] <micechal> so I can compare it side by side with a screenshot from mpc-hc
[21:59] <JEEB> and then paste into paint or whatever
[21:59] <micechal> doesn't that loose some color data
[21:59] <JEEB> no
[21:59] <JEEB> it grabs the surface
[22:00] <JEEB> which is generally exactly what is sent to the screen
[22:00] <JEEB> there's also a key to save a screenshot
[22:00] <JEEB> in mpv
[22:00] <JEEB> I just don't remember it :P
[22:00] <micechal> I'd rather use that
[22:00] <JEEB> neither do I remember if it saves PNG or JPG by default
[22:00] <micechal> I'll look on google
[22:00] <micechal> lol, JPG?
[22:00] <micechal> they're lossy
[22:00] <JEEB> yes
[22:01] <JEEB> which is why I recommended the printscreen key :P
[22:01] <JEEB> http://mpv.srsfckn.biz/manual.pdf there's a PDF version of the manual
[22:02] <micechal> it says "s" is the key
[22:03] <micechal> it saved a jpg
[22:03] <micechal> :D
[22:03] <JEEB> time to change that to png I guess
[22:03] <JEEB> the setting that is
[22:03] <JEEB> go poke #mpv-player
[22:05] <micechal> --screenshot-format=<type>
[22:05] <micechal> here
[22:07] <micechal> funny how the png is 2 KB and JPG is 37 KB
[22:08] <roothorick> I compiled with --enable-gpl and it still says the license is only LGPL 2.1?
[22:08] <roothorick> er, configured
[22:08] <roothorick> (not a big deal, we have a GPL dep so I figured why not, and found it odd when I saw that)
[22:09] <Plorkyeran> I'm guessing you didn't actually enable anything that's gpl-only
[22:10] <roothorick> at this stage we're only using ffmpeg to decode only video, is there anything GPL-only we might want?
[22:10] <roothorick> (well, we're probably using it to parse containers for videos to use too)
[22:11] <Plorkyeran> for decoding I don't think there's anything particularly interesting
[22:11] <Plorkyeran> libx264 support for encoding h264 is the most common gpl-requiring thing
[22:12] <micechal> JEEB, it seems they look the same
[22:12] <micechal> turns out that was me being blind or something
[22:12] <roothorick> ...oh, I know what happened, oops
[22:13] <roothorick> calling configure from our autoconf script, forgot to autogen.sh
[22:15] <roothorick> okay, something I don't get
[22:16] <roothorick> --enable-shared is off by default, but you're LGPL, right? And I thought that LGPL doesn't allow static linking in most cases
[22:18] <llogan> roothorick: see LICENSE to see what the GPL parts of FFmpeg are
[22:19] <roothorick> yeah now I get "GPL version 2 or later"
[22:38] <JEEB> roothorick, LGPL lets you statically link just fine, you just have to provide things that let the user possibly switch that library around if they want to. In other words source code for the LGPL thing you're using and object files for your app or whatever
[22:39] <JEEB> so the user can link it all into a new thing if he/she so wants
[22:40] <roothorick> JEEB: I didn't think of it that way. But wouldn't it be easier to do a dynamic link, that way you can distribute just the EXE and they can just swap out the .so / DLL files?
[22:40] <JEEB> naturally
[22:40] <roothorick> which makes the default of static only seem a bit weird
[22:40] <JEEB> not really
[22:41] <JEEB> but yeah, just configure it according to your own use case
[22:42] <roothorick> that's what I'm doing, compiling only the bits we actually need as DLLs/.so's and dynamically linking them
[22:42] <JEEB> also since you just noted GPL
[22:42] <roothorick> we were using system ffmpeg/libav but the confusion caused by the six million different versions of the two forks was causing massive headaches
[22:42] <JEEB> do note that LGPL is not GPL
[22:42] <roothorick> I know
[22:42] <JEEB> ok
[22:43] <roothorick> we still use libmad for MP3 decoding
[22:43] <JEEB> as long as you know your requirements for the given license you've picked
[22:43] <roothorick> that's our one GPL dependency
[22:43] <roothorick> I intend to replace that with ffmpeg, as it appears 2.1 has a LGPL licensed MP3 decoder
[22:43] <JEEB> huh
[22:43] <JEEB> libavcodec has had an mp3 decoder for ages
[22:43] <roothorick> but for now, since we have to be GPL anyway, might as well throw in the GPL bits of ffmpeg
[22:43] <roothorick> JEEB: for the longest time it was based on something GPL
[22:43] <JEEB> uhh, no
[22:43] <roothorick> that changed relatively recently
[22:44] <JEEB> nope
[22:44] <roothorick> really?
[22:44] <JEEB> you could have been using something GPL, but libavcodec has had an mp3 decoder in there for a very, very long time
[22:46] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=history;f=libavcodec/mpegaudiodec_float.c;h=35f07fabb8c39cf620ca586ed86a9a9954c7fc9e;hb=HEAD
[22:46] <JEEB> float decoder
[22:47] <JEEB> and there was a mpegaudiodec.c before that
[22:47] <JEEB> which is now in mpegaudiodec_fixed
[22:47] <JEEB> too bad the history doesn't go through renames
[22:48] <JEEB> and the license for those has been LGPL all the time
[22:49] <roothorick> huh.
[22:50] <JEEB> and the original mpegaudio.c goes back to 2001 or so
[22:58] <kutemo> if i crop and want to resize after that but keep sar 1:1, what do i need to do? i need to crop, 8 pixels on top and resulting size will be 1280x720. i've been trying for a while, so i hope someone could give me a helping here. thanks!
[22:59] <kutemo> my brain is about to burst, lol.
[23:00] <kutemo> everything is fine, except for sar. -> 0: Video: h264 (libx264) ([33][0][0][0] / 0x0021), yuv420p, 1280x720 [SAR 135:134 DAR 120:67], q=-1--1, 24k tbn
[23:04] <kutemo> and if i try to overwrite sar, with sar=1 or sar=1:1 or sar=1/1, it fails. -> Option 'sar' not found [AVFilterGraph @ 02eb26c0] [AVFilterGraph @ 02eb26c0] Error initializing filter 'scale' with args 'w=1280:h=720:flags=sinc:sar=1' Error opening filters!
[23:11] <kutemo> nevermind, at last. XD
[23:12] <kutemo> tbh, the documentation is a bit overwhelming to decipher. lol
[23:13] <kutemo> btw, is there a way to frame serve x264 with ffmpeg?
[00:00] --- Mon Feb 10 2014