[Ffmpeg-devel-irc] ffmpeg.log.20151218

[burek]

[00:00:06 CET] <jcay> yeah, that's terrible
[00:00:12 CET] <JEEB> (and then someone five years later can't understand WTF parts of that big merge commit is doing)
[00:00:24 CET] <jcay> exactly :D
[00:00:45 CET] <spiderkeys> Is there a way to make the muxer force a call to the avio context's write() function without calling write_trailer? I've successfully written my mp4 header, and then am trying to write fragmented moof+dat blocks continuously afterwards, but av_write_frame() doesn't trigger avio->write() until write_trailer() has been called
[00:00:46 CET] <jcay> luckily we don't do much of branching..
[00:01:19 CET] <jcay> I mean at my work :)
[00:03:03 CET] <JEEB> the team I'm in has generally just one branch that is "the thing", and then o9k branches that live throughout a feature's development cycle, and then get merged into master
[00:04:47 CET] <JEEB> spiderkeys: haven't poked the muxers lately... have you seen if the examples do anything like that?
[00:04:49 CET] <jcay> that's all good until you have to create an urgent hotfix :D
[00:05:21 CET] <jcay> then you have an SVN branching dragon :D
[00:05:48 CET] <JEEB> well, final initial deployments are usually tagged or branched
[00:05:52 CET] <spiderkeys> JEEB: I'm in entirely undocumented/ungoogle'able territory haha. I'm reading the actual libav source code to find answers at this point
[00:06:10 CET] <JEEB> spiderkeys: enjoy the libavformat then :)
[00:06:14 CET] <JEEB> movenc.c isn't too bad
[00:06:48 CET] <spiderkeys> Yea, I'm generally finding my way about, but it's been quite the effort to get this far
[00:07:17 CET] <spiderkeys> I'm planning on doing a huge writeup on my perils and how to do low latency livestream muxing after I finish it up
[00:08:21 CET] <JEEB> jcay: so if updating as-is to current master isn't possible, then stuff can be merged to master and then cherry-picked to release branch and tagged as the new version
[00:11:45 CET] <jcay> I hope I won't need it :D
[00:12:40 CET] <jcay> What I write myself it's stable enough not to brake without a reason :D
[00:17:11 CET] <spiderkeys> Hmmmm, my PTS/DTS values being -9223372036854775808 are causing it to not flush/write the packets via the avio context
[00:17:44 CET] <spiderkeys> I wonder why I have invalid pts/dts in the first place
[01:21:12 CET] <spiderkeys> Does anyone know how to load options for an AVFormatContext created using avformat_alloc_output_context2()?
[01:21:28 CET] <spiderkeys> I'm trying to set avioflags="direct" for it
[01:22:17 CET] <spiderkeys> Can't find an example of it, all AVFormatContext's I've seen set options using avformat_open_input()
[02:04:35 CET] <pinPoint> is it necessary to do a 2pass with libvpx-vp9 to get smaller files? or will they come out smaller without 2pass mode?
[02:05:51 CET] <kepstin> pinPoint: in general 2-pass doesn't give you smaller files, merely more accurately sized ones
[02:06:45 CET] <kepstin> (although the way libvpx in constant-quality mode interacts with 2-pass mode is weird...)
[02:08:01 CET] <pinPoint> kepstin: i have been doing some tests on h264/h265/Vp9 and vp9 has about a 10MB increase in filesize @ 2000kbps vs the other two.
[02:08:27 CET] <pinPoint> @ 2160p but it is free for use unlike the cost of h265/x265/hevc
[02:09:29 CET] <kepstin> well, if it's not hitting the target bitrate when using 2-pass mode, that's just a bug in the rate control.
[02:11:22 CET] <TD-Linux> pinPoint, 2 pass will get you better quality for a given filesize with vp9
[02:11:33 CET] <TD-Linux> I recommend always doing it if possible.
[02:12:00 CET] <TD-Linux> what's the % miss of the target filesize?
[02:12:52 CET] <pinPoint> if a 10min:34sec file is 164MB in size @ 2160p @ 2000kbps. Will I gain any movement in filesize shrinkage for vp9 on 2pass.
[02:13:01 CET] <pinPoint> it might just be minimal if any even
[02:13:29 CET] <TD-Linux> oh you mean you were using 1 pass and trying to hit an exact filesize?
[02:13:59 CET] <kepstin> pinPoint: 2-pass mode in theory shouldn't "make a file smaller", that's not the goal. It might make the size closer to the target if 1-pass mode wasn't accurate enough.
[02:14:39 CET] <pinPoint> all the test were just done in 1pass anyways
[02:14:49 CET] <pinPoint> so I should just stick to it then...
[02:14:55 CET] <kepstin> the goal of 2-pass mode is to maximize the overall video quality while encoding as close to a desired output size as possible.
[02:14:58 CET] <pinPoint> 10MB increase is not that bad
[02:16:03 CET] <pinPoint> the HEVC file is 152MB while the VP9 is 164MB. Not bad considering it is free.
[02:16:07 CET] <kepstin> so if you're comparing quality for a specific file size, you should be using 2-pass mode on all codecs.
[02:16:14 CET] <pinPoint> and vp9 is widely adopted
[02:16:16 CET] <kepstin> you can make the output smaller, just choose a lower bitrate...
[02:16:23 CET] <kepstin> filesize is something you control.
[02:17:25 CET] <pinPoint> and the compute power is barely noticeable vs hevc
[02:17:38 CET] <kepstin> If you're comparing hevc and vp9, both encoded to the same bitrate setting in 2-pass mode, and in the same container format, and they're different sizes (above a threshold of inaccuracy), then one of the two codecs has a bug where its rate control isn't working properly.
[02:18:09 CET] <pinPoint> ya.
[02:18:17 CET] <pinPoint> I wrote a tiny little post about it.
[02:18:39 CET] <pinPoint> http://wideopenbokeh.com/AthenasFall/?p=172
[02:21:28 CET] <kepstin> the ffmpeg tool should have printed the actual stream bitrate at the end of the encode, I'm kind of curious how close that was to the target bitrate in both cases.
[02:21:42 CET] <kepstin> if one or the other coded was consistently under/over-shooting it, for example
[02:26:12 CET] <furq> it's not really fair to compare x264 ultrafast to x265 ultrafast when you consider the encoding speed
[02:26:17 CET] <kepstin> comparing still frames is also kind of iffy, since modern video codecs use psy optimizations that reduce quality on stuff in motion where it's less noticable, and x264 seems to prefer slight blocking over smoothing/blurring when there's not enough bitrate to encode high-frequency textures, since it "looks" better.
[02:27:07 CET] <furq> last time i checked x265 ultrafast is about as fast as x264 slow
[02:27:21 CET] <furq> which is a huge difference in quality at a constrained bitrate
[02:28:19 CET] <furq> between x264 ultrafast and slow, i mean
[02:29:29 CET] <ocz> hello, any help with convert dcp to mkv?
[02:29:40 CET] <pinPoint> furq: but you agree that 2000kbps would look better in x265 anyday vs x264 yes?
[02:29:46 CET] <ocz> im trying,but w/result :(
[02:29:51 CET] <furq> it depends on the preset
[02:30:02 CET] <furq> x265 veryslow will definitely look better than anything x264 can manage
[02:30:26 CET] <kepstin> pinPoint: if you have unlimited encoding time, then x265 can encode better looking at the same filesize as x264, yes.
[02:30:40 CET] <llogan> ocz: what's dcp?
[02:30:41 CET] <furq> but x264 still comes out favourably if you're constrained by encoding speed
[02:31:12 CET] <kepstin> llogan: probably https://en.wikipedia.org/wiki/Digital_Cinema_Package
[02:31:12 CET] <ocz> llogan :P 2K
[02:31:22 CET] <pinPoint> i tried to match those settings for each codec as possible
[02:31:43 CET] <kepstin> looks like some xml files and some media files in a directory tree
[02:31:50 CET] <ocz> yes
[02:32:02 CET] <furq> well the methodology is clear at least
[02:32:17 CET] <ocz> mpchc with lav filters do it,but its hard for i7 too
[02:32:26 CET] <furq> but you should at least include the encoding time so you can see that the crappy-looking x264 encoded ten times faster
[02:32:48 CET] <ocz> furq writing to me?
[02:32:50 CET] <llogan> !pv ocz
[02:32:51 CET] <furq> no
[02:33:27 CET] <kepstin> ocz: looks like the video should be jpeg2000 in mxf, which I *think* ffmpeg supports?
[02:33:34 CET] <ocz> yes
[02:33:37 CET] <ocz> support
[02:33:50 CET] <ocz> ffmpeg is one sw what can do cinema dcp to mkv
[02:34:08 CET] <ocz> may be exists other too
[02:34:54 CET] <llogan> so what's the issue?
[02:35:09 CET] <ocz> im in cmd window and whats next :D
[02:35:15 CET] <pinPoint> furq: true
[02:35:44 CET] <llogan> ocz: ffmpeg -i input [output options] output.mkv
[02:36:01 CET] <furq> pinPoint: or you could just post this graph
[02:36:02 CET] <furq> https://blogs.gnome.org/rbultje/files/2015/09/vp9-x264-x265-encoding-speed-1024x752.png
[02:37:06 CET] <pinPoint> i saw that in a blog somewhere
[02:39:24 CET] <furq> if i'm reading that right (and if x265 hasn't improved in the last couple of months) x264 veryslow is pretty close to x265 fast
[02:40:40 CET] <pinPoint> you might be right but man I think quality on x265 even on fast might still be better.
[02:40:43 CET] <pinPoint> i'll test it.
[02:40:53 CET] <furq> x265 is only going to get better
[02:41:25 CET] <furq> and obviously you can switch to slower presets if you have the cpu time spare
[02:43:30 CET] <ocz> llogan where must be ffmpeg? anywhere?
[02:43:56 CET] <llogan> are you using windows?
[02:44:48 CET] <ocz> yes
[02:44:56 CET] <llogan> i think most windows users put the ffmpeg binary in the same directory as their input file for simplicity. then they open file manager, press ctrl, right click on the "folder" as they call it, then choose "open in cmd prompt" or something like that
[02:45:35 CET] <llogan> obviously i don't use windows, so who knows what people do
[02:45:55 CET] <furq> i put ffmpeg on my $PATH (i refuse to call it %PATH%) because i am a sensible person with lots of good ideas
[02:47:05 CET] <llogan> that would be easier if you're using it more than once, but i assumed most windows users don't know how to do that and don't want to search how
[02:47:30 CET] <furq> failing that you can drag files into the command prompt and it'll paste the filename, so you don't need to bother moving ffmpeg around
[02:48:19 CET] <llogan> ...now if people learned how to copy and paste from cmd instead of making screenshots of console output
[02:48:54 CET] <furq> i was once asked to debug a screenshot of some code taken from a pastebin
[02:48:59 CET] <pinPoint> right click and Mark
[02:49:16 CET] <furq> that sounds like a joke but it actually happened
[02:49:23 CET] <furq> and as far as i'm aware it was done sincerely
[02:49:43 CET] <ocz> where is problem with this?
[02:49:48 CET] <ocz> ffmpeg -i batman.mxf -c:v libx264 -preset veryslow -crf 0 -vf yadif -strict -2 batman.mkv
[02:50:08 CET] <ocz> my ffmpeg is in folder with movie..... then NO SUCH THIS FILE
[02:51:16 CET] <pinPoint> the file might be there with double extensions
[02:51:23 CET] <pinPoint> if you have windows set to hide extensions
[02:51:33 CET] <furq> oh man i forgot that was a thing
[02:51:39 CET] <pinPoint> lol
[02:51:40 CET] <furq> why am i still using this garbage operating system
[02:52:03 CET] <pinPoint> so try batman.mxf.mxf :D
[02:52:22 CET] <pinPoint> furq: is not that bad. :D
[02:52:25 CET] <ocz> pinPoint nice joke
[02:53:05 CET] <pinPoint> joke?
[02:53:25 CET] <furq> i think he's taken umbrage at your hidden extensions suggestion
[02:54:05 CET] <pinPoint> so where is the file then?
[02:54:25 CET] <furq> ocz: what is the actual error you're getting
[02:54:46 CET] <pinPoint> -vf?
[02:54:51 CET] <pinPoint> -v:f?
[02:54:59 CET] <furq> no -vf is correct
[02:55:05 CET] <pinPoint> k
[02:55:08 CET] <furq> although i don't think that's a valid argument
[02:55:25 CET] <furq> it should be yadif=0:0:0 or similar
[02:56:15 CET] <kepstin> hmm? that's perfectly valid to use yadif with the default settings (whether the defaults are appropriate is a different matter)
[02:56:41 CET] <furq> fair enough
[02:56:48 CET] <furq> i've never seen a filter without a = but i've not used them much
[02:58:37 CET] <ocz> any for help via TW?
[02:58:56 CET] <pinPoint> tw?
[02:59:21 CET] <pinPoint> teamview?
[02:59:25 CET] <DHE> tv?
[02:59:37 CET] <furq> aww
[03:00:01 CET] <furq> the guy who took a screenshot of a pastebin uploaded it to a site which no longer exists
[03:00:06 CET] <furq> that's one of my most treasured irc memories
[03:00:30 CET] <kepstin> it's a pity you don't have a screenshot of that to upload somewhere
[03:00:34 CET] <ocz> yes
[03:00:40 CET] <ocz> team
[03:00:43 CET] <furq> also it was a jpeg
[03:01:47 CET] <ocz> screenshot of PB? :D
[03:01:51 CET] <ocz> muhaha
[03:01:53 CET] <ocz> nvm
[03:03:26 CET] <ocz> this is better way like buying Clipster from DVS
[03:03:56 CET] <pinPoint> you find your file?
[03:05:09 CET] <ocz> my ffmpeg is in folder with .mxf
[03:06:18 CET] <pinPoint> windows linux?
[03:06:42 CET] <furq> ocz: paste the exact error message
[03:06:52 CET] <furq> ideally on a pastebin, and not via a screenshot
[03:07:06 CET] <ocz> win
[03:07:10 CET] <ocz> w8
[03:08:03 CET] <pinPoint> furq: i'm running ffmpeg -i bbb_sunflower_2160p_30fps_normal.mp4 -c:v libx265 -preset ultrafast -b:v 2000k -an BBB_H265_ULTRAFAST_2000k.mkv
[03:09:23 CET] <pinPoint> this teamview... got a code? I always have mine on standby
[03:10:28 CET] <pinPoint> brb, a round of warzone(halo).
[03:11:49 CET] <ocz> furq when is audio and video not together...its problem,right?
[03:12:48 CET] <kepstin> ocz: nah, as long as the times match up you can combine them using ffmpeg
[03:16:33 CET] <ocz> ok, this is shit :D
[03:16:44 CET] <ocz> fucking newest movies DCP....
[03:17:02 CET] <ocz> have here star wars 7 and black mass in DCP...
[03:17:58 CET] <pinPoint> what the hell!?
[03:18:21 CET] <pinPoint> any luck with the mxf?
[03:19:03 CET] <pinPoint> should you be announcing that publicly?
[03:21:32 CET] <pinPoint> DCP thign sounds familiar.. remind me again what that means?
[03:21:54 CET] <kepstin> pinPoint: scroll up, i pasted a wikipedia link...
[03:22:10 CET] <furq> ha
[03:23:14 CET] <furq> oh well at least we have no way of knowing where he lives
[03:23:54 CET] <kepstin> hmm? no, levice.club appears to be pointing at a residential internet connection in slovakia...
[03:24:05 CET] <furq> https://en.wikipedia.org/wiki/Levice
[03:25:15 CET] <furq> if only there were some clue
[03:26:04 CET] <pinPoint> wth was that? an inside source? very strange
[03:26:50 CET] <pinPoint> digital cinema package... hmm
[03:27:03 CET] <furq> if he hadn't abruptly left i'd have assumed he was lying to make us more likely to join his teamview session
[03:27:29 CET] <kepstin> yeah, it's how movies are distributed to theatres nowaday. send a hard drive with the dcp files.
[03:28:47 CET] <pinPoint> i was about to join TW session but it was blocked. :/
[03:28:59 CET] <pinPoint> bad thing no?
[03:35:16 CET] <pinPoint> furq: still. ultrafast 2000k still looks better than 4000k h264.
[03:35:30 CET] <furq> what h264 preset
[03:37:35 CET] <pinPoint> libx264
[03:37:40 CET] <pinPoint> brb, another round
[03:40:17 CET] <TD-Linux> oh wow, 2mbit/s on 4k? yeah that's not surprising
[03:40:25 CET] <TD-Linux> h264 is basically DC blocks only at that rate
[03:47:09 CET] <kepstin> that's what, 1/2 a bit per pixel per second, more or less? :)
[03:47:41 CET] <pinPoint> you see my blog-a-roo
[04:21:34 CET] <pinPoint> TD-Linux: yeah even at 2k it will look better than 4k h264
[04:54:37 CET] <Yans> hi
[04:56:03 CET] <Yans> i try to add subtitles to mp4 file, via ffmpeg -i video.mp4 -vf "ass=subtitle.ass" out.mp4
[04:56:08 CET] <Yans> but
[04:56:35 CET] <Yans> i got problem with "-sub_charenc UTF-8"
[04:56:40 CET] <Yans> how to do it correctly?
[04:56:45 CET] <Yans> anyone? please...
[05:17:21 CET] <Yans> nvm done it
[05:22:25 CET] <jimby> Yans: you might find additional help at https://en.wikibooks.org/wiki/FFMPEG_An_Intermediate_Guide/subtitle_options
[05:23:39 CET] <Yans> ffmpeg -i movie.mp4 -sub_charenc cp1250 -i Victoria.srt -vcodec copy -acodec copy -scodec mov_text -metadata:s:s:0 language=666 moviecopy.mp4
[05:23:42 CET] <Yans> and works
[05:23:44 CET] <Yans> niach niach niach
[05:33:38 CET] <k_sze> So ffmpeg's .nut container format is mostly streamable, right?
[05:34:05 CET] <k_sze> I've been encoding videos as .nut files over SMB.
[05:34:42 CET] <k_sze> Sometimes the network connection is so bad that ffmpeg gives up, which means the .nut file isn't properly finalized.
[05:35:08 CET] <k_sze> Is there a way to run ffmpeg on the affected .nut file and just finalize it properly?
[05:35:31 CET] <k_sze> like, putting closing the XML tags, etc, and putting the index at the end (I think .nut is XML-based?)
[05:36:15 CET] <c_14> ffmpeg -i input.nut -c copy -map 0 out.nut should work
[05:36:19 CET] <c_14> Not sure if nut needs that though
[05:43:36 CET] <k_sze> The index at the end of the .nut file is for fast seeking, right?
[05:45:18 CET] <k_sze> c_14: I know the .nut container format is very error resilient.
[05:45:34 CET] <k_sze> It's just nicer to have a completely well-formed file.
[05:45:47 CET] <k_sze> and I suppose building the index helps
[05:45:52 CET] <k_sze> for fast seeking.
[05:48:40 CET] <c_14> the command I mentioned is the only way to add an index using ffmpeg since ffmpeg doesn't modify its input
[06:46:58 CET] <xintox> hey waressearcher2
[07:39:45 CET] <waressearcher2> xintox: hallo
[08:09:10 CET] <xintox> waressearcher2: hey buddy. whatcha working on?
[08:36:35 CET] <waressearcher2> xintox: nichts
[08:41:30 CET] <ented> Hello can C++ structures like vector store easily AVPacket? Or I rather should use AVPacketList?
[08:57:31 CET] <spiderkeys> Does anyone know how to get a custom AVIOContext used for writing to exhibit the behaviour listed under it's "direct" member as mentioned here: https://ffmpeg.org/doxygen/2.2/structAVIOContext.html
[08:59:55 CET] <spiderkeys> I've tried setting that value to 1, as well as setting various flush_packet flags on the avformatcontext and outputformat
[09:01:20 CET] <spiderkeys> I would like my custom write_packet() function to be called on every av_write_frame(), rather than it waiting to fill the entire aviocontext buffer before calling it
[09:53:10 CET] <spiderkeys> Wow, it ended up being as hacky as calling av_write_frame() a second time with a null AVPacket argument
[09:55:46 CET] <spiderkeys> There has to be a better way that that, with all of the direct/flush flags that seem to indicate that someone thought this might be an important feature for a custom aviocontext...
[10:49:15 CET] <codingquark> Need some help, I'm using aformat and amerge on two inputs - audio + video. Need to loop the audio till the video ends. -loop 1 isn't working here. How to do this?
[10:50:43 CET] <c_14> -stream_loop should do it
[10:51:30 CET] <codingquark> Trying.
[11:49:56 CET] <codingquark> c_14: Not working. Here's my command: http://pastebin.com/Wj1FYD4S
[11:50:32 CET] <c_14> You want the audio to loop, put it in front of the audio not in front of the video
[11:50:42 CET] <c_14> also, you'll want -shortest in that case as well (that one's an output option)
[11:53:11 CET] <codingquark> hmm... Trying that.
[11:56:39 CET] <codingquark> Tried -stream_loop -1 -i audio.mp3 (rest the same). Not working.
[11:57:15 CET] <c_14> What about it isn't working, do you get an error?
[11:57:42 CET] <codingquark> No errors. The video ends when the audio.mp3 ends (no loop)
[11:59:57 CET] <codingquark> Oh, the video contains audio. I'm adding one more.
[12:00:04 CET] <codingquark> Does that affect anything?
[12:05:36 CET] <c_14> eh, btw are you sure you want amerge and not amix?
[12:06:11 CET] <c_14> But yeah, it looks like that's a bug/feature request.
[12:06:45 CET] <c_14> If you just want it to work this once, concat the audio file a couple of times, then it should work
[12:08:13 CET] <codingquark> I thought about concat'ing audio files, but then I'll have to get video duration and make the looped audio accordingly.
[12:08:27 CET] <codingquark> I need to look at amix.
[12:09:10 CET] <codingquark> c_14: Thanks for the help!
[13:05:17 CET] <codingquark> Asked the question on StackExchange before submitting bug/feature request - http://video.stackexchange.com/questions/17195/loop-and-mix-audio-on-video
[13:51:34 CET] <Phrk_> Hello, is there a way to download only ffserver ?
[13:51:55 CET] <Phrk_> I don't want to install more than 300 package on my server, just for this app
[14:12:35 CET] <paule32_> hello
[14:12:55 CET] <paule32_> can someone help me to install and configure jwplayer?
[14:13:45 CET] <paule32_> or give it a embeeded player for ffmpeg streams?
[15:38:20 CET] <mosb3rg> hey guys, im running into a common problem that i see others have been able to navigate i have read everything i can find on the subject and im not getting any where. ffplay honors the headers for cookies without any issues., but ffmpeg absolutely refuses to do this anymore. what patching can be done to fix this issue if any. without HLS cookies being noted the majority of playlists these days cannot be dumped correctly..
[15:39:40 CET] <mosb3rg> https://ffmpeg.org/pipermail/ffmpeg-devel/2015-March/170397.html
[15:40:06 CET] <whiskey_time> it is live now
[15:40:18 CET] <whiskey_time> whoops :)
[15:40:25 CET] <mosb3rg> just to show you guys this is a known bug i assume. but i feel as if this worked excellent in the past. now just absolutely doesnt work any further. and in lue of trying to pipe which has failed on all attempts from other source programs to send the feed to ffmpeg even that didnt work either
[16:48:25 CET] <nimzter> Hello: I have just a very simple question? :  : : I'm scrolling invideo text with drawtext but I want to be showed from right to left (not left-right) Thanks
[16:48:26 CET] <droid909> guys, are there some other ways to access mic other than with ffmpeg -f avfoundation -i ":0"
[16:48:47 CET] <droid909> because to have avfoundation i have to install xcode
[16:49:12 CET] <droid909> install xcode just to be able to record audio ..
[16:49:20 CET] <droid909> i'm on mac os
[16:51:23 CET] <c_14> nimzter: just reverse the x y expansion so it moves the other way?
[16:52:39 CET] <droid909> are there alternatives to avfoundation to access mic on on mac os
[16:52:45 CET] <droid909> without installing xcode
[16:53:25 CET] <c_14> installing linux and using v4l2 (assuming that has drivers for the device)
[16:54:36 CET] <nimzter> Can you provide me some sample please :( I'm new with ffmpeg
[16:56:15 CET] <c_14> x=w*-t maybe
[18:49:31 CET] <paule32_> hello
[18:50:33 CET] <paule32_> give it a filter add on, that can display text, scroll text form right to left, or overlay in corners, or down to up?
[18:55:21 CET] <DHE> the drawtext filter can do that. text coordinates are formulas so you can specify a loop with modulos and an offset with the frame counter, for example
[18:56:25 CET] <paule32_> thx
[19:12:44 CET] <paule32_> damm, option not supported
[19:12:54 CET] <paule32_> have to compile ffmpeg
[19:13:09 CET] <paule32_> is there a option --enabled-all ?
[19:13:19 CET] <paule32_> to enable all possible filters?
[19:16:55 CET] <techtopia> how can i encode a 4:2:2 colorspace video as 4:2:0 ?
[19:23:01 CET] <DHE> paule32_: it doesn't force them enabled. broadly speaking it tries to enable any filter whose dependencies are met. what you need is --enable-libfreetype
[19:31:58 CET] <paule32_> DHE: http://fpaste.org/302873/63495145/
[19:34:35 CET] <shincodex> does libavcodec system library of i686 folder handle mjpeg decoding
[19:35:00 CET] <shincodex> context width/height is zero and av_read_packet always gets audio streams only yet it reports a video stream is present
[19:35:16 CET] <Mavrik> techtopia, -pix_fmt yuv420p parameter.
[19:39:53 CET] <techtopia> thanks mavrik
[19:41:52 CET] <paule32_> Mavrik: ?
[19:43:35 CET] <Mavrik> huh?
[19:45:14 CET] <shincodex> mjpeg:
[19:45:17 CET] <paule32_> have you follow
[19:45:24 CET] <shincodex> says unimplemented restart marker
[19:45:38 CET] <paule32_> Mavrik: :  http://fpaste.org/302873/63495145/
[19:46:12 CET] <shincodex> Unimplemented RTP/JPEG restart marker header.\n
[19:46:19 CET] <shincodex> Does that mean mjpeg does not work because of this
[20:02:01 CET] <shincodex> help me im dieing
[20:05:05 CET] <paule32> no ideas?
[20:05:44 CET] <shincodex> is mjpeg supposed to work if i were to install ubuntu?
[20:11:00 CET] <paule32> drawbox is ok
[20:11:15 CET] <paule32> but drawtext, missing
[20:24:35 CET] <shincodex> it appears ubuntu mate System libavcodec mjpeg decoder is not compiled into system
[20:24:39 CET] <shincodex> why would they do this
[20:25:46 CET] <furq> do you want me to phone mark shuttleworth and ask him
[20:26:44 CET] <shincodex> yes cause i dont know him and it seems like you got a speed dial
[20:32:28 CET] <shincodex> maybe there is hope
[20:44:30 CET] <droid909> s/h/p/
[20:52:37 CET] <paule32> so
[20:53:26 CET] <paule32> i have change the config.h, set config_drawtext_filter to 0
[20:53:44 CET] <paule32> but this brings me no effect like 1
[20:53:53 CET] <paule32> the filter is missing
[20:54:37 CET] <paule32> [AVFilterGraph @ 0xbb9900] No such filter: 'drawtext'
[21:14:07 CET] <Mavrik> paule32, when do you ./configure, is it listed under filters?
[21:14:45 CET] <paule32> yes
[21:14:51 CET] <Mavrik> paule32, also did you enable the library required for that filter?
[21:14:59 CET] <Mavrik> "enable libfreetype, needed for drawtext filter [no]"
[21:15:00 CET] <paule32> yes
[22:13:45 CET] <paule32> ok
[22:14:16 CET] <paule32> i have a tip for everyone here where will use drawtext
[22:14:39 CET] <paule32> in ./configure   add --enable-filter=drawtext
[22:14:57 CET] <paule32> and you have support for "drawtext" filter
[22:23:19 CET] <paule32> so, drawtext works now
[22:26:36 CET] <waressearcher2> sehr schön
[22:26:49 CET] <waressearcher2> wunderbar
[22:30:58 CET] <shincodex> Yo want to see my crazy configure?
[22:31:24 CET] <shincodex> its like 20 lines long
[22:33:11 CET] <paule32> yes, with root password, it is possible
[22:36:09 CET] <shincodex> i always use root
[22:36:25 CET] <paule32> so, how is it possible, to change the "drawtext" text in a period by a txt.file?
[22:36:59 CET] <paule32> so i can show 20 sec hello and after this 20 seconds, new 20 seconds world?
[22:43:28 CET] <Mavrik> paule32, so you're trying to make subtitles? :P
[22:43:56 CET] <paule32> ehm, more for infos or scrolling text
[22:54:01 CET] <paule32> like text all 5 minutes (football measures)
[23:04:39 CET] <paule32> how can i stream alsa or pulse?
[23:04:53 CET] <paule32> in combination with x11grab
[23:04:54 CET] <paule32> ?
[23:24:34 CET] <debianuser> paule32: Something like https://unix.stackexchange.com/a/84013 ?
[23:25:37 CET] <paule32> seems good
[23:26:05 CET] <paule32> but, i would like to stream vlc webradio audio for test at the moment
[23:26:17 CET] <paule32> thx debianuser
[23:28:49 CET] <paule32> debianuser: so, now, the question is, how to setup filter to stream only desktop and webradio?
[23:32:44 CET] <debianuser> I'm not sure what "vlc webradio" is, but I guess it should be something like `ffmpeg -f x11grab -i :0.0   -f <yourwebradiotype> -i <yourwebradiosource>  -f <yourstreamformattype> <streamurl>`
[23:33:39 CET] <debianuser> Ffmpeg's command line format is usually like `ffmpeg -i <input1> -i <input2> ... -i <inputN> <output-filters> <output>`
[23:34:52 CET] <debianuser> Ah. Optionally you may have some input-specific options before each input. So that would be `ffmpeg <input1-options> -i <input1> ... <inputN-options> -i <inputN>  <output-filters> <output>`.
[23:34:54 CET] <Prelude_Zzzzz> hey guys.. anyone know how i can trim a video image ( black borders on two sides ) and stretch the image without transcoding it ?
[23:35:31 CET] <debianuser> paule32: You'll also probably need `-s 1920x1080` option for x11grab input, I thing it's something like 640x480 by default.
[23:35:49 CET] <paule32> desktop stream is ok
[23:35:51 CET] <paule32> no problems
[23:35:54 CET] <paule32> but with sound
[23:44:06 CET] <Prelude_Zzzzz> anyone? i don't want to transcode a video  i just want to trim the black lines of the video on both the left and right side
[23:45:32 CET] <paule32> debianuser: [flv @ 0x384c8c0] FLV does not support sample rate 48000, choose from (44100, 22050, 11025)
[23:45:33 CET] <paule32> [flv @ 0x384c8c0] Audio codec pcm_s16le not compatible with flv
[23:47:45 CET] <Mavrik> Prelude_Zzzzz, you can't.
[23:47:57 CET] <Mavrik> Since you have to modify data that's encoded.
[23:48:41 CET] <paule32> debianuser: http://fpaste.org/302963/78888145/
[23:52:54 CET] <pinPoint> So I did this: ffmpeg -i bbb_sunflower_2160p_30fps_normal.mp4 -c:v libvpx-vp9 -b:v 2000k -threads 8 -speed 4 -an BBB_H265_ULTRAFAST_2000k.webm
[23:53:26 CET] <debianuser> paule32: Try dropping "-c:a copy" part. "c:a" == "acodec", and you have "-acodec mp3" followed by "-c:a copy", first says "encode it as mp3", second one says "leave it without reencoding", you can't do both :)
[23:53:48 CET] <pinPoint> the file is only 135MB. HEVC version is 152MB at 2000k as well. does vp9 usually produce smaller files than hevc for now?
[23:54:28 CET] <paule32> debdog: heh
[23:54:52 CET] <Mavrik> pinPoint, that's a really strange metric.
[23:55:01 CET] <Mavrik> Considering you've set a bitrate parameter.
[23:55:14 CET] <Mavrik> Only thing that tells you is that one of the encoder is shit at keeping the bitrate.
[23:55:18 CET] <Mavrik> Nothing about "smaller files" :)
[23:55:30 CET] <pinPoint> so minrate maxrate?
[23:55:39 CET] <pinPoint> does vp9 even have that option?
[23:55:47 CET] <Mavrik> No idea.
[23:56:06 CET] <Mavrik> Last I checked neither x265 or libvpx had good bitrate control.
[23:57:11 CET] <pinPoint> so if I ran the same the command 5 times, I'd get different filesizes always?
[23:58:38 CET] <paule32> debianuser: then:
[23:58:40 CET] <paule32> [auto-inserted resampler 0 @ 0x2efce40] [SWR @ 0x2efd260] Failed to initialize resampler
[23:58:40 CET] <paule32> [auto-inserted resampler 0 @ 0x2efce40] Failed to configure output pad on auto-inserted resampler 0
[00:00:00 CET] --- Sat Dec 19 2015