[Ffmpeg-devel-irc] ffmpeg.log.20161116

[burek]

[00:08:24 CET] <Chloe[m]> OK so it'd be manual. And, I have a lot of DV would h264 lossless + flac be the best way to compress it losslessly? (using veryslow preset)
[00:13:24 CET] <furq> if by "compress" you mean "make bigger" then sure
[00:14:03 CET] <Chloe[m]> you can't compress DV further?
[00:14:17 CET] <furq> dv is already a lossy codec
[00:14:30 CET] <furq> you can compress it further by using another lossy codec but you'll obviously lose quality
[00:14:42 CET] <Chloe[m]> yes, but modern lossless codecs aren't better than DV?
[00:14:51 CET] <furq> i doubt it
[00:15:09 CET] <furq> it probably depends on the source so you'd have to test to be sure
[00:15:18 CET] <furq> lossless h264 is huge, though
[00:15:30 CET] <furq> and it's one of the better lossless codecs
[00:17:03 CET] <furq> also if this is ntsc dv then i'm not sure if h264 supports yuv411p
[00:17:27 CET] <furq> it's not listed in ffmpeg -h encoder=libx264
[00:17:35 CET] <furq> s/h264/x264/
[00:17:42 CET] <Chloe[m]> PAL DV luckily
[00:17:45 CET] <TD-Linux> DV is basically a worse JPEG
[00:20:19 CET] <Chloe[m]> yeah, this isn't going to compress more than DV :/ 50000kb/s where the DV is 30000kb/s
[00:20:46 CET] <llogan> if you want to filter it such as deinterlacing and denoising, the sure, re-encode. if you have a bunch of disk space then just leave them.
[00:20:54 CET] <furq> there are other lossless codecs you can try, but i'd be surprised if any of them beat 30mbps
[00:21:55 CET] <Chloe[m]> I no longer have the tapes, so I want to keep the original streams. But it's quite a lot ~250GB, I was just wondering if I could reduce that
[00:22:12 CET] <furq> compressing the audio should work fine, but i expect that's small consolation
[00:22:21 CET] <Chloe[m]> I probably will reencode lossy with some filtering in the future though
[00:23:43 CET] <llogan> i recommend your favorite deinterlacer with perhaps hqdn3d and maybe even drawbox to cover up any head switching noise on the bottom (unless you want to crop or scale or whatever but i prefer to just keep as it)
[00:24:26 CET] <llogan> how did you capture the tapes?
[00:24:42 CET] <furq> hardcoded black bars make me sad
[00:24:56 CET] <furq> even though i know x264 still encodes them but hides them if you have a non-mod16 crop
[00:25:09 CET] <llogan> furq: you could use a softsub
[00:25:20 CET] <furq> huh
[00:25:47 CET] <llogan> the noise annoys some people
[00:25:56 CET] <furq> well yeah but i just crop that
[00:32:54 CET] <Chloe[m]> llogan: via a custom program I wrote
[00:33:16 CET] <Chloe[m]> well, custom ffmpeg device
[00:33:20 CET] <Chloe[m]> still need to clean that up and send a patch
[00:45:59 CET] <kbarry> I'm wondering if there is a switch for ffprobe to continually probe (i'm looking for changing metdata)
[00:47:10 CET] <Chloe[m]> You can use watch for that
[00:47:58 CET] <Chloe[m]> kbarry: watch -n1 ffprobe myfile.mp4
[00:52:04 CET] <kbarry> yeah, Thanks for the suggestion. Im using Mac. I found a solution that works: http://stackoverflow.com/questions/9574089/osx-bash-watch-command   I should hav googled a little longer.
[01:26:28 CET] <kurufu> so does anyone know how to generate the AVCodecParameters::extradata for a muxer reading encoded AAC data?
[01:26:57 CET] <kurufu> It looks like its just a raw byte array and im not seeing any way to figure out what it should be but the muxer refuses to open an audiostream without it.
[01:29:42 CET] <t4nk388> hello
[01:47:09 CET] <kyleogrg> hello
[01:48:33 CET] <kyleogrg> I'm trying to concat a few MP3s together.  They have the same bitrate, sample rate, etc.  The resulting MP3 has (in VLC) a duration that keeps changing.  How can I fix this?
[01:49:59 CET] <llogan> depends on how you're concatting
[01:51:10 CET] <kyleogrg> ffmpeg -i "concat:file1.mp3|file2.mp3" -c:a copy "output.mp3"
[01:51:34 CET] <furq> you probably want to use the demuxer
[01:51:48 CET] <furq> https://trac.ffmpeg.org/wiki/Concatenate#demuxer
[01:52:13 CET] <kyleogrg> Will this have a different effect?
[01:52:19 CET] <furq> i assume using the protocol will result in mp3 headers in the middle of the file, which i guess explains the changing duration
[01:52:32 CET] <furq> since the protocol is more or less just cat file1.mp3 file2.mp3 > output.mp3
[01:52:55 CET] <kyleogrg> okay
[01:53:03 CET] <furq> the demuxer will actually demux the mp3 stream and write it into a new container
[01:53:10 CET] <furq> s/stream/streams/
[01:53:28 CET] <furq> inasmuch as you can call mp3 a container
[01:54:35 CET] <kyleogrg> okay, i'm trying it
[01:55:17 CET] <kyleogrg> no, same problem...
[01:57:02 CET] <kyleogrg> i put them into a txt file and then did ffmpeg -f concat -i myfile.txt
[02:01:11 CET] <llogan> does it make a difference if you add "-write_xing 0" as output option?
[02:03:15 CET] <kyleogrg> it actually helps a lot, but the duration still wavers by 1 sec at the beginning of the file
[02:05:59 CET] <kyleogrg> ahh, but the duration is 6 minutes longer than it should be
[02:33:27 CET] <kyleogrg> When I concat the MP3s to an MPG (no video), the duration is correct!  Then when I copy this MPG to an MP3, the duration is crazy again.  Why...?
[02:48:07 CET] <kyleogrg> found a solution using a different program
[02:48:10 CET] <kyleogrg> see ya
[05:56:54 CET] <wpm> is this right channel for help building ffmpeg from source?
[06:11:47 CET] <sim590> I have a question which is not related to ffmpeg, but which could most certainly be answered in this channel. I'm trying to use rtmpsrv, as suggested on http://askubuntu.com/questions/262248/record-live-stream, to get the rtmpdump command to use to save a stream. I do execute the iptables command and then run the rtmpsrv. I then, refresh a page with a stream on it (for e.g., twitch.tv), but I never see any
[06:11:49 CET] <sim590> output from rtmpsrv.
[06:12:10 CET] <sim590> Any idea what I could do to find out how to make it work?
[06:19:56 CET] <sim590> or can I directly download a stream from a web page using ffmpeg?
[06:23:20 CET] <codehotter> so I have a microphone and I need to record input from it and split it into a bunch of 200ms mp3 files. So 10 seconds would be 50 files. How do I do that? Can I do it realtime? What would be the lag for the next 200ms file to appear in my folder? What settings do I use to minimize that delay?
[07:26:44 CET] <Pandela> Has anyone had progress with getting a video stream from ffmpeg into Manycams IP Camera feature?
[08:09:09 CET] <fling> [libopus @ 0x562a5275e110] Queue input is backward in time
[08:09:18 CET] <fling> Should I use another codec? ^
[09:48:11 CET] <c00lways> anyone know why am i getting Expected int64 for y but found 1
[09:50:49 CET] <durandal_1707> can you explain what are you doing?
[09:53:09 CET] <james_> i ran ffmpeg command after compilation, and it returns that error when i tried to convert any video file
[09:53:28 CET] <james_> ./ffmpeg -i /opt/webserver/video.h264 -itsoffset -00:00:00.4 -i /opt/webserver/audio.wav -c:v copy -c:a aac -strict -2 -y /opt/webserver/public/videos-source/recording/rpi1.mp4
[09:54:15 CET] <james_> i was using cross compilation on ubuntu for rpi1
[09:56:01 CET] <james_> @duranda1_1707, i can cross compile for rpi on a 64bit ubuntu 14 right?
[09:56:39 CET] <james_> could it be due to rpi is 32bit?
[09:57:03 CET] <james_> ffplayer did comes out its compilation information before segmentation fault
[09:58:32 CET] <james_> btw, i found the static build for armel throws illegal instruction when i tried to encode via libx264
[09:58:51 CET] <james_> that is why iam trying to build it on my own
[09:59:19 CET] <james_> and also i think it does not support openmax
[10:23:33 CET] <c00lways> anyone have any clue? i saw this parameter has been checked inside width and height
[10:24:08 CET] <c00lways> but other version of static build ffmpeg i downloaded doesn't have this issue
[11:49:28 CET] <Pandela> I'm having trouble with udp streaming on my network :p
[11:49:57 CET] <Pandela> I cant pick up the stream in ffplay on another computer in the network, pretty sure my firewall and stuff is out of the way
[11:50:40 CET] <Pandela> Seems to be looping it back to my machine, instead of broadcasting on the network
[12:21:47 CET] <kerio> how does udp streaming work?
[12:21:54 CET] <kerio> just put the video in a mpegts and send it?
[12:56:36 CET] <DHE> kerio: typically yes... multicast is popular as a means of doing so
[12:57:09 CET] <kerio> does mpegts support uncompressed video?
[14:04:53 CET] <DHE> kerio: I don't think so. but I wouldn't want to steam bulk data over UDP anyway. if you're talking uncompressed I assume that's a lot of bandwidth
[18:19:18 CET] <Mysoft> assuming i have raw video for input and... x264 as output using -pix_fmt yuv420p
[18:19:36 CET] <Mysoft> having that raw video as yuv420p speed up things a little?
[18:22:23 CET] <kerio> probably yeah
[18:22:32 CET] <kerio> it's a relatively simple conversion tho
[18:23:11 CET] <Mysoft> yeah, but in my case it has a increased gain
[18:23:41 CET] <Mysoft> because then i can convert the RGB32 capture to yuv420p before piping to ffmpeg
[18:25:09 CET] <Mysoft> also since i have to copy from the 3D surface to a buffer to be able to pipe... doing the conversion during that point is also "extra effective"
[19:02:34 CET] <bencoh> anyone checked libde265 vs ffmpeg native hevc decoder recently?
[19:02:42 CET] <bencoh> (in term of speed)
[19:24:45 CET] <sweLogan> Hi
[20:02:33 CET] <sweLogan> Hello?
[20:06:52 CET] <c_14> Just ask your question, and if someone can help you they will.
[20:11:33 CET] <shincodex> heres my question
[20:11:45 CET] <shincodex> send me gcc 6.1.0 binary for i686
[20:11:52 CET] <shincodex> all of them
[20:13:01 CET] <c_14> That's neither a question nor ffmpeg related.
[20:13:13 CET] <shincodex> ffmpeg doesnt exist without gcc
[20:13:25 CET] <shincodex> but as for the question part.... hmmm
[20:13:26 CET] <c_14> You can use clang
[20:13:31 CET] <shincodex> lol shite
[20:13:43 CET] <c_14> And ffmpeg doesn't exist without electricity but you don't see us talking about power generators in here.
[20:13:54 CET] <kerio> cool story bro
[20:14:11 CET] <shincodex> energy.... gotta stop using it up
[20:37:17 CET] <sweLogan> I im trying to stream a webpage to twitch, for example like OBS where you enter your URL for donation alerts
[22:06:35 CET] <ksk> hey, I read about "faststart movflag" - however in man ffmpeg on debian there is no mentioning of it. could you please help me out? thanks!
[22:07:02 CET] <c_14> -movflags +faststart
[22:07:09 CET] <c_14> Is in ffmpeg-formats iirc
[22:09:34 CET] <ksk> I was missing the "+". thanks!
[23:03:52 CET] <ksk> mhhm, more generally speaking, is it possible to convert $input into x264/mp3->mp4 and play that output as it is converting? added -movflags +faststart but it does not seem to play (vlc eg. says "moov atom not found")
[23:04:16 CET] <TD-Linux> not with the mp4 container, no.
[23:07:29 CET] <ksk> ah okay. which ones can do that? Is there an "easy" way I could find that out?
[23:08:52 CET] <c_14> Pretty much every container that isn't mp4
[23:09:13 CET] <c_14> Though you can try enabling fragmentation (if the player you're using supports that)
[23:13:34 CET] <ksk> hah, okay, gonna dig that. many thanks!
[23:15:38 CET] <haasn> ./configure --enable-lto && make # fails with ar: libavdevice/alldevices.o: plugin needed to handle lto object etc.
[23:16:05 CET] <haasn> Sounds like `ar` is being used by the ffmpeg build process incorrectly? you need to add --plugin <path> to get LTO support into ar
[00:00:00 CET] --- Thu Nov 17 2016