[Ffmpeg-devel-irc] ffmpeg.log.20170416

[burek]

[02:13:42 CEST] <evgeniy> how can i use ffmpeg
[02:13:45 CEST] <evgeniy> to record my
[02:13:47 CEST] <evgeniy> audio
[02:13:50 CEST] <evgeniy> what i hear..
[02:14:00 CEST] <evgeniy> please just give me the command
[02:14:03 CEST] <evgeniy> i need it right now
[02:14:10 CEST] <evgeniy> and i can't find anything to do it
[02:18:14 CEST] <jerome-> That was fast to give up :)
[02:18:45 CEST] <jerome-> Hello, I'm trying to understand how to record my ip webcam with ffmpeg, but I'm getting an error: http://uu.zoy.org/p/eSIrww#x=umwyANNBdgAIFBcA  I'm new with it, and I don't understand the error
[02:20:53 CEST] <BtbN> Update your ffmpeg
[02:21:09 CEST] <BtbN> (Or do as the error says, but you don't want to use that old aac encoder)
[02:21:38 CEST] <jerome-> ah that's the latest one with ubunut
[02:21:44 CEST] <jerome-> ok thanks
[02:22:24 CEST] <BtbN> 2.8 is almost 2 years old
[02:40:14 CEST] <jerome-> yes, I have an old version of ubuntu 16.04
[02:40:17 CEST] <jerome-> thanks!
[02:41:26 CEST] <BtbN> Ubuntu is not going to change ffmpeg version within the same release
[02:43:13 CEST] <jerome-> yes, I should at least update to a more recent ubuntu (or install myself a newer version of ffmpeg)
[03:21:19 CEST] <anonymousey> i have a video that is 3840x2178. on my 2010 basic laptop, it's virtually unwatcheable due to choppiness. would reducing resolution to 1366x768 (my laptop's res) or to 720p help? Can ffmpeg do this?
[04:29:54 CEST] <agentsim> when using libavcodec to read jpeg files, is there a way to detect the orientation of the image from exif metadata?
[05:42:12 CEST] <damdai> how do i play  ratdvd file
[05:52:46 CEST] <dystopia_> work out what codec the video uses and install one damdai
[05:52:59 CEST] <damdai> no idea
[05:53:00 CEST] <damdai> what it uses
[05:53:17 CEST] <dystopia_> RatDVD (originally stylized "ratDVD") is the name of a proprietary container format for digital video, developed by Peter Jensen and a group of Russian and Danish university students. The container format is a compressed archive format that holds all features of DVD-Video in a single file.
[05:53:20 CEST] <dystopia_> so
[05:53:30 CEST] <dystopia_> it's a container
[05:53:35 CEST] <dystopia_> and inside it is your video
[05:53:39 CEST] <dystopia_> which will be mpeg2
[05:54:29 CEST] <damdai> no video. just get sound using  mpc
[05:54:40 CEST] <dystopia_> seems mpc and mplayer both have native support for it
[05:55:45 CEST] <dystopia_> you can install the "ratdvd" software and convert the file back to a normal dvd
[05:55:52 CEST] <dystopia_> then play that or encode it into somthing else
[05:55:59 CEST] <dystopia_> http://ratdvd.ca/convertrat.php
[08:50:43 CEST] <CyberShadow> Hello. I'm trying to imitate an MPEGTS stream for an STB, but I can't get ffmpeg to produce a format that the STB will accept.
[08:50:43 CEST] <CyberShadow> https://dump.thecybershadow.net/e40b6a8cd9fb9423d1bfe30e1a87fd6b/05%3A55%3A23-upload.png <- on the left is what a working stream looks like in Wireshark. On the right is ffmpeg's output for the mpegts format. How can I make ffmpeg's output more like the stream on the left?
[08:50:57 CEST] <CyberShadow> I might be wrong but I think the problem might be that ffmpeg is not producing sync bytes
[09:31:44 CEST] <Immatix> I might be missing something obvious, but is there an explanation how a direct stream copy (acodec copy, vcodec copy) results in a file some ~410MB smaller than the source file? Here's the details of the before and after files https://pastebin.com/9SYiyCS1
[09:34:06 CEST] <furq> Immatix: mpegts has a lot of overhead
[09:44:06 CEST] <Immatix> furq: I figured it was something like that, so it's normal? I didn't think it was actually encoding anything given how fast the operation is
[10:17:02 CEST] <furq> it seems about right
[10:17:56 CEST] <furq> mpegts was designed for broadcast/streaming so it has a lot of overhead in every packet
[10:18:21 CEST] <furq> in my experience you'll save about 10% when remuxing to mkv/mp4
[10:18:36 CEST] <kerio> can ffmpeg manipulate a mjpeg stream to drop or duplicate frames without reencoding?
[10:19:47 CEST] <furq> probably not in one pass
[10:19:55 CEST] <furq> you could do it via image2
[12:42:07 CEST] <thebombzen> hm... I have a CFR video which is 120 fps - it's H.264-inside-mp4
[12:42:38 CEST] <thebombzen> I'd like to slow it down by a factor of 2, so it's half-speed and 60 fps
[12:43:04 CEST] <thebombzen> this should definitely be possible. the issue I'm having is that the setpts filter in libavfilter works on uncompressed frames
[12:43:11 CEST] <furq> do you mean without reencoding
[12:43:19 CEST] <thebombzen> well yes
[12:43:35 CEST] <furq> -i foo.mp4 -c copy tmp.264; -r 60 -i tmp.264
[12:43:47 CEST] <thebombzen> ah okay you have to use the raw h264 muxer
[12:43:51 CEST] <furq> yeah
[12:44:06 CEST] <thebombzen> I tried using "-r 60 -i input.mp4" but it still used the embedded 120 fps
[12:44:17 CEST] <thebombzen> thx
[12:44:37 CEST] <furq> i wonder if a dedicated muxer can do it in fewer keystrokes
[12:44:48 CEST] <furq> and fewer temp files ofc
[12:45:07 CEST] <thebombzen> I've noticed that timestamp manipulation is generally difficult with ffmpeg.c unless you're using uncompressed frames and libavfilter's setpts filter
[12:45:23 CEST] <thebombzen> in particular, it's hard to manipulate timestamps with ffmpeg.c while not transcoding
[12:45:58 CEST] <thebombzen> you can rebase everything to a constant framerate by using -r as an input option (sometimes)
[12:46:31 CEST] <thebombzen> but the answer you just provided doesn't work if you're trying to slow VFR content
[12:46:57 CEST] <furq> oh cool
[12:47:06 CEST] <furq> l-smash will do it in one shot if you have it installed
[12:47:14 CEST] <thebombzen> lsmash? I probably do
[12:47:23 CEST] <furq> muxer -i src.mp4?fps=60 -o dest.mp4
[12:48:02 CEST] <furq> gpac will do it as well but if you use gpac then JEEB veins will fill with tar
[12:48:05 CEST] <furq> +'s
[12:48:34 CEST] <thebombzen> why "muxer" and not "remuxer"
[12:48:45 CEST] <furq> remuxer doesn't have that option
[12:49:00 CEST] <furq> don't ask me why, or why muxer takes muxed files as inputs when there's a tool called remuxer
[12:49:13 CEST] <furq> i'm not japanese enough to understand#
[12:49:45 CEST] <thebombzen> what does japanese have anything to do with it?
[12:49:59 CEST] <furq> https://l-smash.github.io/l-smash/
[12:51:06 CEST] <thebombzen> oh I didn't realize it was a Japanese project
[12:51:10 CEST] <thebombzen> given that JEEB works on it
[12:51:27 CEST] <furq> what do you think the J stands for
[12:51:42 CEST] <thebombzen> probably Jan
[12:51:52 CEST] <furq> you're 60% right
[12:51:55 CEST] <thebombzen> Jan Ekström
[12:52:14 CEST] <thebombzen> you know
[12:52:16 CEST] <thebombzen> Jeeb's name
[14:22:53 CEST] <grublet> JEB
[14:22:59 CEST] <grublet> short for JEBEDIAH
[14:23:26 CEST] <grublet> yelling his name is mandatory
[14:27:31 CEST] <thebombzen> it's jeebediah
[14:28:48 CEST] <furq> i thought it was short for john ellis bush
[14:28:53 CEST] <furq> or is that Jeb!
[14:48:16 CEST] <thebombzen> yea it's not jeb bush unless you have an exclamation point
[14:48:57 CEST] <thebombzen> but jeeb isn't secretly jeb! though - you can tell by the cut of his jeeb
[15:02:56 CEST] <furq> jeeb stands for john extremely ellis bush
[15:46:27 CEST] <inflex> I have  a 1hr long MKV file which was recorded via vokoscreen, and I need to remove a portion of that video about 1 minute long midway through. While I can find guides to extracting a portion of a file and saving it to another, I cannot find anything in ffmpeg to *remove* a segment and reform the video
[15:46:42 CEST] <inflex> I presume maybe I have to cut my wanted segments out, then magically concat them back together?
[15:48:35 CEST] <furq> if you want to do it without reencoding then yes
[15:49:43 CEST] <inflex> furq, okay, best method to concat them?
[15:50:08 CEST] <inflex> wonder if someone's made a gui tool for just this sort of thing (tried avidemux but no such luck, and of course shotcut renecodes)
[15:50:47 CEST] <furq> -i foo.mkv -t 00:30:00 -c copy part1.ts; -i foo.mkv -ss 00:31:00 -c copy part2.ts; cat part1.ts part2.ts > out.ts
[15:51:19 CEST] <furq> that won't cut exactly at the timestamps, it'll be at the nearest keyframe
[15:52:14 CEST] <inflex> tx, that's fine, I appreciate your time.
[16:18:49 CEST] <inflex> furq, cuts and concats fine on the video side, but part2 audio is out of sync by a second or two.  Will have to investigate
[16:19:05 CEST] <inflex> ( the actual cut portion is out of sync )
[18:18:28 CEST] <thebombzen> inflex: that's because the audio frames aren't in alignment with the video
[18:19:09 CEST] <thebombzen> if you're okay with re-encoding the audio (which is probably fine, tbh) then you can figure out where the keyframes are
[18:19:28 CEST] <thebombzen> and then use -t and -ss for those timestamps. also, use -c:v copy but set the audio settings as you want
[21:01:39 CEST] <djk> it is possible within one ffmpeg to output to rtmp and have it save a still image every X seconds?
[21:05:44 CEST] <thebombzen> yea. you can have separate outputs
[21:06:07 CEST] <thebombzen> ffmpeg -i input output_opts output1 output_opts2 output2
[21:06:26 CEST] <furq> -i input rtmp://foo -r 1/10 out%03d.png
[21:06:48 CEST] <thebombzen> for your thing you'd want ffmpeg -i input rtmp_output options rtmp://url -r 1/X out%4d.png
[21:06:51 CEST] <thebombzen> or something like that
[21:07:01 CEST] <furq> hi
[21:07:11 CEST] <thebombzen> oh hello, it looks like you beat me to it
[21:40:10 CEST] <Tatsh> i have some mono content and i want to encode so it plays the same channel in all 5 of my speakers
[21:40:14 CEST] <Tatsh> what are my options?
[21:40:29 CEST] <Tatsh> usually mp4 container, fdk AAC?
[21:41:16 CEST] <c3r1c3-Win> Tatsh: OBS doesn't do surround sound.
[21:41:25 CEST] <Tatsh> not doing OBS
[21:41:37 CEST] <Tatsh> this is a recording off old VHS definitely mono content
[21:41:46 CEST] <Tatsh> and i just want to make it not be annoying for my dad
[21:41:52 CEST] <Tatsh> in the surround sound system he has
[21:41:57 CEST] <c3r1c3-Win> LOL, sorry... I thought I was in the OBS channel.
[21:42:24 CEST] <c3r1c3-Win> I think they have an example of that on the website.
[21:42:53 CEST] <c3r1c3-Win> https://trac.ffmpeg.org/wiki/AudioChannelManipulation
[21:42:55 CEST] <Tatsh> it would be appear i can encode in ac3 5.1
[21:43:12 CEST] <Tatsh> his receiver only does old DTS or dolby
[21:43:39 CEST] <c3r1c3-Win> https://trac.ffmpeg.org/wiki/Map
[21:43:39 CEST] <Tatsh> yea the 6 x mono thing
[21:43:41 CEST] <Tatsh> that's cool
[21:43:46 CEST] <Tatsh> good documentation = good
[21:44:20 CEST] <Tatsh> i'll be doing that for the final encode as one of the audio channels then
[21:44:22 CEST] <Tatsh> fake 5.1
[21:55:21 CEST] <rjp421> is there a preferred way to vetical flip a dshow input? just -vf?
[21:57:21 CEST] <rjp421> https://superuser.com/questions/578321/how-to-rotate-a-video-180-with-ffmpeg gives diff ways, but is there a preferred or more performant?
[21:58:10 CEST] <durandal_1707> vflip
[22:01:28 CEST] <rjp421> ty
[22:05:34 CEST] <rjp421> this is on a pi using the pi-cam... is that still the best way then? usually raspivid does the flip
[22:06:27 CEST] <rjp421> and its not dshow its v4l, derp
[22:06:50 CEST] <rjp421> ffmpeg -f v4l2 -loglevel debug -fflags nobuffer -r 24 -re -input_format h264 -i /dev/video0 -c:v copy -x264opts keyint=48 -an -sn -timecode `date '+%H:%m:%S.00'` -f flv 'rtmp://192.168.10.104/testplaylist/pi2cam4 flashver=FMLE/3.0\20(compatible;\20FMSc/1.0) live=1'
[22:07:15 CEST] <furq> an awful lot of those options do nothing with -c copy
[22:08:17 CEST] <furq> you can't vflip (or apply any filters) when copying
[22:08:27 CEST] <furq> maybe flv supports rotation metadata but that seems unlikely
[22:09:38 CEST] <rjp421> Filtergraph 'hflip,vflip,format=yuv420p' was defined for video output stream 0:0 but codec copy was selected.
[22:09:38 CEST] <rjp421> Filtering and streamcopy cannot be used together.
[22:09:41 CEST] <furq> yes
[22:09:42 CEST] <rjp421> :(
[22:10:50 CEST] <rjp421> furq, do i need to use raspivid then? :( that cmd is long and confusing, id rather not lol
[22:11:30 CEST] <furq> maybe
[22:11:53 CEST] <furq> you could pull rawvideo from the pi cam and use -c:v h264_omx
[22:12:08 CEST] <furq> that might be slower than raspivid though depending on what it's actually doing to rotate
[22:14:09 CEST] <rjp421> furq, thats what i was assuming i had to do originally when you mentioned it.... ill try ty.  does that mean i can set keyframe interval and gop too? :) it helps down the road converting to hls
[22:14:17 CEST] <furq> yeah you can do whatever
[22:16:50 CEST] <norbert> hi, when I use ffmpeg to add .ass subtitles, can I make the italic (<i>...</i>) appear as italic?
[22:29:11 CEST] <norbert> or any kind of subtitle format that works with italic, then I'll just convert it to whatever ffmpeg wants
[22:29:40 CEST] <JEEB> "add" as in?
[22:29:51 CEST] <norbert> hardcoded, to make it part of the video
[22:30:00 CEST] <JEEB> for hardcoding there's the libass-based ass filter which should support all that stuff
[22:30:13 CEST] <norbert> yes, ffmpeg -i untitled.mp4 -vf "ass=untitled.ass" output.mp4
[22:30:29 CEST] <norbert> but it doesn't process the italic, it just prints <i>...</i>
[22:30:51 CEST] <JEEB> I don't exactly remember how italic works in ASS but if aegisub shows it as italic then it should be "OK"
[22:31:03 CEST] <JEEB> as far as I've seen libass render stuff italic has worked
[22:31:28 CEST] <JEEB> although <i> might be a thing that's not ASS
[22:31:39 CEST] <JEEB> as that's the HTML thing, wasn't it?
[22:32:00 CEST] <JEEB> aegisub outputs "{\i1}Test{\i0}"
[22:32:12 CEST] <JEEB> so yes, <i></i> is not part of ASS
[22:32:14 CEST] <norbert> yeah, so it seems
[22:32:27 CEST] <norbert> thanks
[22:34:52 CEST] <furq> <i> is srt
[22:35:14 CEST] <norbert> I started in subtitleeditor with an .srt and then did save as to .ass, but I guess the program doesn't automatically convert <i> to {\i1} etc.
[22:35:31 CEST] <furq> -vf subtitles should work with either
[22:35:37 CEST] <furq> assuming you don't have any fancy ass-specific stuff in there
[22:49:42 CEST] <norbert> very nice, it does exactly what I want
[22:49:53 CEST] <norbert> tried vlc and mencoder and they didn't want to cooperate
[22:50:16 CEST] <norbert> but here I can force_style the font size and color easily and it properly centers and splits lines etc.
[22:50:44 CEST] <norbert> thanks again for the pointers
[23:56:23 CEST] <unomystEz> hey all
[23:57:59 CEST] <unomystEz> this might sound bad but anyone have any ideas on which effects might help against an auto copyright bot that is flagging sound in my video that is in the public domain?
[23:58:24 CEST] <unomystEz> I tried changing the playback to 0.98 speed and 1.02 but that didn't seem to help
[23:58:38 CEST] <BtbN> dispute the claims?
[23:58:54 CEST] <unomystEz> BtbN: it's happening on all of my uploads
[23:58:57 CEST] <unomystEz> BtbN: http://www.theorchard.com/
[23:59:05 CEST] <unomystEz> that's the company who keeps flagging it
[23:59:14 CEST] <unomystEz> yes, I already sent them emails and disputed this
[23:59:31 CEST] <unomystEz> but was just curious if there might be effects to at least help mitigate it from flagging
[00:00:00 CEST] --- Mon Apr 17 2017