[Ffmpeg-devel-irc] ffmpeg.log.20170421

burek burek021 at gmail.com
Sat Apr 22 03:05:02 EEST 2017


[00:00:15 CEST] <BtbN> xhost local:
[00:00:18 CEST] <BtbN> is probably safer
[00:00:28 CEST] <thebombzen> does xhost local: disable xauth for localhost only?
[00:00:30 CEST] <djk> that script is working just fine at this point I can't mess with setting now
[00:00:38 CEST] <BtbN> it disables it for non-network connections.
[00:00:42 CEST] <thebombzen> that seems nice. I'm a singleuser system
[00:00:54 CEST] <llogan> djk: can you not use screen?
[00:00:59 CEST] <BtbN> it still makes sense to restrict access to X to the one user using it
[00:01:24 CEST] <thebombzen> I've had issues because my x server is owned by root
[00:01:43 CEST] <djk> what I like to understand is how to kick remotely because FB times out at 4 hours or just randomly without reason if unlimited
[00:01:48 CEST] <BtbN> that's who runs it normally
[00:01:55 CEST] <BtbN> rootless X is a pretty new development
[00:02:11 CEST] <djk> llogan: what do you mean by screen?
[00:02:17 CEST] <thebombzen> GNU Screen
[00:02:35 CEST] <thebombzen> djk: if you send SIGINT to ffmpeg's process it'll exit gracefully
[00:03:05 CEST] <llogan> what he said. alternatively, look into using the fifo muxer. it apparently allows reconnection in some cases.
[00:03:09 CEST] <djk> I have to update the key and then start the ffmpeg again.
[00:04:00 CEST] <djk> Facebook stops the stream (expires) so you have get a new key and start it up again
[00:04:27 CEST] <djk> I can kill the the running ffmpeg that isn't the problem
[00:05:24 CEST] <djk> I'd rather not have to be at the local machine or have to VNC to it. A quick ssh session from the phone to update the key and restart would be ideal.
[00:06:14 CEST] <llogan> why does fcaebook expire?
[00:07:00 CEST] <djk> good question they don't show an error and my stream is still running but if I select unlimited it just randomly without reason expires
[00:07:33 CEST] <djk> if I use the default that has a 4 hour limit it runs to the end no problem
[00:07:52 CEST] <djk> either way I need to get a new key and restart
[00:09:14 CEST] <llogan> is it only ffmpeg that has the same issue? IIRC, another user had to contact facebook regarding dropped connection. it was a problem on facebooks end...or maybe it was some CDN. can't remember.
[00:10:41 CEST] <djk> don't know I am only using ffmpeg to rtmp:// to FB
[00:12:56 CEST] <djk> llogan: so that user was fine on their end and FB had to fix it on their end or the CDN
[00:13:43 CEST] <djk> Now to find a way to get FB contact and not the blast to the community QA
[00:23:01 CEST] <llogan> djk: oh, it was edgecast CDN, not facebook
[00:28:40 CEST] <djk> so there was indication I might be able to do a remote start of the x11grab but I am missing the detail how I would do that with a remote ssh console session no GUI
[00:34:57 CEST] <thebombzen> djk: same way
[00:35:02 CEST] <thebombzen> run ffmpeg -f x11grab -i :0.0
[00:35:08 CEST] <thebombzen> you give the display # in the ffmpeg command
[00:36:05 CEST] <djk> so you're saying that command should run remote?
[00:36:22 CEST] <thebombzen> yea I'm saying your above command is the same
[00:36:53 CEST] <thebombzen> it doesn't need to be launched from the display, because you specify the display # in the command
[00:37:07 CEST] <thebombzen> you might need to set the DISPLAY environment variable tho
[00:37:29 CEST] <djk> ah that might be the issue DISPLAY
[00:37:31 CEST] <djk> Cannot open display :0.0+1280,10, error 1.
[00:37:31 CEST] <djk> :0.0+1280,10: Input/output error
[00:43:15 CEST] <djk> That looks like might work will be giving it a try on the next run. Thank you
[02:25:09 CEST] <JamEngulfer> I'm having the most *infuriating* time trying to do anything with a .mkv file I have. I am running "ffmpeg -i video.mkv" and just getting "Format matroska,webm detected only with low score of 1, misdetection possible! Read error at pos. 1 (0x1). EBML header parsing failed."
[02:26:12 CEST] <klaxa> well sounds like your file is corrupt
[02:27:06 CEST] <klaxa> does "file video.mkv" say anything useful?
[02:27:26 CEST] <JamEngulfer> The thing is, it plays absolutely fine on my phone
[02:27:49 CEST] <klaxa> huh, that's very weird
[02:28:03 CEST] <JamEngulfer> Nope, I get "file is not recognised as an internal or external command"
[02:28:15 CEST] <JamEngulfer> I'm on windows BTW, probably should have mentioned :P
[02:28:21 CEST] <klaxa> ah
[02:29:16 CEST] <klaxa> well there's this http://gnuwin32.sourceforge.net/packages/file.htm but not sure if it will help
[02:29:45 CEST] <klaxa> i'm inclined to think it's not actually an mkv file though
[02:30:13 CEST] <klaxa> if it's a video and plays on a phone, but has no valid ebml header, it might just be something else
[02:30:54 CEST] <JamEngulfer> Yeah, I suspected that
[02:31:10 CEST] <JamEngulfer> It's h265 encoded, I'm pretty sure
[02:31:38 CEST] <klaxa> well that shouldn't change anything though :)
[02:33:45 CEST] <JamEngulfer> I'm going to copy the file to OSX and prod it, I figure I might get better results
[02:35:39 CEST] <JamEngulfer> klaxa: Ok, running 'file' on it just tells me "data"
[02:38:29 CEST] <furq> well it's not technically wrong
[02:38:29 CEST] <klaxa> nice... uh...
[02:38:48 CEST] <klaxa> well it's the case default: string i'd guess
[02:40:21 CEST] <JamEngulfer> It's even playing on my phone using VLC...
[02:40:47 CEST] <furq> does osx come with xxd
[02:41:11 CEST] <furq> if so then paste the output of `xxd file.mkv | head -1`
[02:41:29 CEST] <JamEngulfer> I believe so, yes
[02:42:03 CEST] <JamEngulfer> "0000000: 0000 0000 0000 0000 0000 0000 0000 0000  ................"
[02:42:11 CEST] <furq> lol
[02:42:12 CEST] <furq> nice
[02:42:53 CEST] <JamEngulfer> Well, I guess we know that much...
[02:43:45 CEST] <thebombzen> what's the filesize?
[02:44:00 CEST] <thebombzen> if it's small enough you could upload it to 0x0.st
[02:44:14 CEST] <thebombzen> and file never reports anything useful abouta matroska file
[02:44:29 CEST] <thebombzen> it'll just say "matroska data" if it even detects it as matroska
[02:44:29 CEST] <furq> er
[02:44:42 CEST] <furq> how is that not useful
[02:45:04 CEST] <thebombzen> or rather, it doesn't provide any information other than that it's a matroska file
[02:45:11 CEST] <thebombzen> but it does for nearly every other file format
[02:46:49 CEST] <thebombzen> such as something like, JPEG image data, JFIF standard 1.01, aspect ratio, density 1x1, segment length 16, progressive, precision 8, 1280x720, frames 3, or like ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=b605512bd055497aeb40c6b285535504a0b3d354, stripped, with debug_info
[02:48:14 CEST] <furq> thank you for this wonderful information
[02:55:13 CEST] <thebombzen> no thank file
[02:55:40 CEST] <JamEngulfer> Hmm, apparently 99.997% of the file is blank data
[02:59:50 CEST] <JamEngulfer> I mean, that *would* explain why it wasn't working
[03:02:53 CEST] <JamEngulfer> I just copied the file back off of my phone and it's magically transformed into a working .mp4 file that's a bit larger. It works now, I'm not going to question any of this
[03:03:25 CEST] <JamEngulfer> Thanks for the help though!
[03:08:55 CEST] <klaxa> heh wow
[03:13:43 CEST] <Afshaal> Well JEEB, I managed to finally compile untrunc (after loads of compile warnings from both libav-0.8.7 and untrunc itself), and...  it managed to recover my audio at least.  Video recovery is pretty garbled though.
[07:29:38 CEST] <Mista_D> Is that supoposed to take longer: "-ss $t -i $file" VS. "-ss $t -f concat -i $file.concat" ??? Wondering why is seeking much slower with concat...
[08:00:09 CEST] <Kadigan> Hm... It -sounds- like it needs to use slow (ie. output) seek in that case, which is slower.
[08:01:01 CEST] <Kadigan> Considerably, as in.
[09:49:12 CEST] <delicado> hi,  how to build ffmpeg in release mode? is --disable-debug enough? what else?
[10:06:00 CEST] <JEEB> delicado: it's release by default
[10:06:08 CEST] <JEEB> --enable-debug is required for optimizations to be disabled
[10:15:42 CEST] <delicado> JEEB: I tried configure without flags, it built an static library and linked to my program, now my program is 80MB+.. Is this correct size when statically linking ffmpeg? maybe it should be around 30MB?
[10:16:03 CEST] <tdr> hehe
[10:17:30 CEST] <tdr> were you wanting a dynamic one?  --disable-static  will do that
[10:19:47 CEST] <delicado> tdr: thanks, i wanted to linked statically though. I just building with --disable-debug, and I see it has reasonable size now, libavcodec.a was 79MB before, now it's 13MB. I think it's correct size now
[10:19:56 CEST] <delicado> *finished
[10:24:51 CEST] <tdr> yeah that sounds a bit less crufty
[10:24:52 CEST] <delicado> thanks guys, my program is down to 15MB.
[10:27:13 CEST] <delicado> indeed
[10:28:47 CEST] <JEEB> tdr: you mean --enable-shared
[10:28:57 CEST] <JEEB> and yes, there's debug symbols
[10:29:23 CEST] <tdr> JEEB, also that.  i think if you do the other it defaults to shared tho
[10:29:44 CEST] <JEEB> might be since one *has* to be enabled
[10:29:48 CEST] <tdr> yeah
[10:30:10 CEST] <tdr> i just looked at ./configure --help | grep static and got my option
[10:46:26 CEST] <tommy``> hello
[10:49:57 CEST] <tommy``> ffmpeg -i input.mkv -sameq -r 24 -y input2.mkv <---- is this good to change fps without reeconding?
[10:52:46 CEST] <durandal_1707> remove sameq
[10:53:28 CEST] <durandal_1707> you cant change fps without reencoding iirc
[10:54:54 CEST] <tommy``> ah
[10:55:20 CEST] <tommy``> i've 29fps and i want to convert 23.976
[11:10:08 CEST] <thebombzen> tommy``: do you actually want to run an inverse telecine?
[11:10:15 CEST] <thebombzen> or do you just want to drop 30 to 24
[11:10:33 CEST] <tommy``> inverse telecine thebombzen
[11:11:40 CEST] <thebombzen> check the documentation on the fieldmatch filter
[11:12:07 CEST] <durandal_1707> or qtgmc?
[11:12:26 CEST] <tommy``> ok thebombzen
[11:13:03 CEST] <thebombzen> there's also the pullup filter
[11:13:14 CEST] <thebombzen> I'm not sure which one works best, but those are a few
[11:13:54 CEST] <tommy``> https://ffmpeg.org/ffmpeg-filters.html#fieldmatch <------
[11:14:58 CEST] <tommy``> thebombzen what0s the difference from IT and drop 30 to 24?
[11:16:12 CEST] <notdaniel> -vf pullup,fps=24000/1001
[11:16:17 CEST] <notdaniel> should get you there
[11:17:26 CEST] <tommy``> i've this mkv with 29fps and the audio is 23,976 from another video, i've tried with mkvmerge but there is desync using assumefps 23,976 fro video
[11:17:45 CEST] <tommy``> notdaniel i'll try that too
[11:23:50 CEST] <BtbN> The audio is 23,976? That's not how audio works.
[11:24:16 CEST] <tommy``> audio is taken from video 23,976
[11:24:31 CEST] <BtbN> the audio doesn't care about the framerate of the video
[11:25:22 CEST] <BtbN> if you convert the framerate of some video, the audio does not need to be altered at all
[11:25:24 CEST] <tommy``> usually when i have 25fps video and audio from 23,976 video i stretch by 23,976/25 = 0,95904 value
[11:25:29 CEST] <BtbN> unless you speed up or slow down the video
[12:06:42 CEST] <zxd> hi
[12:06:52 CEST] <zxd> can someone give me a quick line   to convert a video to H.264 video up to 720p, 30 frames per second
[12:07:45 CEST] <zxd> H.264 video up to 720p, 30 frames per second, Main Profile level 3.1
[12:08:36 CEST] <c_14> ffmpeg -i video -c:v libx264 -vf scale=-2:720,fps=30 -profile:v main -level 3.1 out.mp4
[12:26:47 CEST] <zxd> hey thanks c_14
[12:27:15 CEST] <zxd> I am using avconv fork of ffmpeg only thing I have in debian 8
[12:27:21 CEST] <zxd> [scale @ 0x1d2b620] The <w>:<h>:flags=<flags> option syntax is deprecated. Use either <w>:<h>:<flags> or w=<w>:h=<h>:flags=<flags>.
[12:27:24 CEST] <zxd> [scale @ 0x1d2b620] Size values less than -1 are not acceptable.
[12:27:27 CEST] <zxd> [scale @ 0x1d2b620] Failed to configure output pad on Parsed filter 0 scale
[12:27:29 CEST] <zxd> Error opening filters!
[12:27:32 CEST] <zxd> I get this error
[12:27:46 CEST] <sfan5> you should install ffmpeg
[12:28:20 CEST] <sfan5> but assuming your vid is 16:9 you can use -vf scale=1280:720,fps=30 instead
[12:28:51 CEST] <c_14> either upgrade debian or grab a static build
[12:28:53 CEST] <c_14> http://johnvansickle.com/ffmpeg/
[12:48:12 CEST] <zxd> k
[12:48:29 CEST] <zxd> how to remove the audio all together don't need the audio track in the video
[12:48:37 CEST] <zxd> from the file being converted
[12:51:06 CEST] <dystopia_> add -an to the line
[12:51:30 CEST] <dystopia_> -an ignores audio, -sn ignores subtitles, -vn ignores video
[12:51:47 CEST] <zxd> cool
[14:35:34 CEST] <sdf> Hi, i need to help. i want to optimize this ffmpeg script https://pastebin.com/66ck8jK7
[14:38:18 CEST] <DHE> umm.. I see -an which means discard audio in the second command, yet you've explicitly pulled audio in the first. also you've got -vcodec twice
[14:39:04 CEST] <DHE> after -map [a] you can just put all your secodn command options into the first command before the output filename and do it all in one shot
[14:44:24 CEST] <sdf> DHE yea i just noticed i am missing an audio
[14:44:45 CEST] <sdf> DHE thanks and is there any way that i can make it faster
[14:44:49 CEST] <sdf> it seems so slow
[14:48:50 CEST] <durandal_1707> slow is because of preset right?
[14:50:50 CEST] <furq> most of the options for preset slow will be overriden by profile baseline
[14:50:55 CEST] <furq> so i'd expect that to be pretty quick
[14:51:06 CEST] <DHE> motion estimation searches will still be pretty hard
[14:51:08 CEST] <furq> also don't use baseline unless you really need to
[14:51:34 CEST] <DHE> well, libopenh264 is baseline only. depends on what you're targeting... :/
[14:52:06 CEST] <DHE> still, for any other application yeah, main and high are actually really well supported on modern devices and hardware
[14:52:12 CEST] <sdf> hey furq, remember me
[14:52:14 CEST] <sdf> i am furqan
[15:47:01 CEST] <cdidd> hi, when I reencode high-bitrate files into hevc with crf=18 I get good compression ratio, when I do it with lower bitrate videos, the resulting compression ratio is noticeably worse, sometimes video becomes even bigger than it was before
[15:48:01 CEST] <cdidd> Is there a way to avoid spending bitrate encoding blocking artifacts, like an automatic quality detector or something?
[15:49:45 CEST] <DHE> no. it's trying to encode what it sees. there are filters to try and reduce mpeg artifacts but they may negatively impact image quality in other ways.
[15:49:54 CEST] <furq> you could not reencode them
[15:50:31 CEST] <cdidd> furq what do you mean I can't reencode them?
[15:51:02 CEST] <DHE> what are you doing that requires a codec change into h265?
[15:51:25 CEST] <DHE> the best way to preserve quality is to do nothing to the originals
[15:51:30 CEST] <cdidd> trying to save couple of terabytes of space
[15:52:43 CEST] <cdidd> Then there is at least should be a way to automatically detect which files are worth to reencode, and which are not.
[15:53:11 CEST] <cdidd> It's definitely worth to reencode 12k kb/s 1080p h264 video
[15:55:51 CEST] <cdidd> Any ideas?
[16:54:45 CEST] <logicom> how to remove locked audio from the video stream? I have rewraped 4 separate mono channels with the video in MOV but one that was interleaved within DV stream remains. Is it possible to remove this without retranscoding?
[16:54:58 CEST] <logicom> here is the mediainfo:
[16:55:06 CEST] <logicom> Audio #5 ID                                       : 1-0 Format                                   : PCM Muxing mode                              : DV Muxing mode, more info                   : Muxed in Video #1 Duration                                 : 29 min Bit rate mode                            : Constant Bit rate                                 : 1 536 kb/s Encoded bit rate                         : 0 b/s Channel(s)            
[17:02:32 CEST] <mbparsa> aham
[17:12:48 CEST] <durandal_1707> logicom: currently not afaik
[17:45:19 CEST] <pihpah> find . -name "*.mkv" -type f -exec ffmpeg -i {} -map 0:s:4 {}.srt \; does not extract subtitles. What's wrong?
[17:45:23 CEST] <pihpah>     Stream #0:4(eng): Subtitle: dvd_subtitle, 1440x1080
[17:45:57 CEST] <furq> dvd_subtitle is an image subtitle format
[17:46:05 CEST] <furq> ffmpeg can't convert that to text, you need to ocr it
[17:46:44 CEST] <pihpah> furq: what is ocr?
[17:46:59 CEST] <furq> https://en.wikipedia.org/wiki/Optical_character_recognition
[17:47:20 CEST] <pihpah> Ah
[17:47:35 CEST] <pihpah> So ffmpeg won't help me in this case.
[17:48:17 CEST] <furq> there are cli tools which do it but i don't know how good they are
[17:48:38 CEST] <furq> vobsub2srt is one, although i don't think ffmpeg has a vobsub muxer
[18:03:42 CEST] <durandal_1707> ffmpeg have ocr filter
[18:08:09 CEST] <faLUCE> hello, how can I convert av_gettime_relative() to date (year + day + hh + minutes + seconds) ?
[18:33:15 CEST] <sefen> printf 'file %s\n' * | ffmpeg -f concat -i - -c copy out.mkv
[18:33:17 CEST] <sefen> Protocol not on whitelist 'crypto'!
[18:33:46 CEST] <sefen> it used to work...
[19:24:02 CEST] <josemslopes> Hello
[19:25:04 CEST] <josemslopes> i am trying to use Freeswitch as Video Conference  with ffmpeg  and nvidia card ... but i can only do 6 video calls
[19:26:38 CEST] <josemslopes> * i am trying to use Freeswitch as Video Conference  with ffmpeg  and nvidia card ... but i can only do 6 video conferences , when i start the 7 video conference,  i get an error
[19:26:50 CEST] <josemslopes> it get an error from nvenc.c file  Failed creating CUDA context for NVENC: 0x3e7
[19:27:30 CEST] <josemslopes> i analyse the code 0x3e7 and it means   unsupported device (2)
[19:28:22 CEST] <josemslopes> there is any reason to i have only 6 video conferences?  ... there is some configuration ?
[19:29:47 CEST] <josemslopes> i tried to run simultaneos commands of ffmpeg using nvenc ... and i didn't get this error
[19:30:10 CEST] <c3r1c3-Win> josemslopes: Sounds like a NVidia limitation.
[19:30:48 CEST] <josemslopes> but with multiple ffmpeg commands  (~30 commands) running at same time ... i don't have this issue
[19:31:49 CEST] <josemslopes> c3r1c3-Win: but with multiple ffmpeg commands  (~30 commands) running at same time ... i don't have this issue.
[19:31:53 CEST] <llogan> using the same commands as whatever this freeswitch thing is using? or is it using the libraries and not cmd tool?
[19:32:18 CEST] <josemslopes> yes i am using the same libs ... i am using ffmpeg 3.2
[19:32:58 CEST] <josemslopes> i put a log on source code of libavcodec ... and i saw that log using ffmpeg and freeswitch
[19:36:30 CEST] <josemslopes> the diference that i notice is ... with freeswitch it is the same process invoking multiple times the encoder.  using multiples commands of ffmpeg it is multiple commands each invoking encoder
[19:36:57 CEST] <llogan> i guess you're using the libraries incorrectly. blame freeswitch?
[19:37:28 CEST] <josemslopes> i compile freeswitch to use the same libraries :)
[19:40:05 CEST] <josemslopes> cu_res = dl_fn->cu_ctx_create(&ctx->cu_context_internal, 0, cu_device)  -> this is where, after 6 video conference, stop to use nvenc for video conference
[20:11:05 CEST] <BtbN> nvenc on consumer cards is limited to two concurrent encode sessions
[20:11:28 CEST] <BtbN> that's a driver limitation, nothing some magic flag could change
[20:13:37 CEST] <thebombzen> GeForce cards have quadro-capable chips but they limit the functionality in the software and not the hardware yea
[20:13:58 CEST] <thebombzen> this means if you really want to tinker you could trick the driver by using really lowlevel stuff but it's not worth it
[22:54:06 CEST] <Tatsh> hmm, piping with vspipe to ffmpeg and the audio goes out of sync
[22:54:11 CEST] <Tatsh> by about 1.5 seconds
[22:54:18 CEST] <Tatsh> not sure if this is a constant throughout the stream
[22:54:45 CEST] <Tatsh> anyone familiar with vapoursynth and seeing this sort of thing? i'm muxing the audio at the same time as using vspipe
[22:54:49 CEST] <JEEB> you can find the vs author on #avisynth
[22:55:00 CEST] <JEEB> and no, I haven't seen that sort of behavior
[22:55:05 CEST] <JEEB> although I haven't utilized audio with it
[22:55:12 CEST] <JEEB> did vspipe ever get NUT support for it?
[22:55:18 CEST] <JEEB> I know it does y4m for video
[22:57:35 CEST] <Tatsh> if this gives an idea
[22:57:35 CEST] <Tatsh> https://gist.github.com/Tatsh/6301c2c24099687ff321804e4ad61f2f
[22:58:37 CEST] <JEEB> right
[22:58:40 CEST] <Tatsh> it's more that with QTGMC which doubles the framerate
[22:58:44 CEST] <Tatsh> not sure if that's the exact reason
[22:58:52 CEST] <Tatsh> but perhaps it's not an exact doubling
[22:59:05 CEST] <JEEB> in any case, you have your timestamps coming from two separate sources
[22:59:15 CEST] <JEEB> so something's going bonkers there
[22:59:18 CEST] <Tatsh> ok
[22:59:21 CEST] <Tatsh> would the delay be a constant?
[22:59:27 CEST] <Tatsh> or would it shift throughout the video?
[22:59:34 CEST] <JEEB> most likely unless stuff goes VFR
[22:59:41 CEST] <JEEB> because y4m is constant frame rate
[22:59:43 CEST] <Tatsh> that's my main concern; if it's a constant i can figure out the constant and set it
[22:59:46 CEST] <Tatsh> in -af
[23:00:33 CEST] <Tatsh> i hate audio delays though because there's no such thing as perfect when you adjust it
[23:01:06 CEST] <Tatsh> when y4m does its processing it creates new timestamps that are not guaranteed to match up
[23:13:34 CEST] <Tatsh> since it's a negative offset, i can't use adelay
[23:13:39 CEST] <Tatsh> is asetpts the correct filter?
[23:13:57 CEST] <Tatsh> -af 'asetpts=PTS-(1.5/TB)'
[23:14:10 CEST] <JEEB> I'd probably use -itsoffset paired with -vsync passthrough -copyts or so
[23:14:19 CEST] <JEEB> that way no transcoding of audio is needed
[23:15:19 CEST] <Tatsh> -copyts copies timestamps from the y4m stream right?
[23:16:15 CEST] <JEEB> it disables some timestamp meddling with the input its set to
[23:16:22 CEST] <JEEB> -vsync passthrough is another similar option
[23:20:54 CEST] <Tatsh> so i used -copyts and -itsoffset
[23:20:56 CEST] <Tatsh> works good
[23:21:18 CEST] <Tatsh> ffmpeg -thread_queue_size 4096 -copyts -i pipe: -itsoffset -1.5 -i raw2/fa.mkv -preset slow -crf 0 -pix_fmt yuv420p -aspect 4:3 -codec:a copy -copyts -map 0:0 -map 1:1 raw2/fa-progressive.mkv
[23:21:23 CEST] <Tatsh> that make sense?
[23:21:53 CEST] <Tatsh> i hate that i have to do -crf 0 on this because the file is huge, but i need 2-pass to get a close to valid bitrate for youtube uploading (2.5 Mbps for 480p)
[23:22:09 CEST] <Tatsh> i don't see how i would do 2-pass without running qtgmc twice
[23:22:16 CEST] <Tatsh> otherwise
[23:27:36 CEST] <Tatsh> i wonder if this, because the video source is basically the same format, would the -1.5 seconds be constant for all my QTGMC processing
[23:31:29 CEST] <llogan> "valid bitrate". youtube isn't picky and it will re-encode whatever you feed it. use a single pass with -crf 18 or so.
[23:33:24 CEST] <furq> yeah unless you're livestreaming then any bitrate is fine
[23:33:30 CEST] <furq> also fwiw i've never had audio sync issues when using qtgmc
[23:34:18 CEST] <furq> maybe there's an offset in the source file
[23:35:53 CEST] <furq> also you can just ignore the -thread_queue_size warning
[23:36:13 CEST] <furq> i'm not really sure why vspipe causes that but it doesn't seem to indicate an actual problem
[00:00:00 CEST] --- Sat Apr 22 2017


More information about the Ffmpeg-devel-irc mailing list