January 2019 - Ffmpeg-devel-irc

ffmpeg-devel.log.20190105
by burek 06 Jan '19

06 Jan '19

[09:43:22 CET] <cone-392> ffmpeg 03Paul B Mahol 07master:f2e245629466: avfilter/af_afir: adjust min partition size [12:18:48 CET] <durandal_1707> atomnuker: how to reduce ssd calculation overlap? [13:47:08 CET] <atomnuker> I'd grab a pen and paper, figure out which parts of the buffer each step uses for 2 small sizes and see what the pattern is [22:37:05 CET] <cone-223> ffmpeg 03Michael Niedermayer 07master:2d580d727cf2: avcodec/gdv: Optimize unscaled loop [23:29:25 CET] <ubitux> any reason why the avutil blend api isn't used in the overlay filter? [23:30:02 CET] <ubitux> i was looking for some kind of blend utils for the subtitles overlay and naively checked vf overlay but it seems not to be using any shared code [23:30:13 CET] <ubitux> otoh i'm seeing ff_blend_rectangle & friends [23:30:24 CET] <ubitux> maybe the pixel format support differing? [23:43:01 CET] <durandal_1707> overlay code is optimized [23:45:11 CET] <ubitux> and i guess transfering the optimization from lavfi to lavu would complexify the code? [23:45:28 CET] <ubitux> actually the drawutils are in lavfi [23:45:35 CET] <ubitux> from overlay to drawutils [23:45:50 CET] <ubitux> complexify the api* [23:48:02 CET] <durandal_1707> special vs generic [00:00:00 CET] --- Sun Jan 6 2019

1 0

ffmpeg.log.20190105
by burek 06 Jan '19

06 Jan '19

[02:46:47 CET] <cluelessperson> hey, I want to convert a lossless h264 video to a lossless compressed format? [02:46:49 CET] <cluelessperson> or is that possible? [02:46:52 CET] <cluelessperson> what do you suggest? [03:28:26 CET] <furq_> cluelessperson: lossless h264 is already compressed pretty well [03:39:49 CET] <kepstin> you can of course convert from any lossless compressed codec to any other lossless compressed codec without loss. [03:40:09 CET] <kepstin> if you'd prefer it to be in ffv1 or something instead. It might be bigger or smaller depending on content. [03:40:49 CET] <kepstin> (if you have a *lossy* h264 video, don't do that - it'll just become huge for no reason) [04:15:17 CET] <cluelessperson> furq_: well this it's probably lossy to begin with [04:15:26 CET] <cluelessperson> I like to convert small clips to share into webm [11:57:28 CET] <strk> how can I create a video from an audio file and a set of images ? Ideally with transition for images [12:05:01 CET] <ariyasu> https://trac.ffmpeg.org/wiki/Slideshow#Addingaudio [12:13:09 CET] <strk> and, how to reduce quality of audio ? I have a .flac but it is too big to publish [12:13:43 CET] <durandal_1707> strk: transcode to lossy codec [12:17:11 CET] <strk> ok, doing that part, now back to the slideshow [12:18:49 CET] <strk> ffmpeg -framerate 1/5 -i Img/pic%04d.jpg -i in.mp3 out.avi # generates a file that VLC considers broken [12:19:09 CET] <strk> maybe this is related: [swscaler @ 0x55654ee8bda0] deprecated pixel format used, make sure you did set range correctly [12:19:16 CET] <durandal_1707> nope [12:19:25 CET] <strk> vlc says: Because this file index is broken or missing, seeking will not work correctly. [12:19:46 CET] <strk> pSelecting "Play as is" still complains: [00007fe838c178a0] avi demux error: no key frame set for track 0 [12:31:44 CET] <strk> I don't understand that wiki page, it mentions "adding audio" but it doesn't add the audio to the slideshow animation, only to a still image ? [12:31:52 CET] <strk> ie the "poster" image [12:40:08 CET] <strk> I'm copying some instructions elsewhere, they mention using -b 4M, but ffmpeg says: Please use -b:a or -b:v, -b is ambiguous [12:45:06 CET] <strk> odd, mplayer plays it "fine" [12:48:15 CET] <strk> Stream #1:1: Video: png, rgb24(pc), 700x700 [SAR 1:1 DAR 1:1], 90k tbr, 90k tbn, 90k tbc [12:48:24 CET] <strk> but the input is made of .jpg, can that be the problem with vlc ? [13:08:38 CET] <strk> can I ask ffmpeg to "fill" input images that are not in the output size ? [13:09:00 CET] <strk> or do I have to do this before putting them togheter ? [14:55:12 CET] <strk> how do I add an audio stream to an mp4 ? [14:56:27 CET] <furq_> strk: -i foo.mp4 -i bar.m4a -map 0 -map 1:a -c copy baz.mp4 [14:57:33 CET] <strk> does that work with mp3 input ? [15:05:05 CET] <furq_> sure [15:08:59 CET] <strk> cool, I'm almost there, complex generator script but very close [15:09:21 CET] <strk> what's the switch to force already exists. Overwrite ? [15:11:49 CET] <furq> -y [15:13:51 CET] <strk> thanks [15:17:57 CET] <strk> how to deal with [swscaler @ 0x5578173a8560] deprecated pixel format used, make sure you did set range correctly [15:50:18 CET] <th3_v0ice> Guys, what does this error returned from decoder mean: "Error number -16976906 occurred"? [15:57:40 CET] <strk> alright, I'm all set now [16:23:38 CET] <strk> (how) can I extract length of an audio file via ffmpeg ? [16:23:40 CET] <strk> in seconds [16:26:24 CET] <strk> ffmpeg -i thing.mp3 2>&1 | grep Duration # gets me close [16:47:18 CET] <Mavrik> strk, use ffprobe and one of its structured outputs [16:47:28 CET] <Mavrik> never parse user-facing / debut output of tools, c'mon :) [16:48:28 CET] <Peetz0r> Hey! I'm trying to mix some different inputs (x11grap, mjpeg stream, v4l2 device, and alsa audio) into one stream and stream that to twitch. All this on a Raspberry Pi 3A+ using h264_omx. This is now 50% working. [16:49:28 CET] <Peetz0r> But I'm having some issues with performance and audio. I think that one of the problems might be that my mjpeg source has a very irregular framerate. I have tried using the fps filter to fix it at 30fps, but that has only reduced performance even further [16:49:53 CET] <BtbN> Decoding the mjpeg stream might be too much for the CPU [16:50:08 CET] <BtbN> Specially paired with all the other stuff you make it do [16:50:57 CET] <Peetz0r> When I make it only transcode the mjpeg thing into h264 and stream that to twitch, performance is finw [16:51:10 CET] <Peetz0r> When it's running all the things, only 1 cpu core is loaded, the other 3 are mostly idle. So I guess there's only 1 thread in ffmpeg doing muhc work [16:51:18 CET] <Peetz0r> Is there a way to find out what that would be? [16:51:31 CET] <BtbN> Yes, ffmpeg.c is entirely single-threaded. [16:51:46 CET] <BtbN> And the task isn't something that could be easily multithreaded [16:52:09 CET] <BtbN> So it's very unlikely a Pi is ever going to be able to do all that [16:52:37 CET] <Peetz0r> aahw [16:52:43 CET] <Peetz0r> but it's coming pretty close actually [16:53:42 CET] <Peetz0r> But if fmpeg is singlethreaded, what does the -threads option do? [16:54:56 CET] <Peetz0r> brb putting all my stuff in a git repo to make my life easier [16:55:15 CET] <BtbN> ffmpeg.c is single threaded [16:55:27 CET] <BtbN> individual codecs/filters aren't. [16:56:30 CET] <strk> Mavrik: the dirty way is working fine for now. I just gave ffprobe a go but it outputs the non-structured output by default (same as ffmpeg -i ?) [16:57:25 CET] <Mavrik> Yes, there's a parameter to give you a structured output. [16:57:47 CET] <Mavrik> "just fire for now" means it'll just break if you feed it a slightly different file that will produce a slightly different output [16:57:51 CET] <Mavrik> not to mention if you update ffmpeg [16:58:42 CET] <Peetz0r> Good, because I use a few filters [16:59:08 CET] <BtbN> Not much of what you said you use uses more than one thread [16:59:23 CET] <BtbN> Specially x11grab is dirt slow [16:59:35 CET] <Peetz0r> Oh, maybe I should try disabling that [16:59:47 CET] <Peetz0r> I was expecting x11grab to be the least of my problems [17:00:05 CET] <BtbN> It's struggling to hit 30 fps on a proper PC [17:00:13 CET] <BtbN> Oh higher resoltuions at least [17:00:46 CET] <Peetz0r> would there be another way to have a webbrowser screen capture in ffmpeg without x11? [17:01:18 CET] <Peetz0r> Lemme first try this whole thing with x11grab turned off then [17:06:28 CET] <Peetz0r> it looks like an improvement so far [17:07:20 CET] <Peetz0r> I still get output like this: [mpjpeg @ 0xa75e30] Thread message queue blocking; consider raising the thread_queue_size option (current value: 8) [17:07:37 CET] <Peetz0r> would increasing that actually help (it didn't before, but I don't really know what I'm doing) [17:08:40 CET] <Peetz0r> yeah, it's still maxing out 1 cpu core [17:09:18 CET] <Peetz0r> This is my current commandline: https://paste.sigio.nl/view/d625ddda [17:10:30 CET] <Peetz0r> but if the mjpeg stream is holding me back, does anyone have a better way to capture an android phone screen? [17:17:16 CET] <Peetz0r> I am currently using https://play.google.com/store/apps/details?id=info.dvkr.screenstream, which does mjpeg and sends one frame whenever something happens on screen [17:17:42 CET] <Peetz0r> Which probably means it's trying to do 60fps [17:18:27 CET] <BtbN> Those 3 overlay filters alone will max out a CPU core. [17:19:11 CET] <BtbN> plus crop, plus transpose, plus decoding mjpeg [17:20:07 CET] <Peetz0r> But those filters do have their own threads, right? [17:20:34 CET] <BtbN> no [17:20:48 CET] <BtbN> They are executed on the main thread sequentially [17:20:53 CET] <Peetz0r> oh, that sucks :\ [17:21:07 CET] <BtbN> Each depends on the results of the previous, how would you multi-thread that? [17:21:09 CET] <DHE> (it's a feature I've wanted to develop..) [17:21:21 CET] <DHE> you can pipeline, having each filter work on a frame at a time and pass them along [17:21:48 CET] <Peetz0r> DHE: anything I can do to help? [17:21:48 CET] <BtbN> With such a complex setup with 3 parallel chains that get merged again, that seems extremely hard to get right [17:22:09 CET] <Peetz0r> (I'm not really a developer) [17:22:12 CET] <BtbN> And even with multithreading I have high doubts a Pi could do that [17:22:31 CET] <BtbN> It's just really really slow [17:22:49 CET] <BtbN> The slowest x86 will beat it by magnitudes [17:23:04 CET] <Peetz0r> the latest Pi's are quite powerful, and so far I haven't used nearly all of the performance that's available [17:23:19 CET] <Peetz0r> but heah, threading is hard, and I have maxed out it's singlecore performance I guess [17:23:37 CET] <BtbN> It's 1.4GHz 32bit ARM still [17:23:59 CET] <BtbN> x86_64 will trash it [17:25:28 CET] <Peetz0r> isn't the A53 a 64 bit arm? [17:25:33 CET] <BtbN> The CPU is 64 bit capable, the OS and firmware usually isn't. [17:25:52 CET] <Peetz0r> that's lame :p [17:26:12 CET] <BtbN> I think there is no 64bit driver for the GPU, so no omx or anything on 64bit [17:26:38 CET] <Peetz0r> but the omx part seems to perform fine as it is right now [17:28:09 CET] <strk> how can I specify mp4 title ? What I can do with VLC [17:28:22 CET] <strk> Mavrik: I switched to ffmprobe meanwhile :) [17:32:49 CET] <Peetz0r> meh, raspbian is 32 bit only for now indeed [17:34:26 CET] <Peetz0r> brb trying https://raspberrypi.stackexchange.com/a/77707 [17:34:47 CET] <Peetz0r> I guess that will only give me the 64 bit kernel and I'll have to build a 64bit ffmpeg myself? [17:35:05 CET] <BtbN> You won't have omx if you forcefully use a 64bit userland [17:35:13 CET] <Peetz0r> that would be bad [17:35:26 CET] <BtbN> The driver is closed source and 32bit only [17:36:00 CET] <Peetz0r> h264 without omx sounds like a terrible thing to even try [17:36:21 CET] <Mavrik> 64-bit on RPi is just a headache and nothing more [17:36:25 CET] <Mavrik> Why do you need it? :) [17:36:53 CET] <Peetz0r> trying to squeeze more singlecore performance out of it [17:37:00 CET] <Peetz0r> but it probably won't help me at all [17:37:08 CET] <Peetz0r> losing h264_omx support would kill me [17:37:42 CET] <Peetz0r> I still believe that the Pi 3A+ should be capable of what I am doing... somehow [17:38:10 CET] <Peetz0r> but if it isn't, is there any other sub-¬50 singleboard device that you guys would recommend? [17:39:00 CET] <Peetz0r> (needs to be somewhat power efficient as I'm going to run it from a powerbank - but it's not too critical, it's a 100 Wh multivoltage powerbank) [17:39:04 CET] <BtbN> I usually use Intel NUCs for everything that has to be small scale [17:39:24 CET] <BtbN> Experience with Cheap Boards has been to horrible every single time, so I gave up on cheap. [17:39:25 CET] <Peetz0r> Yeah, but those are somewhat above my budget for this project :( [17:39:47 CET] <BtbN> Get a used Xeon Server from ebay [17:40:08 CET] <Peetz0r> yeah and carry that in my backpack together with a nuclear reactor from a submarine or somthing? [17:40:14 CET] <Peetz0r> sorry :p [17:40:24 CET] <BtbN> You never mentioned anything about mobile. [17:40:28 CET] <BtbN> In that case use a phone? [17:40:30 CET] <Peetz0r> I actually have a used 2008-era dual-xeon thing at home :D [17:41:09 CET] <Peetz0r> could I run vanilla ffmpeg on a non-rooted android phone? and connect a usb wemcam to it? [17:41:17 CET] <Peetz0r> because if I can, that would work [17:41:45 CET] <BtbN> no idea about USB Webcam, but a lot of apps use ffmpeg [17:41:49 CET] <Peetz0r> (my endgoal is a portable Twitch mobile gaming setup based on 2 phones, 1 webcam, the Pi, a huge powerbank and an unlimited 4G plan) [17:42:47 CET] <BtbN> That's most likely not going to happen for cheap [17:43:03 CET] <BtbN> The few people I know that do that have a fully powered PC on their back [17:43:38 CET] <BtbN> And there's a commercial product for that very purpose. They sell a backpack that does it, for 6k $ [17:43:53 CET] <Peetz0r> ew [17:44:21 CET] <Peetz0r> I could try rigging my laptop to not overheat while in my laptop somehow [17:44:27 CET] <BtbN> A more cheapish setup usually involved streaming to your home PC from the phone, and doing all the heavy work there [17:44:30 CET] <Peetz0r> eh, backpack* [17:44:41 CET] <Peetz0r> I don't really have a home pc :p [17:45:02 CET] <BtbN> Well, you will need something powerful to do the heavy lifting, somewhere [17:45:10 CET] <Peetz0r> I have the nuc which is the most low power one with a 2014 celeron, and I have my laptop with a i5 7200U [17:45:33 CET] <BtbN> They will both outperform the Pi [17:45:35 CET] <Peetz0r> the laptop could do it, but then it wil lneed to survive running in my backpack without thermal issues [17:45:59 CET] <Peetz0r> so I guess I'll buy a 2nd backback and modify it with usb powered fans and airducts :p [17:47:01 CET] <BtbN> You usually want something that's on a stable connection, so the stream doesn't die every 5 minutes [17:49:42 CET] <furq> Peetz0r: if the mjpeg input is from a usb2 camera then you might want to try the offical picam thingy [17:49:56 CET] <furq> iirc that isn't over usb2 so you can get 720p30 out of it without having to decode mjpeg [17:50:37 CET] <furq> mmal on a pi does support decoding mjpeg but i don't think ffmpeg supports it yet [17:50:46 CET] <Peetz0r> no, the mjpeg is from my phone capturing Pokemon Go using this app: https://play.google.com/store/apps/details?id=info.dvkr.screenstream [17:50:57 CET] <furq> oh fun [17:51:27 CET] <furq> well yeah if you can get rawvideo or h264 out of that or a similar app then that's worth looking into [17:51:28 CET] <Peetz0r> THis is what I am trying: https://www.twitch.tv/videos/359206040 [17:51:31 CET] <BtbN> The GPU will be busy encoding h264 anyway, but not of much help decoding [17:51:42 CET] <furq> i remember other people in here have had terrible issues trying to decode mjpeg in realtime on a pi [17:51:49 CET] <Peetz0r> but with smooth 30fps, and audio :D [17:51:57 CET] <BtbN> The Pi only has USB interfaces as well [17:52:04 CET] <furq> the camera interface isn't usb iirc [17:52:10 CET] <BtbN> so more than 100Mbit/s isn't going to come in stable via network [17:52:12 CET] <furq> but yeah everything else is kind of bad [17:52:19 CET] <furq> you won't even get 100mbps on a 3A [17:54:32 CET] <Peetz0r> my current camera is a usb webcam. would using the official pi cam give me better performance, or would it be the same? [17:54:53 CET] <Mavrik> I don't get why you don't use builtin streaming on your phone :( [17:54:54 CET] <Peetz0r> I also don;t need 720p from the webcam, 1/3rd of that is fine. But I do need it rotated [17:55:13 CET] <furq> Mavrik: i'm guessing he wants to do some trick overlay stuff [17:55:14 CET] <Peetz0r> Mavrik: because that won't let me have a layout like that [17:55:25 CET] <furq> you could maybe try running ffmpeg on your phone [17:55:27 CET] <furq> if it's rooted [17:55:35 CET] <Mavrik> That'll eat away all the CPU :) [17:55:54 CET] <Mavrik> Anyway, still seems like an insane setup when that could be done on some datacenter :P [17:56:17 CET] <Peetz0r> but I like insane :D [17:56:20 CET] <furq> i was going to say you could use omx/mediacodec depending on the phone [17:56:29 CET] <furq> but honestly yeah just rent a cheap vps and send your screen to that [17:56:38 CET] <furq> then run ffmpeg on there [17:57:06 CET] <Peetz0r> What method/app would you guys use to send your phone screen to somewhere? [17:57:43 CET] <Peetz0r> I think the app I am currently using is not the best choice from a performance standpoint, but most alternatives are nonfree or have watermarks or ads or timelimits [17:57:51 CET] <Peetz0r> (or all of the above) [17:58:13 CET] <Mavrik> furq, mhm, there's a whole video capture pipeline which is rather efficient, but as soon as you want to pull frames from the compositor and feed them to MediaCodec yourself, you add a lot of CPU use [17:58:21 CET] <Mavrik> And CPU will be busy chewing through pokemons :P [17:58:22 CET] <Peetz0r> maybe I should dive in and make my own app specifically for my own purpose [17:59:25 CET] <Peetz0r> well, my current setup already has 2 phones, 1 for running pokemon go and the other basically just as a glorified 4G modem [17:59:58 CET] <Peetz0r> one reason to keep those seperate is to never get any personal notifications on my stream [19:16:31 CET] <strk> how to deal with: Input link in14:v0 parameters (size 3120x4160, SAR 0:1) do not match the corresponding output link in0:v0 parameters (1280x960, SAR 1:1) [19:16:52 CET] <strk> the size is the size of one of the images passed to ffmpeg for creating the slideshow [19:18:29 CET] <strk> I'm passing it something like: -filter_complex " [0:v]scale=1280:960:force_original_aspect_ratio=decrease,pad=1280:960:(ow-iw)/2:(oh-ih)/2,setsar=1,fade=t=out:st=16:d=1[v0]; [19:18:33 CET] <strk> etc.etc. [19:18:55 CET] <strk> ah, my fault [19:26:45 CET] <strk> ok, now my fault [19:29:55 CET] <strk> what's setsar ? [19:32:02 CET] <strk> passing 3120:4160 to the scale and pad params seems to fix it [19:35:06 CET] <strk> but produces a huge video [19:37:55 CET] <iive> sar = sample aspect ratio [19:38:40 CET] <strk> so "setsar=1" means ? [19:43:34 CET] <iive> square pixels [19:45:16 CET] <strk> why does ffmpeg complains of non-matching inputs ? in an earlier run it did not complain [19:46:18 CET] <strk> and the other run also had a number of disparate sizes [19:46:19 CET] <iive> DVD 16/9 movies are using 720x576 resolution image, the resolution is 4/3, but since it has non-square pixels. [19:46:26 CET] <strk> could it be that the very first image rules the error ? [19:46:40 CET] <strk> like, the very first input in my previous run was JPEG 720x1280 720x1280+0+0 8-bit sRGB 247KB 0.020u 0:00.019 [19:46:58 CET] <strk> while this run is a JPEG 4160x3120 4160x3120+0+0 8-bit sRGB 5.404MB 0.000u 0:00.000 [19:47:20 CET] <iive> not really sure. but you can start with a single filter and add them until you get the error. [19:47:30 CET] <strk> I was using 1280x720 before [19:47:35 CET] <iive> filters are separated by "," [19:48:00 CET] <strk> my goal is to _set_ an output size and either scale larger images or pad smaller ones [19:48:41 CET] <strk> the slideshow is now completed, but it is a 90MB file, with 3120x4160 resolution [19:48:48 CET] <strk> anything smaller complained ... [19:53:34 CET] <strk> is there a quick way now to reduce the resolution of that video stream ? [20:17:43 CET] <furq> strk: pastebin the entire command somewhere [20:49:45 CET] <strk> furq: http://strk.kbt.io/tmp/slideshow.cmd [23:18:44 CET] <fuxxy> Is there a way to scan a folder of movies and show the default audio stream for each? [23:20:20 CET] <pink_mist> should be easy to script, probably [23:20:58 CET] <fuxxy> That's what I was thinking, but can't find the option to show the audio stream [23:45:52 CET] <arejay> fuxxy - could use ffprobe [23:47:34 CET] <fuxxy> I saw that, I'm playing with that now [23:48:49 CET] <fuxxy> I'm only looking to display the default audio stream's language [00:00:00 CET] --- Sun Jan 6 2019

1 0

ffmpeg-devel.log.20190104
by burek 05 Jan '19

05 Jan '19

[01:44:38 CET] <jamrial> michaelni: valgrind complains about a recent change to rangecoder by you: http://fate.ffmpeg.org/report.cgi?time=20190103234808&slot=x86_64-archlinux… [03:37:42 CET] <cone-757> ffmpeg 03James Almer 07master:ef67af31ffec: x86/af_afir: use three operand form forat some instructions [04:16:43 CET] <cone-757> ffmpeg 03Steven Liu 07master:2780cd33d938: fate: add tests/fate/hlsenc.mak for hls FATE [04:23:48 CET] <cone-757> ffmpeg 03Steven Liu 07master:eb81fd792fa8: avcodec/fft_template: improve performance of the ff_fft_init in fft_template [06:30:00 CET] <cone-757> ffmpeg 03Steven Liu 07master:b9523f2fd316: avformat/hlsenc: free variant streams after write all variant streams info [10:53:14 CET] <cone-899> ffmpeg 03Gyan Doshi 07master:d7ef10f16d56: doc/ffmpeg: update entry for stream disable options [11:20:15 CET] <cone-899> ffmpeg 03Gyan Doshi 07master:1087b792c376: doc/ffmpeg: clarify behaviour of -discard [13:05:11 CET] <Easyfab> Trying to build with msys2 I got this error for af_sofalizer https://pastebin.com/jgrr731p . Is it my build system ? [13:07:00 CET] <nevcairiel> your sofa library is probably incompatible, possibly out of date [13:20:25 CET] <Easyfab> Msys2 latest version is mingw64/mingw-w64-x86_64-libmysofa 0.6-1 [13:20:25 CET] <Easyfab> . So I need to build a more recent dev version ? [13:31:51 CET] <Easyfab> It's Ok now, thanks [14:18:15 CET] <cone-899> ffmpeg 03Jun Zhao 07master:807b57a31b4e: lavc/libdavs2: Cosmetics: Fix indentation for switch statement [14:18:16 CET] <cone-899> ffmpeg 03Jun Zhao 07master:796da4f3e083: lavc/libxavs2: Cosmetics: Fix indentation for switch statement [14:18:17 CET] <cone-899> ffmpeg 03Jun Zhao 07master:4fd6df6934d1: doc/encoders: Update docs for libxavs2 [16:01:25 CET] <cone-899> ffmpeg 03nu774 07master:e0686318ddd7: avformat/vapoursynth: properly initialize err variable in read_header_vs() [22:19:34 CET] <durandal_1707> i wrote almost perfect audio denoiser but it is extremly slow [22:20:47 CET] <atomnuker> how so? [22:20:52 CET] <jamrial> rewrite it in opencl [22:21:14 CET] <cone-297> ffmpeg 03Michael Niedermayer 07master:5faa1b8b5a04: tools/target_dec_fate: Add entries from around issue 500 to 700 [22:22:00 CET] <durandal_1707> atomnuker: ~0.06 realtime for 48000 sample rate [22:23:08 CET] <atomnuker> what's it using as a basis? ffts or iir filters? [22:23:30 CET] <durandal_1707> non-local mean [22:26:18 CET] <atomnuker> ah, yeah, that'd be slow [22:26:37 CET] <atomnuker> you can slice thread it though, can't you? [22:32:12 CET] <durandal_1707> i could, but i do only channels separately [22:32:55 CET] <durandal_1707> this thing only works rt with 8k rate, at least as described in scientific papers [22:38:13 CET] <atomnuker> can it be simd'd? [22:38:36 CET] <atomnuker> assuming floats and avx, 0.06*8 = 0.48, halfway there [22:41:18 CET] <durandal_1707> sad/ssd could be simdable, but big slowup is caused by expf * (Search window size) - 600 for 48000 rate [22:45:33 CET] <atomnuker> got code up somewhere? [22:56:35 CET] <durandal_1707> atomnuker: https://github.com/richardpl/FFmpeg/tree/anlmdn [23:01:18 CET] <durandal_1707> actually >90% time is spend on ssd [23:01:33 CET] <durandal_1707> so no point in touching expf [23:02:20 CET] <durandal_1707> also volume needs to be renormalized somehow [23:04:44 CET] <atomnuker> mov m0, [src1] ; subps m0, [src2] ; mulps m0, m0 ; addps m1, m0 ; jmp start ; <horizontal add left as an exercise for the reader> [23:05:49 CET] <atomnuker> I can see plenty of overlap during ssd computation so further on with some reordering you could probably reduce it [00:00:00 CET] --- Sat Jan 5 2019

1 0

ffmpeg.log.20190104
by burek 05 Jan '19

05 Jan '19

[00:44:34 CET] <b_jonas> hmm, so apparently I have previously encoded audio to mp3 with either (-q 4) or (-abr -b 128). I'll have to figure out what quality setting I want in the future. [00:46:39 CET] <b_jonas> oh, but I should also experiment with mobile phones and other devices in case I can encode ogg or webm in some way that they are willing to play itt [00:46:55 CET] <b_jonas> so far I only got these stupid mobile phones to play mp3 or lossless audio [00:47:22 CET] <JEEB> vorbis in ogg, aac in mp4 are two things that work on androids etc, idevices want aac in mp4 [00:48:57 CET] <b_jonas> JEEB: these phones are not smartphones and not android [00:49:28 CET] <b_jonas> the hard part is, of course, that I would ideally like them play on future devices that I haven't even bought yet [00:49:34 CET] <JEEB> well generally then raw mp3 or aac in mp4 are the likely things to work [00:49:39 CET] <b_jonas> but I can make different encodes later [00:49:49 CET] <JEEB> yes, just keep oroginsl data [00:50:15 CET] <JEEB> since audio cds are just 1411kbps or so [00:50:24 CET] <b_jonas> JEEB: the mobile phones can record video, so I can test what video they record, and hope they can play audio files with the same audio codec as in the video they recorded, if I just set parameters correctly [00:50:25 CET] <JEEB> small compared to raw video data [00:50:39 CET] <b_jonas> it gets ugly, eg. some cheap mobile phones won't even render progressive (multi-plane) jpegs [00:50:42 CET] <b_jonas> only baseline jpegs [00:50:43 CET] <JEEB> b_jonas: overthinking I think :) [00:50:57 CET] <JEEB> try cbr mp3 and aac in mp4 [00:51:08 CET] <JEEB> put both onto device abd see if works [00:51:15 CET] <JEEB> quicker to test (tm) [00:51:18 CET] <b_jonas> JEEB: sure, or flac-compressed data because that's lossless and half or quarter of the size of the CD data [00:51:23 CET] <JEEB> yup [00:51:26 CET] <b_jonas> I have already tested mp3, that works [00:51:42 CET] <b_jonas> even mp3 files that I have encoded before these phones even existed work [00:51:59 CET] <furq> what cds are you buying that compress by 75% [00:52:07 CET] <b_jonas> ok, I will test aac in mp4 too then [00:52:19 CET] <b_jonas> furq: hmm, let me check the actual compression rate [00:52:21 CET] <JEEB> -b:a aac -b:a 192k -movflags faststart and then output file as out.mp4 [00:52:33 CET] <JEEB> argh first is -c:a aac [00:52:37 CET] <JEEB> of course :p [00:52:41 CET] <furq> flac averages about 65% compression ratio [00:52:53 CET] <furq> i.e. the output is 65% of the input [00:53:25 CET] <b_jonas> darn, I deleted the cd data ... I don't want to re-read the cd, so I'll have to guess the length from the duration [00:53:42 CET] <furq> i've got discs that are less than 50% but not many [00:58:34 CET] <b_jonas> yes, that's probably a better approximation [01:00:02 CET] <b_jonas> this specific album has total size of 296568832 bytes of flac, and a total duration of 2646.2 seconds [01:00:22 CET] <b_jonas> what's the bitrate for CD audio? [01:00:46 CET] <b_jonas> JEEB said 1411kbps [01:01:00 CET] <JEEB> yup, something like that [01:01:41 CET] <b_jonas> that can't work, that wouldn't fit these 45 minutes on a CD [01:02:03 CET] <b_jonas> 2646.2 seconds would be DVD sized at that bitrate [01:05:05 CET] <b_jonas> ah, it's 1411 kiloBITS per second [01:05:06 CET] <b_jonas> sorry [01:06:56 CET] <b_jonas> so the grabbed CD data is about 466789690 bytes, so flac compression rate is 63% [01:06:58 CET] <b_jonas> you're right [04:30:33 CET] <haasn> is it possible to "reinterpret" the video format without conversion? say I have a rgbp video that I want to "cast" to yuv444p [04:30:51 CET] <haasn> i.e. I have an RGB video where red contains the luma information and so on [04:31:14 CET] <haasn> I can try using something like swscale to convert from rgbp to yuv444p but this applies colorspace conversions [04:35:12 CET] <klaxa> haasn: you can use -pix_fmt as an input option [04:36:03 CET] <haasn> Option pixel_format not found. [04:36:30 CET] <haasn> ffmpeg -pix_fmt yuv444p -i in.mov -c copy out.mkv [04:36:44 CET] <klaxa> hmm... just checked with my old code, i only used it on rawvideo [04:37:14 CET] <haasn> I would mostly be interested in an avfilter though, so I can use it with -vf in mpv [04:37:32 CET] <haasn> extractplanes + combineplanes I guess [04:37:43 CET] <haasn> except I have no idea how to "combine planes" [04:38:29 CET] <haasn> ah it's called 'mergeplanes' [04:41:12 CET] <haasn> -lavfi-complex '[vid1] extractplanes=r+g+b [v][y][u]; [y][u][v] mergeplanes=0x001020:yuv444p [vo]' [04:41:14 CET] <haasn> this works [04:41:41 CET] <haasn> sort of annoying having to do it this way but at least that works [06:01:44 CET] <hungrywolf> How do I figure out what kind of library omxplayer uses to display video to the screen, and ofcourse its audio library [07:02:31 CET] <analogical> when I type "ffmpeg.exe -i video.mkv -c:v copy -c:a copy video.mp4" I only want to copy the video and the audio from the source file. What do I type to NOT copy the subtitles from the source file? [09:10:05 CET] <TheAMM> analogical: -sn [09:10:16 CET] <TheAMM> ffmpeg -i video.mkv -c copy -sn video.mp4 [09:10:49 CET] <TheAMM> -vn and -an for video/audio, there was a fourth one but I forget what it was about (metadata?) [09:30:45 CET] <CoreX> :P [09:30:56 CET] <CoreX> [20:14:48] <analogical> can FFmpeg be used to remove metadata and tags from videofiles? [09:30:56 CET] <CoreX> [20:15:18] <furq> analogical: -map_metadata -1 [09:30:56 CET] <CoreX> [20:15:24] <furq> it only removes container metadata though [09:30:56 CET] <CoreX> [20:20:42] <analogical> furq, it worked! Now how do I apply that to all the files in the same directory? [09:30:58 CET] <CoreX> [06:02:31] <analogical> when I type "ffmpeg.exe -i video.mkv -c:v copy -c:a copy video.mp4" I only want to copy the video and the audio from the source file. What do I type to NOT copy the subtitles from the source file? [09:31:14 CET] <CoreX> TheAMM those are the only ones he asked about from what i can see [09:33:11 CET] <TheAMM> k [09:34:34 CET] <analogical> I've tried lots of things but I still cant remove the subtitles from the videofile [09:43:30 CET] <analogical> seems like I have to remux the file just to remove the subtitles [09:43:44 CET] <analogical> with mkvtoolnix [09:53:27 CET] <hungrywolf> This is a very small video player >> https://github.com/illuusio/ffmpeg-tutorial/blob/master/tutorial07.c [09:53:37 CET] <hungrywolf> This is built using sdl and ffmpeg [09:53:47 CET] <hungrywolf> but this does not work on raspberrypi 0 [09:54:01 CET] <hungrywolf> The video is too laggy [09:54:10 CET] <hungrywolf> is there any way I can make it work on rpi0? [16:00:24 CET] <lays147> Good afternoon guys... I am trying to convert a mxf with 8 tracks of audio to a wav file with only the audios, but apparently my code just get's the first track of the audio from the mxf to the wav. is there a way to merge them all? because some videos has audio on the second track and only the first is being used on the wav conversion [16:16:27 CET] <DHE> you'll have to use a filter to merge multiple audio tracks into a single audio track [16:16:54 CET] <lays147> DHE: any tips on how to do that? [16:17:33 CET] <DHE> lays147: https://ffmpeg.org/ffmpeg-filters.html#amix I assume you want a simple mono or stereo track that sounds like playing all audio tracks simultaneously... I dunno about mxf specifically but this might get you started [16:20:33 CET] <lays147> DHE: you got the idea, however, the only input is one video with 8 tracks, and this amix sounds like it needs a few files for input? [16:21:16 CET] <DHE> that's the example they provide. you can do it with one file but you'll have to build the filter_complex manually [16:57:45 CET] <SQUelcher> Hello. [17:09:30 CET] <SQUelcher> I've noticed that libfdk_aac got an upgrade a few days ago (0.1.6 -> 2.0.0). [ https://sourceforge.net/projects/opencore-amr/files/fdk-aac/ ] - After upgrading my library, ffmpeg doesn't seem to work with it anymore. It seems that it is trying to encode something, but everything stays in the buffer, until I quit ffmpeg. (like: [libfdk_aac @ 0x55a786462900] 613 frames left in the queue on closing). After [17:09:36 CET] <SQUelcher> downgrading, the problem is gone. Current git also does not work with the new library. [17:09:39 CET] <SQUelcher> Should I open a bug report about this? [17:12:32 CET] <c_14> If there's not already an open bug report about that, yeah [17:14:35 CET] <SQUelcher> I haven't seen one, so I will. Ok, thanks. [19:17:00 CET] <Mavrik> um... what happened to the open broadcast encoder project or what was it? The live streaming server someone was developing? [21:13:15 CET] <Peetz0r> Hey! I'm trying to mix some different inputs (x11grap, mjpeg stream, v4l2 device, and alsa audio) into one stream and stream that to twitch. All this on a Raspberry Pi 3A+ using h264_omx. This is now 50% working. [21:14:42 CET] <Peetz0r> But I'm having some issues with performance and audio. I think that one of the problems might be that my mjpeg source has a very irregular framerate. Is there a way to buffer frames so that it can just do it's work at 30fps (or whatever rate I specify) amd duplicating or dropping frames whenever needed? [21:17:08 CET] <Peetz0r> Here's my current commandline: https://paste.sigio.nl/view/45e772d7 [21:18:43 CET] <Peetz0r> (my endgoal is a portable Twitch mobile gaming setup based on 2 phones, 1 webcam, a Pi, a huge powerbank and an unlimited 4G plan) [21:19:38 CET] <Peetz0r> Also, what would be the best CPU/GPU memory split on a 512MB Pi 3A+? [22:17:52 CET] <Skandalist> Hello. I'm trying to capture screen with good quality with ffmpeg, but the video comes slowly out of sync with audio (I think, because of slow HDD), I was trying different formats, some are better, the others are worse. Can VFR help with that? Should I use -vsync 2 -r 25 to enable it? [00:00:00 CET] --- Sat Jan 5 2019

1 0

ffmpeg-devel.log.20190103
by burek 04 Jan '19

04 Jan '19

[10:45:36 CET] <thardin> so did everyone hear about the widevine drm break? [10:45:51 CET] <durandal_1707> nope [10:47:14 CET] <thardin> https://mobile.twitter.com/david3141593/status/1080606827384131590 [11:07:33 CET] <nevcairiel> the stuff was somehow vulnerable anyway, you see bit-by-bit copies unencrypted floating around the web since forever [11:07:57 CET] <nevcairiel> (ie. the so called WEB-DL thingys) [11:16:00 CET] <thardin> you mean youtube-dl and friends? [11:16:56 CET] <thardin> oh wait, as in the warez term. yeah [11:22:24 CET] <durandal_1707> that have nothing to do with warez [11:25:23 CET] <thardin> hm, wikipedia thinks so [11:45:02 CET] <cone-284> ffmpeg 03Gyan Doshi 07master:3ba5eef2c7b2: ffmpeg: allow disabling streams by type for inputs [14:13:18 CET] <cone-284> ffmpeg 03James Almer 07master:9b5bd665e105: x86/af_afir: fix processing the last element [14:13:19 CET] <cone-284> ffmpeg 03James Almer 07master:82043dfd2e50: avfilter/af_afir: split off fcmul_add into a DSP context [14:13:20 CET] <cone-284> ffmpeg 03James Almer 07master:ba89dc27b50c: checkasm: add an af_afir test [14:13:21 CET] <cone-284> ffmpeg 03James Almer 07master:5402c1886b97: x86/af_afir: add ff_fcmul_add_avx() [17:32:11 CET] <cone-618> ffmpeg 03Oliver Collyer 07master:bf96937a961b: avdevice/dshow: Fixed some minor memory leaks [22:59:44 CET] <cone-757> ffmpeg 03Manoj Gupta 07master:6fcf813110fe: libavcodec: Remove dynamic relocs from aarch64/h264idct_neon.S [00:00:00 CET] --- Fri Jan 4 2019

1 0

ffmpeg.log.20190103
by burek 04 Jan '19

04 Jan '19

[05:21:44 CET] <johnjay> is there a way to capture the audio of the system but on android with ffmpeg in termux? [07:32:08 CET] <hungrywolf> can some one tell me how to get AVCodecContext from AVFormatContext [07:39:41 CET] <hungrywolf> or even just from a stream [08:42:30 CET] <kepstin> hungrywolf: decoding? just look up the decoder by the stream codec id: https://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/demuxing_decod… [08:46:22 CET] <kepstin> for encoding, well, containers usually support multiple codecs. ffmpeg has a "default" audio and video codec for each container, but it's not always the best choice. You'll probably want to look up a specific encoder by name (e.g. "libx264") when encoding. [09:18:21 CET] <meiamsome> Hello. I am currently trying to use the command line to stream a .mov file into a fake v4l2 camera and a pulseaudio output. What I have so far is that this: `ffmpeg -re -i big_buck_bunny_480p_h264.mov -map 0:0 -flush_packets 1 -vf scale=640x480 -f v4l2 /dev/video0` (output here: https://pastebin.com/qutCvYmf ) will stream to the webcam correctly, and setting up pulse with this: `ffmpeg -re -i big_buck_bunny_480p_h264.mov -map 0:0 -flush_packets 1 [09:18:21 CET] <meiamsome> -vf scale=640x480 -f v4l2 /dev/video0 -map 0:2 -f pulse a` (output here: https://pastebin.com/YjvMVGAf ) works correctly for the audio output, but it causes the webcam video stream to start lagging (It looks like a couple of frames all get pushed out at once and then there is a time gap). Does anyone have any suggestions? [11:58:34 CET] <meiamsome> I have a HLS file here that appears to have some encoding issues, but I can't work out what the root problem is - does anyone know of a good way to debug these? https://live-test.ugcpro.tv/5c2dd0704e90820203149ef1/master.m3u8 [13:03:56 CET] <hungrywolf> kepstin: thank you [21:14:48 CET] <analogical> can FFmpeg be used to remove metadata and tags from videofiles? [21:15:18 CET] <furq> analogical: -map_metadata -1 [21:15:24 CET] <furq> it only removes container metadata though [21:20:42 CET] <analogical> furq, it worked! Now how do I apply that to all the files in the same directory? [22:01:51 CET] <seni_> how big of a performance difference does ASM make? [22:03:14 CET] <Mavrik> In ffmpeg, a lot. [22:03:42 CET] <Mavrik> As in - don't run encoders/decoders/filters for video without it.\ [22:04:19 CET] <seni_> what about audio? [22:05:02 CET] <durandal_1707> same [22:10:34 CET] <seni_> what's a lot? is it 2x slower? because I can live with that? or is it 1000x slower? [22:11:21 CET] <Mavrik> Well, measure. [22:11:27 CET] <seni_> also, is there a recommended stable version? getting errors when compiling for android, this is after disabling ASM [22:12:01 CET] <Mavrik> Since the speed difference is different depending on what codec you're using (and if codec even has ASM code). [22:12:27 CET] <Mavrik> CPU encoding on Android will be fun :D [23:54:40 CET] <MichaelJoel> I see how to use the -to but can not find how to force it to begin at a certain time within a video??? [23:56:31 CET] <c_14> -ss [23:56:57 CET] <MichaelJoel> thanks! [00:00:00 CET] --- Fri Jan 4 2019

1 0

ffmpeg-devel.log.20190102
by burek 03 Jan '19

03 Jan '19

[00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:07607a1db879: avcodec/4xm: Fix returned error codes [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:fe17f9b95683: swscale/yuv2rgb: Return a more specific error code from ff_yuv2rgb_c_init_tables() [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:5d9a69167cb1: avcodec/aacps: Return more specific error codes. [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:057328b29d74: avcodec/bitstream: Return specific error codes when building vlc tables [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:9520d51e21f9: avcodec/avpacket: Avoid unspecific return -1 for av_grow_packet() [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:6747a6339759: avcodec/asvenc: Replace bitstream space check by assert [00:24:41 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:34f942c6c41f: avcodec/assenc: Return more specific error codes for ass_encode_frame() [00:24:42 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:c6db2e97dcd2: avcodec/ac3enc: Use specific error code in validate_float_option() [00:24:43 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:8e1b5ba53839: libavcodec/ac3dec.c: Avoid unspecific -1 error code [00:24:44 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:5079eca22a16: avcodec/alac: Avoid unspecific error codes and forward error codes [00:24:45 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:2d9fe6321e32: avcodec/adpcm: use assert for codec id instead or error return [00:24:45 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:bb2d8708f7bc: avcodec/binkaudio: Use assert instead of return for use of compile time disabled codecs [00:24:47 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:0505b35a38c5: avcodec/dct: Propagate error code from ff_rdft_init() [00:24:48 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:8b1099c288ab: avcodec/dcaenc: Return specific error code from subband_bufer_alloc() [00:24:49 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:eb49d259f420: avcodec/eatqi: Return more specific error code from tqi_decode_mb() [00:24:50 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:3ee323583993: avcodec/eacmv: Return more specific error codes and forward error codes [00:24:51 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:b33de5574704: doc/indevs: fix upto typo [00:24:52 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:47c3a10b16f2: avfilter/vf_tonemap_opencl: Make static tables const [00:24:53 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:796807022c93: avfilter/ebur128: Simplify by using log10() [00:24:54 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:5785e0e9c401: avfilter/af_crossfeed: Use ff_exp10() [00:24:55 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:1b6695354d9e: avfilter/af_biquads: minor simplification by using ff_exp10() [00:24:56 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:044be48faead: avcodec/acelp_pitch_delay: Minor simplification by using ff_exp10() [00:24:56 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:33d1fb06279c: avcodec/aacdec_fixed: Provide context to av_log() [00:24:58 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:052ad4d25540: avcodec/imc: Provide non NULL context to av_log() [00:24:59 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:9446c5887f99: avcodec/mss2: Provide non NULL context to av_log() [00:25:00 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:b89fc3d01b8e: avcodec/bink: Provide non NULL context to av_log() [00:25:00 CET] <cone-364> ffmpeg 03Michael Niedermayer 07master:1dcb5b7dca2b: avcodec/bfi: Provide non NULL context to av_log() [15:15:58 CET] <ubitux> why is ffmpeg.c mux_timebase set to the encoder timebase instead of the output AVStream timebase set by the muxer itself? [15:16:07 CET] <ubitux> i'm confused. [15:17:49 CET] <Compn> in case codec and container timebase differ ? [15:18:18 CET] <Compn> maybe hardware encoders ? [15:32:16 CET] <ubitux> the pkt timestamps are rebased from the encoder timebase to the muxer timebase before being sent to the muxer [15:32:24 CET] <ubitux> but in ffmpeg it seems kind of different [15:32:40 CET] <ubitux> it's like, it's converted from encoder time base, to a muxer timebase and then to the ost tb [15:32:59 CET] <ubitux> there is this intermediate mux timebase which isn't actually about the muxer [15:33:16 CET] <ubitux> since the muxer expect it in ost timebase [15:33:25 CET] <ubitux> (defined by itself) [15:36:12 CET] <cone-314> ffmpeg 03Paul B Mahol 07master:af5b6b45b4d8: avfilter/vf_detelecine: fix obvious frame memory leaks [16:15:46 CET] <cone-314> ffmpeg 03Paul B Mahol 07master:25c7aa9974af: aformat/bfi: use ff_get_extradata() [16:35:22 CET] <cone-314> ffmpeg 03Paul B Mahol 07master:1139887ace7f: avformat/acm: use ff_get_extradata() [16:35:23 CET] <cone-314> ffmpeg 03Paul B Mahol 07master:e8d36ba47753: avformat/yop: use ff_get_extradata() [16:35:24 CET] <cone-314> ffmpeg 03Paul B Mahol 07master:37af319086a6: avformat/anm: use ff_get_extradata() [16:51:48 CET] <cone-314> ffmpeg 03Derek Buitenhuis 07master:8f585c3c550c: mov: Remove duration-of-last-frame heuristic hack [17:05:20 CET] <atomnuker> kierank: https://www.scotsman.com/news/uk-government-fighting-eu-plans-to-end-daylig… [17:05:50 CET] <thardin> haha what [17:05:58 CET] <nevcairiel> what does the UK care, dont you want to be gone by then [17:05:59 CET] <atomnuker> are they just doing it because they can and it won't matter soon? [17:06:26 CET] <thardin> they're out by what, march this year? [17:06:32 CET] <thardin> april maybe [17:08:48 CET] <atomnuker> nevcairiel: wish I knew where to go, it seems to me its all bad everywhere [20:08:21 CET] <atomnuker> it'd be nice if taxes weren't inversely proportional to internet speeds, >= 40% is insane [20:28:33 CET] <durandal_1707> using avx should be faster that pure sse3 by what amount? [20:28:47 CET] <nevcairiel> anything from 0 to 100% [20:29:08 CET] <durandal_1707> why? [20:29:41 CET] <nevcairiel> if its entirely compute bound and stuff scales perfectly with bigger registers, you can get 100% increase (well, almost, anyway) [20:29:54 CET] <nevcairiel> if other factors are limiting it, then it'll be lower then that [20:29:57 CET] <nevcairiel> there is no one answer [20:33:56 CET] <durandal_1707> what other factors? [20:37:44 CET] <kurosu> amd [20:38:50 CET] <nevcairiel> memory access may be a concern, and there is always the lane business to consider depending on how the data is structured and has to be moved [20:40:25 CET] <nevcairiel> pretty short algorithms may also cost a bit more due to avx clock changes, although that factor is disappearing more and more with newer gen cpus [20:40:58 CET] <nevcairiel> and of course AMD only has 128-bit ALUs, so they dont benefit from avx much, if at all. presumably thats changing with zen2 though, so yay for that, i guess [22:55:50 CET] <cone-256> ffmpeg 03Gyan Doshi 07master:54109b1d1472: ffmpeg: skip disabled streams [00:00:00 CET] --- Thu Jan 3 2019

1 0

ffmpeg.log.20190102
by burek 03 Jan '19

03 Jan '19

[00:27:01 CET] <Skandalist> Hello. Is it possible to make ffmpeg always to choose only default video and audio streams while transcoding? [00:30:07 CET] <klaxa> -map 0:v -map 0:a will use the first video and audio track from the first input file [00:32:30 CET] <Skandalist> I've found information, that this would copy all video and all audio tracks [00:33:05 CET] <klaxa> hmm -map 0:v:0 and -map 0:a:0 ? [00:34:09 CET] <Skandalist> yes this would take the first ones, but what if the first one is not the default? [00:35:30 CET] <klaxa> i think in that case you're best off reading/parsing ffmpeg's or ffprobe's output with a script and build your command line from the results of that [00:47:42 CET] <Skandalist> Ok. Thank You. [05:43:41 CET] <BenderRodriguez> How do I specify a sink to use with ffplay [05:43:53 CET] <BenderRodriguez> it doesn't seem to work and segfaults if I just let it all default [10:25:22 CET] <skident> Hi all. [10:26:16 CET] <skident> Guys, how to build FFMPEG library on macOS (not for the latest version of macOS) [10:26:36 CET] <skident> ./configure [10:27:00 CET] <skident> make CXXFLAGS="-mmacosx-version-min=10.10" [10:27:23 CET] <Tatsh> is it possible to convert VideoToolbox pixel format to YUYV422? [10:28:15 CET] <Tatsh> i am working with swscale and swscale_isSupported(videotoolbox_fmt) returns 0; is it just not possible? [10:28:20 CET] <skident> it builds all needed libraries but when I link it to my project I got the warning that my object files inside library were built for macos 10.14 (I specified 10.10) [10:28:41 CET] <Tatsh> skident, that's because -mmin-macosx- is set for some files but not your other files [10:29:02 CET] <Tatsh> if you're in cmake you should set CMAKE_MACOSX_DEPLOYMENT_TARGET before your project() call [10:29:14 CET] <Tatsh> otherwise do the equivalent for your system [10:29:39 CET] <analogical> how do I use FFmpeg to remove metadata from all the videofiles in a folder? (windows) [10:29:42 CET] <skident> I'm not on CMAKE [10:30:09 CET] <skident> Tatsh: I'm using the default autotools files for build ffmpeg [10:30:46 CET] <skident> ./configure && make CXXFLAGS="-mmacosx-version-min=10.10" [10:31:00 CET] <Tatsh> what about CFLAGS? [10:31:10 CET] <Tatsh> CXXFLAGS only applies to C++ [10:32:23 CET] <skident> I tried CFLAGS and got error that some header (SDL.h) not found [10:34:00 CET] <skident> also, I tries both CFLAGS and CXXFLAGS and it didn't help too [10:36:14 CET] <skident> maybe I need to add some option to 'configure'? [10:40:55 CET] <skident> Seems like the default compiler is gcc and when I set CFLAGS - compiler is changed to clang and it doesn't know where to find some headers [10:42:54 CET] <analogical> how do I use FFmpeg to remove metadata from all the videofiles in a folder? (windows) [11:09:11 CET] <skident> Tatsh: thank you. I added a path to an SDL2 headers and it helped [11:18:07 CET] <AnpMoot> hi all. I'm making a timelapse from lots of jpg files that contain no exif data. running this command creates a movie with the correct dimensions (352x288), but the aspect ratio is wrong (in VLC). is this because the lack of exif, and what can be done? [11:18:17 CET] <AnpMoot> ffmpeg -f concat -r 25 -i filelist.txt -vcodec libx264 out.mp4 [11:23:43 CET] <Tatsh> -aspect [11:28:38 CET] <AnpMoot> thanks. -aspect 352:288 made it great [12:47:07 CET] <analogical> how do I use FFmpeg to remove all metadata from all the videofiles in a folder? (windows) [13:44:17 CET] <tsglove> analogical, what have you tried? [13:51:31 CET] <analogical> tsglove, I've tried mkvtoolnix but that only works for one file at a time and need to remove tags and metadata from hundreds of videofiles [14:05:06 CET] <c_14> use a bash script? [14:06:39 CET] <tsglove> analogical, if I were you: nail down converting ONE file. Remove the tags and metadata from that one. Nail down your one-liner and the ffmpeg incantations. [14:07:07 CET] <tsglove> Then, as c_14 states, create a simple bash script. For a for each file in files_list: blah_blah -graph-thingy-here [14:07:16 CET] <tsglove> And make sure to backup your files. [14:07:21 CET] <tsglove> Pre-tests. [14:11:36 CET] <analogical> tsglove, I have no idea how to make bash script [14:11:53 CET] <tsglove> Maybe you can learn? [14:12:07 CET] <tsglove> Dude, get your one-liner nailed down. Make sure you can convert ONE file as you desire. [14:12:17 CET] <analogical> no time for that [14:12:20 CET] <tsglove> Once that is done... time to "wrap it" in a for [14:12:21 CET] <tsglove> oh [14:12:24 CET] <tsglove> then no time to help you [14:13:09 CET] <analogical> so you want me to become a programmer just to remove some tags? [14:13:31 CET] <tsglove> hi-yo!! Good luck buddy. I hope 2019 brings awesome things to you and your loved ones. [14:14:19 CET] <analogical> I already know how to edit ONE file as I mention earlier but I need to edit hundres of files at once [14:27:55 CET] <iive> analogical, you automate it, by making the computer remove them one file at a time. [16:03:10 CET] <poopBot> hi i have a video of 9mb [16:03:12 CET] <poopBot> ffmpeg -i z.mp4 -vcodec copy -acodec copy -ss 0 -fs 8000000 zz.mp4 [16:03:30 CET] <poopBot> and this gives me video of 4.5 mb [16:04:35 CET] <poopBot> duno wtf is happening, my logic was its probbaly decresing size somehow else so it dosent even reach -fs 8000000 [16:04:35 CET] <durandal_1707> poopBot: full uncut console output missing [16:05:00 CET] <poopBot> duno how to full console output -vv ? [16:05:15 CET] <pink_mist> poopBot: is this a twitch vod or something? [16:05:52 CET] <poopBot> nah its video i recoreded of my desktop using simpleScrenRecorder [16:06:08 CET] <pink_mist> oh [16:06:26 CET] <poopBot> now when i want to share on discord it crys 8mb limit so i want to write a script to split video in some size chunks [16:07:19 CET] <poopBot> yep just tested whitout -fs it still decreses size from 9mb to 4.5 [16:08:43 CET] <poopBot> https://pastebin.com/VKaVii2H [16:08:58 CET] <poopBot> meybe this "[mp4 @ 0x55e437b184c0] track 1: codec frame size is not set [16:08:58 CET] <poopBot> " [16:09:14 CET] <poopBot> is there copy frame size ? [16:10:06 CET] <pink_mist> is the video wrong? [16:10:18 CET] <poopBot> dont think so i think its same 100% [16:10:37 CET] <pink_mist> so probably ssr is just shit at writing files :P [16:12:28 CET] <poopBot> will ffprobe both files and compere if there is any diference [16:14:55 CET] <DHE> So, crazy idea. I have a long-running (over the air broadcast receiver) decoder that sometimes gets into a bad state and needs a restart to fix glitching issues. How crazy is it that when I get an AVPacket marked as a keyframe I quickly close and re-open the decoder before proceeding? [16:15:21 CET] <poopBot> hmm for 1 it shows [16:15:22 CET] <poopBot> Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 1920x1080 [SAR 1:1 DAR 16:9], 527 kb/s, 30.02 fps, 30 tbr, 15360 tbn, 60 tbc (default) [16:15:37 CET] <poopBot> Stream #0:0(und): Video: h264 (avc1 / 0x31637661), none, 1920x1080, 496 kb/s, SAR 1:1 DAR 16:9, 30 fps, 30 tbr, 15360 tbn, 30720 tbc (default) [16:16:03 CET] <poopBot> am i converting it same or what? [16:18:30 CET] <poopBot> also on output file i get [mov,mp4,m4a,3gp,3g2,mj2 @ 0x55ddcc2ea680] Could not find codec parameters for stream 0 (Video: h264 (avc1 / 0x31637661), none, 1920x1080, 496 kb/s): unspecified pixel format [16:26:33 CET] <pzy> semi-annual reminder that ffmpeg [16:26:36 CET] <pzy> IS COOL [16:26:44 CET] <poopBot> :) [16:27:03 CET] <pzy> in 2018 I learned about the -listen flag and it got cooler [16:42:01 CET] <DWa1ker> any way one can fix the screen orientation for videos with static images e.g. https://youtu.be/0x-w2o1I4gI, youtube keep making it horizontal no matter which size I make the picture, can't figure it out but possibly this is something to do with picture ratio [16:42:42 CET] <pzy> https://stackoverflow.com/questions/3937387/rotating-videos-with-ffmpeg [16:43:14 CET] <pzy> ffmpeg -i in.mov -vf "transpose=1" out.mov [16:43:14 CET] <pzy> For the transpose parameter you can pass: 0 = 90CounterCLockwise and Vertical Flip (default); 1 = 90Clockwise; 2 = 90CounterClockwise; 3 = 90Clockwise and Vertical Flip [16:44:04 CET] <DWa1ker> thanks, will read the tread and report back, on the road atm. [16:48:02 CET] <DWa1ker> this seems to be about downloaded videos but what I want is it to make youtube not to rotate it, might be wrong, will read more careful later [16:48:57 CET] <poopBot> anyone know whats wrong with my issue, why ffmpeg -i y.mp4 -vcodec copy -acodec copy -ss 0 zz.mp4 makes same video just smaller size [16:49:41 CET] <durandal_1707> poopBot: have you sent full uncut console output using pastebin? [16:51:16 CET] <durandal_1707> also application my insert junk chunks in container [16:52:11 CET] <poopBot> https://pastebin.com/7AvmjViH [16:53:46 CET] <durandal_1707> poopBot: is audio working in output file? [16:54:43 CET] <poopBot> puls audio shouws its playing some [16:55:00 CET] <poopBot> tought i just recorded desktop there was no sound atm [16:55:12 CET] <poopBot> i can record new one with youtube on to test [16:56:51 CET] <durandal_1707> you could inspect atoms in mp4 files, is size difference always same or it increases with content duration? [17:15:53 CET] <poopBot> durandal_1707, nah just tested have audio on both files [17:16:12 CET] <poopBot> how to inspect atoms [17:18:22 CET] <pink_mist> I'm still unclear what the problem is - you wanted files small enough that you could post them on discord, you got files that were small enough to post on discord [17:21:21 CET] <durandal_1707> poopBot: you havent answered what happens if you record much longer video, does copy then crate smaller files for same amount or it scales up? [17:21:40 CET] <poopBot> i found what was issue [17:22:45 CET] <poopBot> or at least i fugure out what makes it smaller [17:23:11 CET] <poopBot> if i have -ss 0 it makes it smaller, tought lol dont get why [17:23:30 CET] <poopBot> dosent -ss mean to seek to 0 is first frame? [17:23:53 CET] <poopBot> running same command whiotut -ss 0 leaves same file size [17:25:43 CET] <durandal_1707> negative start? [17:25:52 CET] <poopBot> would love to know why, just for why not [17:25:59 CET] <poopBot> what you mean by negative start? [17:26:32 CET] <durandal_1707> timestamps do not start from 0 [17:26:35 CET] <poopBot> anyway i think i can write it down like this in my script it will spit it anyway with fs [17:27:19 CET] <poopBot> what, from what i see both video are same , dident notice any cutting, from when do they start if not form 0 [17:27:36 CET] <poopBot> or you talking about some magic behind that i dont know? [17:28:55 CET] <poopBot> all have Duration: 00:00:19.94, start: 0.000000 [17:29:31 CET] <poopBot> vanila one, converted with ss 0 and converted whiout [18:26:26 CET] <the_gamer> on ffprobe Duration: 05:10:00.73 means a duration of over 5 hours? [18:32:16 CET] <c_14> yes [19:28:42 CET] <poopBot> i writed a scipt it it steels me shomehow 6 seconds from evry split i think [19:38:55 CET] <poopBot> hmm duno why it stills me seocnds [19:39:12 CET] <poopBot> does using -fs brakes something or what [19:44:25 CET] <poopBot> ok so when i open with vlc it shows that 6 sec blank then it starts playing video [19:45:05 CET] <poopBot> so guess -fs brakes video somehow, anyone know how to fix this? [19:51:29 CET] <pink_mist> poopBot: splitting can only happen on keyframes if you're using a copy vcodec, so if the -fs limit happens somewhere other than on a keyframe, you can get such issues if you're using copy [19:51:40 CET] <pink_mist> afaik [19:51:47 CET] <poopBot> any solution? [19:52:08 CET] <poopBot> i saw some post -g 1 tought duno what it does [19:52:45 CET] <pink_mist> I'm not sure ... either re-encode which may cause quality degradation, or somehow find out which keyframe you need to split at, and explicitly split on that one instead [19:53:49 CET] <pink_mist> I have no clue how to do the latter though [20:07:36 CET] <poopBot> stupit think wasted whole day on it so i can end up 8 sec short [20:40:42 CET] <poopBot> i just reoncoded it counded find a better fix [20:58:00 CET] <pa> hi [20:58:02 CET] <pa> question [20:58:12 CET] <pa> what's a suitable container to stream audio, video and subtitles? [20:58:17 CET] <pa> ffmpeg tells me flv is not [20:58:30 CET] <pa> mkv as well [20:58:34 CET] <pa> and mp4 neither [20:58:45 CET] <pa> what should i use? [20:58:58 CET] <Mavrik> why aren't shose? [20:58:59 CET] <pa> (for h264 video, aac audio and subs) [20:59:08 CET] <furq> pa: streaming to what [20:59:12 CET] <pa> Mavrik: no idea [20:59:17 CET] <pa> furq: streaming to pipe [20:59:26 CET] <furq> probably just mkv then [20:59:28 CET] <pa> | nc -l -p xxxx [20:59:33 CET] <pa> i tried [20:59:34 CET] <furq> or more likely mpegts [20:59:47 CET] <pa> NULL @ 0x34377c0] Requested output format 'mkv' is not a suitable output format [21:00:44 CET] <pa> furq: mpegts seems to work [21:00:50 CET] <pa> but on the decode side, the timing is wrong [21:00:56 CET] <pa> it decodes as fast as possible [21:01:16 CET] <furq> if you're trying to simulate a live input then use -re before -i [21:01:17 CET] <pa> probably fps aren't passed through [21:02:12 CET] <pa> furq: hm.. no that doesnt work [21:02:17 CET] <pa> cache gets flushed first [21:02:27 CET] <pa> probably i have to force fps on the decoding side [21:03:24 CET] <pa> that works [21:07:14 CET] <BtbN> it's called matroska if you use -f [00:00:00 CET] --- Thu Jan 3 2019

1 0

ffmpeg-devel.log.20190101
by burek 02 Jan '19

02 Jan '19

[00:05:36 CET] <thardin> happy GNU+year! [11:59:11 CET] <cone-113> ffmpeg 03Paul B Mahol 07master:f7745edeaaee: doc/filters: improve afir description a little [12:30:08 CET] <cone-113> ffmpeg 03Paul B Mahol 07master:e1f7881d926f: afilter/af_sofalizer: check explicitly other type [13:35:15 CET] <cone-113> ffmpeg 03Paul B Mahol 07master:35a8179149c1: avfilter/af_adelay: allow setting delays in seconds [18:12:06 CET] <j-b> HappyNewYear! [19:13:03 CET] <durandal_1707> i always forgot what to call when counting number of instructions [19:15:06 CET] <durandal_1707> so what code count cycles? [19:16:50 CET] <atomnuker> libavutil/timer.h [19:17:12 CET] <atomnuker> START_TIMER <code> STOP_TIMER("name"); [19:27:36 CET] <cone-597> ffmpeg 03James Almer 07master:3209d7b3930b: configure: bump year [19:33:13 CET] <kurosu> and you can't intermingle several [19:52:19 CET] <durandal_1707> i get slightly more decycles but overall speed is about same, and perf reports smaller usage of time in fcmul_add [19:53:00 CET] <durandal_1707> that is probably avx overhead [19:53:22 CET] <kierank> durandal_1707: turn cpu govenor to perofmrance [19:53:30 CET] <kierank> also set affinity to one cpu [19:53:47 CET] <durandal_1707> how? [19:54:52 CET] <kierank> google [20:08:30 CET] <durandal_1707> yes, avx is slower somehow on this cpu [20:10:42 CET] <atomnuker> compared to sse? [20:12:02 CET] <durandal_1707> yes, 23,20% ffmpeg libavfilter.so.7.46.101 [.] ff_fcmul_add_avx.loop VS 23,09% ffmpeg libavfilter.so.7.46.101 [.] ff_fcmul_add_sse3.loop [20:16:28 CET] <atomnuker> meh, seems like the difference is down in the noise [20:16:56 CET] <atomnuker> though avx should definitely be faster than sse due to register sizes [20:21:21 CET] <durandal_1707> https://pastebin.com/pgQJMz6K [20:32:54 CET] <atomnuker> can you post timer cycles instead of relative perf percentages? [20:33:51 CET] <atomnuker> also be sure the amount of data per call doesn't change or you'll be introducing noise [20:36:18 CET] <kierank> 19:16:53 <"atomnuker> though avx should definitely be faster than sse due to register sizes [20:36:21 CET] <kierank> not if it's memory bound [20:38:21 CET] <atomnuker> should still be faster, movs take the same amount of time for either register size [20:38:35 CET] <atomnuker> and page size >> register size [20:47:28 CET] <jamrial> durandal_1707: write a checkasm module for it. much better than timer.h benchmarks, as those include function call overhead [20:48:15 CET] <jamrial> should be trivial, just look at other float based modules, like float_dsp and synth_filter [20:48:24 CET] <durandal_1707> 703500 decicycles in fcmul_add, 261908 runs, 236 skips VS 687032 decicycles in fcmul_add, 261985 runs, 159 skips [21:09:52 CET] <durandal_1707> maybe asm can be overwritten to get better performance [21:16:50 CET] <iive> is the sse3 and avx assembly the same, with just different register sizes? [21:18:27 CET] <nevcairiel> I don't even see avx assembly in the source [21:18:29 CET] <nevcairiel> did you make it? [21:22:17 CET] <kurosu> from the name of the function, avx downclock may also be a thing [21:22:18 CET] <iive> durandal_1707, btw, on intel avx 256 bit processing is not enabled right away, the module boot up time is quite significant. [21:22:40 CET] <cone-597> ffmpeg 03Marton Balint 07release/4.1:ddc284300eb0: avfilter/af_asetnsamples: fix last frame props [21:22:41 CET] <iive> you may either boot it up during cpuid or use banchmark that runs for longer than second. [21:22:47 CET] <kurosu> also, what cpu? amd are almost all 128b AU, so no benefit of AVX vs xmm [21:23:08 CET] <kurosu> (except code size, sometimes) [21:23:29 CET] <iive> isn't xmm like mostly bigger? [21:23:42 CET] <kurosu> not always, iirc [21:23:42 CET] <iive> (as code size) [21:23:48 CET] <kurosu> different encoding of insn [21:24:01 CET] <iive> i thought it uses more prefixes. [21:27:34 CET] <kurosu> I don't think so, the opposite actually? evx prefix makes the insn bigger [21:41:12 CET] <durandal_1707> nevcairiel: what avx instructions i could use? [21:58:10 CET] <durandal_1707> are thy going to remove FMA support in CPUs? [21:58:48 CET] <jamrial> no [00:00:00 CET] --- Wed Jan 2 2019

1 0

ffmpeg.log.20190101
by burek 02 Jan '19

02 Jan '19

[01:00:53 CET] <ariyasu> Tue Jan 01 00:00:51 2019 [01:00:59 CET] <ariyasu> happy new year #ffmpeg [01:20:07 CET] <Tachyon> hny! :) [01:49:11 CET] <pomaranc> Tachyon: no [06:20:30 CET] <hungrywolf> does any one know any good tutorial on making a video player using ffmpeg? [06:21:03 CET] <hungrywolf> http://dranger.com/ffmpeg/tutorial01.html << this looks like a good one, but I am seeing many errors when I type in the code [14:27:12 CET] <hungrywolf_> can some one tell me what is the best resource for learning ffmpeg working with c++ [15:32:25 CET] <Reventlov> Hi. How to escape filenames to allow ffmpeg to use them? [15:32:46 CET] <Reventlov> (in a zsh function, e.g. force_subs() { ffmpeg -i "$@" -vf subtitles="$@" "forced_$@"; } ) [15:34:23 CET] <c_14> in the filter with single quotes [15:35:26 CET] <Reventlov> it seems it's not working as expected (still gives error when executed) [15:35:40 CET] <c_14> what error? [15:35:51 CET] <Reventlov> (what do you mean exactly? replace subtitles="$@" with what?) [15:36:23 CET] <c_14> subtitles=\'"$@"\' probably [15:40:11 CET] <Reventlov> well, yeah, thank you c_14 [15:40:14 CET] <Reventlov> bye [16:03:59 CET] <dv_> uh, terminology question. "frame dropped" vs "frame skipped", both during decoding, is there any meaningful difference? it sounds to me like a frameskip is a planned event, like skipping altref VP8 frames, while a framedrop is something unexpected, for example due to corrupted frames. [16:13:15 CET] <DHE> frames can be dropped for legit reasons as well. if you're reducing the framerate, those can be dropped frames. [17:24:27 CET] <dv_> DHE: yeah, but I am solely looking at frame skipping/dropping in a *decoder* [17:30:22 CET] <DHE> you mean the AVCodecContext specifically as opposed to ffmpeg? [17:34:18 CET] <dv_> I mean in general, not specifically ffmpeg [17:34:41 CET] <dv_> I keep reading about frameskipping and framedropping in decoders, and I wonder about their de facto meaning [17:35:09 CET] <dv_> the only distinction I found was that, as said, skipping = planned, normal ; dropping = abnormal, typically because of corrupted frames [17:48:11 CET] <DHE> oh... well, when you're trying to do realtime decoding (ie video playback) and you can't keep up - presumably because CPU is insufficient - something has to give. Maybe you skip drawing the frame on the screen, maybe you skip decoding entirely for some frames and hope to recover some other way (wait for a new keyframe?) [17:48:54 CET] <DHE> one interpretation might be skipping = not rendered on screen, dropped = not given to the decoder at all [17:52:37 CET] <iive> yes, libavcodec support such type of decode skipping. it could be tuned. e.g. skip only frames that are not used for reference (B-Framces) [17:53:01 CET] <iive> skip also forward reference frames (P-Frames).. [17:53:45 CET] <iive> you can also skip only parts of decoding e.g. skip loopfilter (internal codec post-processing). [18:09:31 CET] <DHE> right, but in this channel most people think either in terms of real-time encoding or non-realtime transcoding where decoding is largely irrelevant in the grand scheme of things [22:28:30 CET] <Celmor> how does ffmpeg decide parameters when I omit them? Usually it seems re-encoding downloaded files like for example a twitch vod makes them significantly smaller but I'm not sure how much quality I'd lose [22:29:02 CET] <BtbN> It uses whatever is set as default for the omitted parameter [22:29:06 CET] <BtbN> usually the defaults are bad [22:33:06 CET] <Celmor> so I assume quality loss would be significant? [22:34:56 CET] <Celmor> join #awk [22:35:00 CET] <Celmor> oops [22:35:06 CET] <furq> the defaults for mkv and mp4 are ok [22:35:11 CET] <furq> not good but ok [22:35:25 CET] <BtbN> Depends on the version [22:35:33 CET] <BtbN> it used to be like 200kbit or something [22:35:33 CET] <pink_mist> I thought the fileending has nearly no bearing on the quality of the encoding [22:35:38 CET] <furq> i'm assuming you have a version from the last couple of years with libx264 [22:35:39 CET] <BtbN> and was changed to some crf mode recently [22:35:40 CET] <furq> which most people do [22:36:09 CET] <furq> pink_mist: it has a bearing on what encoders ffmpeg picks [22:36:12 CET] <furq> by default [22:38:07 CET] <Celmor> I'm simply converting stream "dumps" from twitch, mp4 -> mp4, since the bitrate often is constantly unnecessarily around 5k I often get huge decrease in file size if I re-encode the file, omitting any parameters like -c, using latest version and default h264 transcoders [22:40:20 CET] <BtbN> Twitch is using a lot of filler to keep the bitrate constant. Just re-muxing their raw mpeg-ts to mp4 without re-encoding will shrink it. [22:42:44 CET] <Celmor> does "-c copy" also remux? [22:43:16 CET] <Celmor> till now I only experiences decrease in output file size if I omitt the "copy" for the codec [22:43:59 CET] <arejay> i have a .mkv file which consists of both 1080 and 720 resolution frames. curious if there a way ffmpeg can remove all the 1080 frames from the video file [23:27:29 CET] <DHE> arejay: you'll have to figure out which streams are the ones you want to keep (say, 0:0 and 0:2 (video and audio)), then: ffmpeg -i input.mkv -map 0:0 -map 0:2 -c copy output.mkv [23:28:03 CET] <DHE> "ffprobe input.mkv", or "ffmpeg -i input.mkv", will provide a summary of the input file which should be sufficient [23:54:59 CET] <arejay> DHE: dude, thank you!! exactly what i needed [00:00:00 CET] --- Wed Jan 2 2019

1 0