[Ffmpeg-devel-irc] ffmpeg.log.20161129

[burek]

[00:00:23 CET] <Guest94097> I am taking in a mpts via udp and trying to select a program and output to a file, but I am getting I/O error, anything wrong with this cli?
[00:00:51 CET] <Guest94097> ffmpeg -i udp://244.13.9.13:3100 -map 0:p:450 -c:v copy -c:a copy c:\temp\test.ts
[00:03:00 CET] <lvella> JEEB: from the MPEG-2 standard, the elements of a GOP are "pictures", and a "picture" is either a field, in interlaced mode, or a frame, in progressive mode, so that seems very weird that ffmpeg encoder would group the fields
[00:21:26 CET] <iive> mpeg2 standard mandates that two complementary fields are always encoded, so you alway have a full frame.
[00:29:43 CET] <durandal_1707> Hello71: play with overlap and video size
[00:33:08 CET] <DHE> Guest24481: IP addresses starting in 244 are in the reserved block. that's not a multicast address
[00:33:18 CET] <DHE> did you mean 224 instead?
[00:33:26 CET] <PlanC> does anyone know why the image that's in the meta data/ID3 of a file is removed when it's copied with ffmpeg?
[00:33:39 CET] <PlanC> everything else is still there but the album art/image/cover is missing
[00:33:56 CET] <PlanC> this guy had the same problem as me: http://superuser.com/questions/758338/keep-album-art-with-ffmpeg-while-cutting-a-mp3-file
[00:34:15 CET] <PlanC> but he never shared the solution... (https://xkcd.com/979/)
[00:34:35 CET] <DHE> there's an xkcd for everything. (I don't even have to click it)
[00:35:58 CET] <c_14> PlanC: internal reasons needs a patch
[00:36:18 CET] <PlanC> c_14: so it's a bug?
[00:36:26 CET] <c_14> Well
[00:36:38 CET] <c_14> It's a "feature" that wasn't very well thought out.
[00:37:03 CET] <PlanC> strange
[00:37:15 CET] <PlanC> because all the other data is coming with the output
[00:37:24 CET] <PlanC> the image is the only thing that's not
[00:37:50 CET] <PlanC> tried playing around with  -map_metadata too without any luck
[00:38:14 CET] <PlanC> is there any sort of "hack" that I can use to fix it?
[00:38:34 CET] <c_14> https://ffmpeg.org/ffmpeg-formats.html#mp3
[00:38:44 CET] <c_14> the second example
[00:39:09 CET] <c_14> the cover art should automatically be turned into a video stream in the input file
[00:40:38 CET] <PlanC> the problem is that the image is in the file though
[00:40:53 CET] <c_14> the cover art should automatically be turned into a video stream in the input file
[00:40:56 CET] <c_14> ^like I said
[00:41:05 CET] <c_14> you (probably) just need the -metadata stuff
[00:46:49 CET] <PlanC> tried it just now
[00:46:59 CET] <PlanC> it gives this in the output
[00:47:05 CET] <PlanC> "No packets were sent for some of the attached pictures."
[00:47:21 CET] <c_14> try explicitly mapping it -map 0:v
[00:47:29 CET] <c_14> you'll also need -map 0:a
[00:50:15 CET] <PlanC> I'm doing something wrong
[00:50:19 CET] <PlanC> this is the command that I'm running:
[00:50:32 CET] <PlanC> ffmpeg -i art.mp3 -c copy -map 0:v -map 0:a -metadata:s:v title="Test cover" -metadata:s:v comment="Cover (Front)" -ss 20 -t 30 out.mp3
[00:50:47 CET] <PlanC> that last "-ss 20 -t 30" is the part that I'm cutting
[00:51:12 CET] <PlanC> art.mp3 contains the image
[00:51:32 CET] <PlanC> out.mp3 which is the trimmed output doesn't have it
[00:56:15 CET] <c_14> It doesn't seem to work with seeking
[00:57:21 CET] <PlanC> ah crap
[00:57:34 CET] <PlanC> and that's going to dramatically increase the time it takes to process the file, right?
[00:58:17 CET] <Hello71> durandal_1707: I halved the horizontal resolution and stretched it but then it looks jerky
[00:58:44 CET] <c_14> PlanC: ffmpeg -ss 10 -t 20 -i art.mp3 -i art.mp3 -map 0:a -map 1:v -metadata:s:v title="Test cover" -metadata:s:v comment="Cover (Front)" -c:a copy out.mp3
[00:58:46 CET] <c_14> ^that works
[01:00:38 CET] <PlanC> checking it out
[01:01:41 CET] <PlanC> c_14: wow that really does work
[01:01:45 CET] <PlanC> c_14: thanks a lot mate!
[01:02:53 CET] <PlanC> so you basically selected the same file twice and got the content and ID3 from the first file and the image from the second, right?
[01:03:07 CET] <c_14> pretty much
[01:03:09 CET] <c_14> well
[01:03:17 CET] <c_14> I'm not sure which file it's getting the ID3 from
[01:03:19 CET] <c_14> one of them
[01:03:25 CET] <c_14> but since they're the same file it shouldn't matter
[01:03:49 CET] <PlanC> ah that's clever
[01:03:51 CET] <PlanC> thanks again!
[01:09:10 CET] <Freakshow> quick and potentially silly question...
[01:09:46 CET] <Freakshow> does the ffmpeg process wait for the log printf statement to complete before working on the next frame of video?
[01:10:44 CET] <c_14> depends on where the log is coming from, but probaly yes
[01:10:47 CET] <c_14> *probably
[01:12:10 CET] <Freakshow> c_14, the log I am referencing is the output log from ffmpeg itself&
[01:12:31 CET] <Freakshow> whether its on console or just written to disk
[01:12:48 CET] <c_14> Like I said, it depends. But probably yes.
[01:12:54 CET] <Freakshow> sure sure
[01:12:57 CET] <Freakshow> thanks
[01:32:07 CET] <PlanC> c_14, I just noticed that there seems to be something wrong with the start and end times though
[01:32:42 CET] <PlanC> c_14, for example -ss 10 -t 20 should output a 20 a file with a 20 second duration
[01:33:13 CET] <PlanC> c_14, but when I run it with the -map 0:a -map 1:v that you sent me it outputs a 30 second file
[01:35:45 CET] <PlanC> and when I move the -ss and -t part it doesn't come with the cover
[01:36:33 CET] <PlanC> it's respecting -ss but not the -t
[01:37:03 CET] <PlanC> so if I use "-ss 10 -t 10" for a 1 minute file
[01:37:24 CET] <PlanC> then it'll actually start from 10 seconds but it'll ignore the "-t" and continue to the end of the file
[01:37:39 CET] <PlanC> so the output would be 50 seconds long instead of 10 seconds
[01:54:41 CET] <klaxa> PlanC: afaik -t is an output option
[01:54:53 CET] <klaxa> use ffmpeg -ss 10 -i somefile.mp4 -t 10 output.mp4
[01:59:54 CET] <PlanC> klaxa, I played around with it earlier and moving it to the end did the trick just like you showed
[02:05:52 CET] <PlanC> thanks for the tip btw
[05:16:50 CET] <tomed> what advantage do i get by using  "very slow" setting over "fast" setting
[05:41:00 CET] Last message repeated 1 time(s).
[06:13:26 CET] <naylo44> Hey, I was wondering if anyone was able to compile ffmpeg using the --enable-nvresize and --enable-nvenc flags. I'm trying to follow these directions: http://developer.download.nvidia.com/compute/redist/ffmpeg/1511-patch/FFMPEG-with-NVIDIA-Acceleration-on-Ubuntu_UG_v01.pdf but they seem pretty outdated. Thanks!
[06:23:32 CET] <c_14> naylo44: nvenc should be enabled automatically if it's supported
[06:23:35 CET] <c_14> nvresize no idea
[06:24:19 CET] <c_14> (in recent enough versions)
[06:24:22 CET] <c_14> so 3.2 or git-master
[06:30:42 CET] <naylo44> Thanks! I tried transcoding a video I had and only achieved ~23FPS on a GTX 770 using NVENC. I think this bad performance might be because of the missing nvresize flag. I might be worng though. Thanks!
[06:48:36 CET] <Guest60667> Is there any flags that I can use to ignore the bad response (missing Transport?) for RTSP SETUP from the embedded RTSP server on an Amino STB? http://pastebin.com/NEAQBX8e
[09:00:57 CET] <kurufu> So, av_interleaved_write_frame expects frames across all streams to have monotonically increasing dts? Not just frames within a single stream?
[10:18:11 CET] <MZ_> Hi, I'm new to ffmpeg. I've added swscale, avcodec and avutil dll in my code. compile and link is done successfully but when I run my program I'll get this message: "the program can't start because out.dll is missing from your computer". So the question is: what is out.dll? I missed some dependencies? how can I fix this error?   thank you for your time
[10:45:42 CET] <alexvf> MZ_: are you linking with out.dll? sounds like a mock library taken from some compilation sample?
[10:46:11 CET] <alexvf> MZ_: i'm no expert but that out.dll shouldn't be anything related to ffmpeg
[10:46:13 CET] <MZ_> i do not know what is out.dll. i don't have it
[10:46:58 CET] <MZ_> alexvf, i really cannot understand why there is that error message. i never used anything named "out.dll"
[10:47:35 CET] <alexvf> MZ_: can you show how do you build your program?
[10:48:18 CET] <MZ_> alexvf, I'm using visual studio, what I need to show you?
[10:49:14 CET] <alexvf> bfff, i've never used that, in fact i rarely compile nothing on windows
[10:49:51 CET] <alexvf> you should look where you add avcodec, avutil and all the dependencies that your code uses
[10:50:10 CET] <superware> I'm trying to have https://trac.ffmpeg.org/ticket/5615 resolved, problem is I don't have sufficient knowledge to debug this myself. Can someone please help? it's extremely easy to reproduce..
[10:51:52 CET] <MZ_> alexvf, I added avcodec.lib, avutil.lib and swscale.lib to the Linker Additional Dependencies. And in the code I #include the headers inside extern "C"{} block. (I'm coding in C++). nothing more..
[10:55:42 CET] <alexvf> MZ_: sorry, i can't help with the Visual Studio tool, if it is a small program, you could try to compile it without Visual Studio but, again, i'm of little help because i don't develop under Windows
[10:56:14 CET] <MZ_> alexvf, maybe I found the problem. I found some "out.dll" string inside the binary of the avcodec.lib file... I will work on that.   thank you for your time!
[10:56:41 CET] <alexvf> MZ_: i was about suggesting you that
[10:56:49 CET] <alexvf> MZ_: looking in the binaries
[10:57:22 CET] <alexvf> MZ_: it may be a problem in the libraries itself, good luck!
[11:01:04 CET] <MZ_> alexvf, thank you XD
[12:20:15 CET] <MZ_> Hi, avcodec_send_packet() give me an error code that I cannot understand. The error is  -1094995529   How/Where can I look for understanding it?   (i solved the previous problem about the dll)
[12:27:41 CET] <jkqxz> MZ_:  See libavutil/error.h.
[12:29:50 CET] <MZ_> jkqxz, ok, thanks
[13:29:05 CET] <fling> Hello!
[13:29:14 CET] <fling> I have a video from a dancing contest.
[13:29:40 CET] <fling> There is a head from audience appearing on top of the background in a part of the video.
[13:30:06 CET] <fling> The idea is to overlay this part with the static solid background from the different time of the same video to hide this head.
[13:30:13 CET] <fling> What is the proper way doing so?
[13:45:25 CET] <superware> fling: it sounds like you need a video editing software.
[13:45:27 CET] <ikevin> fling, maybe adding a static png with transparency and your background can do the job
[13:54:14 CET] <fling> superware: but I thought ffmpeg cat overlay videos!
[13:54:32 CET] <fling> ikevin: why static?
[13:54:45 CET] <fling> s/cat/can/
[13:54:56 CET] <ikevin> a mistake, png are always statics
[13:55:52 CET] <fling> But why not video?
[13:56:01 CET] <fling> Is it just simplier?
[13:56:08 CET] <ikevin> create a png with the same size than the video and with transparent background, add your logo where you want to hide the logo
[13:56:12 CET] <ikevin> yes
[13:56:35 CET] <fling> It is not a logo but a moving head of a person but I got your idea.
[13:56:57 CET] <fling> I will try to do this!
[13:58:11 CET] <ikevin> oh, if it's moving that will need more work
[14:10:43 CET] <fling> Which picture file format for a single yuv 422 frame?
[16:11:24 CET] <superguest> http://pastebin.com/FfpeLt6m
[16:13:00 CET] <superguest> Basically ffmpeg doesn't recognize the 'localtime' format argument for 'pts' function for the 'text' parameter.
[16:13:24 CET] <superguest> https://ffmpeg.org/ffmpeg-filters.html#Text-expansion
[16:14:14 CET] <superguest> [Parsed_drawtext_0 @ 0x992dc0] Invalid format 'localtime'
[16:14:16 CET] <DHE> but you're running version 2.8 whereas those docs will be based on the most up-to-date version available
[16:15:18 CET] <superguest> :< shiieeet.
[16:15:48 CET] <superguest> DHE, stable or devel?
[16:16:26 CET] <DHE> umm... not entirely sure actually
[16:16:42 CET] <DHE> I would guess release version
[16:37:40 CET] <superguest> I am using Ubuntu Xenial (16.04.1), and 2.8.8 is the latest version in their official repo
[16:37:56 CET] <superguest> I don't want to build ffmpeg from source
[17:24:46 CET] <elmarikon> cheers! Can someone help me with the loudnorm Filter!? Is it still in development? It produces results that vary big time from what I am telling it to do...
[17:25:16 CET] <elmarikon> my settings: loudnorm=-23:15:-1
[17:26:39 CET] <elmarikon> result: Output Integrated: -25.1 LUFS (should be -23), Output True Peak: -1.0 dBTP (ok), Output LRA: 20.5 LU (should be mx. 15)
[19:20:14 CET] <tomed> what advantage do i get by using  "very slow" setting over "fast" setting
[19:21:27 CET] <BtbN> quality vs speed
[19:23:31 CET] <DHE> it will take a lot longer, but the image quality should be better or the resulting file size will be smaller (depending on settings)
[19:54:37 CET] <KDDLB> happy birthday `md!
[19:54:41 CET] <kode54> yes
[19:55:57 CET] <tomed> dhe  i've seen slower setting producing bigger file
[19:56:42 CET] <DHE> with what settings though?
[19:56:54 CET] <DHE> I mean using -qp, -crf or -b may result in different effects
[20:01:51 CET] <tomed> crf20/very slow  vs   crf20/fast
[20:06:11 CET] <DHE> hmm.. not what I would have expected
[20:09:46 CET] <JEEB> the result of crf depends completely on the preset, and you should not expect the same crf to be comparable between presets
[20:10:05 CET] <JEEB> depending on the content you will also get results where a much slower preset gives you a larger size for the same crf
[20:10:48 CET] <JEEB> you can think of it as the encoder being able to "see" more things and do more things as you enable a slower preset, thus the meaning of the CRF value changes
[20:11:24 CET] <JEEB> CRF was never meant to be a stable base between presets, QED
[20:11:43 CET] <JEEB> or well, by the pure fact that it isn't such a thing and never was such
[20:41:26 CET] <kerio> qp ftw
[21:02:14 CET] <kriskropd> hello - I have a ton of sound effects and stuff, but the problem is a lot of them are collected in groups of 3 or 4 per file - they are each separated by some time of silence though
[21:04:08 CET] <kriskropd> I wonder if I might be able to use fmfpeg or some other tool to extract each sound effect from their grouped wav file using the audio silence as a delimeter
[21:08:17 CET] <kriskropd> i think sox might be able to do this too actually
[21:12:09 CET] <durandal_1707> kriskropd: there are ffmpeg filters like silencedetect
[21:13:27 CET] <kriskropd> durandal_1707: http://unix.stackexchange.com/questions/318164/sox-split-audio-on-silence-but-keep-silence I found this and I just adjust the silence threshold to a shorter distance
[21:13:31 CET] <kriskropd> adjusted*
[21:13:36 CET] <kriskropd> worked for me
[21:16:42 CET] <durandal_1707> kriskropd: same filter is available in ffmpeg, silenceremove
[21:18:37 CET] <kriskropd> noted - I'll give that a try too then, since I don't have sox on the cygwin computer at home
[21:18:55 CET] <kriskropd> this is going to be a massive time saver for me x.x
[21:20:09 CET] <kriskropd> durandal_1707: I think what I'm not understanding is how to split the audio with ffmpeg
[21:20:26 CET] <kriskropd> I've done it before using timestamps and a very lengthy file, but I don't see anything in this man page about splitting files based on silence
[21:25:10 CET] <durandal_1707> kriskropd: there is special muxer for that...
[21:31:46 CET] <Filarius> how do YUV420p looks like ? (Height) lines of (Width) bytes of Y, after that (Height/4) lines of (Width/4) bytes of U and same for V ?
[21:34:34 CET] <durandal_1707> Filarius: that's more like yuv410 iirc
[21:35:02 CET] <durandal_1707> 440 actually
[21:35:41 CET] <durandal_1707> 420 is divided by 2
[22:02:48 CET] <Filarius> oh, ye, w/2 and h/2. So i'm right? just need to undertand coordinates of pixels in 8x8 rectangles in yuv420p
[22:15:30 CET] <jermz> http://pastebin.com/5bkB9RV9 gives me the error "Warning: data is not aligned! This can lead to a speedloss", while http://pastebin.com/XqvS8LhS does not.
[22:15:48 CET] <jermz> The only difference between these two commands is the padding x coordinate which moves it over 280 px to center it.
[22:16:48 CET] <jermz> Any idea how to center the video without getting the error?
[22:23:33 CET] <Filarius> maybe it do not know what put at background ?
[22:24:01 CET] <jermz> the background is just supposed to be a blank black background. I am using a PNG as an overlay in a separate command to fill the space
[23:14:32 CET] <Guest60667> Is there any flags that I can use to ignore the bad response (missing Transport?) for RTSP SETUP from the embedded RTSP server on an Amino STB? http://pastebin.com/NEAQBX8e
[00:00:00 CET] --- Wed Nov 30 2016