[Ffmpeg-devel-irc] ffmpeg.log.20190412

burek burek021 at gmail.com
Sat Apr 13 03:05:02 EEST 2019


[00:01:02 CEST] <faLUCE> JEEB: it's strange that this discussion is of about 10 years ago...
[00:01:08 CEST] <faLUCE> what happened meanwhile?
[00:01:18 CEST] <JEEB> (´4@)
[00:04:32 CEST] <faLUCE> JEEB: note that:
[00:04:33 CEST] <faLUCE> https://www.openh264.org/faq.html
[00:04:54 CEST] <faLUCE> " In order for Cisco to be responsible for the MPEG LA licensing royalties for the module, Cisco must provide the packaging and distribution of this code in a binary module format (think of it like a plug-in, but not using the same APIs as existing plugins), in addition to several other constraints"
[00:05:08 CEST] <JEEB> yes, cisco distributes binaries that f.ex. firefox then downloads
[00:05:21 CEST] <JEEB> this way cisco is distributing the binaries
[00:05:25 CEST] <faLUCE> JEEB: it confirmes what the quoting of ryan said
[00:05:58 CEST] <JEEB> it's the more or less common understanding of what *MPEG-LA* wants
[00:07:19 CEST] <faLUCE> JEEB: so, openh264 source code doesn't pay MPEG-LA ?
[00:07:43 CEST] <JEEB> what
[00:08:39 CEST] <faLUCE> JEEB: yes, look at that. The binary has another license than the source code
[00:08:43 CEST] <faLUCE> "Additionally, this binary is licensed under Ciscos AVC/H.264 Patent Portfolio License from MPEG LA, at no cost to you, provided that the requirements and conditions shown below in the AVC/H.264 Patent Portfolio sections are met.  "
[00:08:53 CEST] <JEEB> that is not a software license
[00:08:56 CEST] <faLUCE> while the source code is BSD
[00:08:59 CEST] <JEEB> that is a patent license
[00:09:02 CEST] <JEEB> separate things
[00:09:22 CEST] <faLUCE> ok, but I mean: the patent license is applied only to the binary
[00:09:28 CEST] <faLUCE> not to the source code
[00:09:41 CEST] <faLUCE> even if the source code has patented algos
[00:10:19 CEST] <JEEB> that is the common understanding of how it goes with MPEG-LA, yes. do note that unless cisco has this in writing from mpeg-la, mpeg-la could choose to change its mind
[00:10:34 CEST] <JEEB> of course probably, maybe that is unlikely to happen
[00:10:38 CEST] <JEEB> but welcome to legalese
[00:10:58 CEST] <JEEB> although cisco already pays the maximum license amount
[00:11:01 CEST] <faLUCE> JEEB: ok, but this license clarifies all
[00:11:05 CEST] <JEEB> so for cisco it wouldn't matter at all :P
[00:12:15 CEST] <faLUCE> FOSS source code doesn't require license from MPEG LA, while binary requires patent license
[00:12:31 CEST] <faLUCE> they WROTE that, so....
[00:12:37 CEST] <JEEB> cisco did, yes
[00:12:51 CEST] <JEEB> I'm not saying this is not the common understanding
[00:12:53 CEST] <JEEB> just details :P
[00:13:19 CEST] <faLUCE> JEEB: but note this is an important doc, because your thread is old and without explicit quotes
[00:13:30 CEST] <faLUCE> instead, this is a sort of legal doc by cisco
[00:13:31 CEST] <JEEB> is this an explicit quote?
[00:13:40 CEST] <faLUCE> yes, it's the cisco license
[00:13:45 CEST] <faLUCE> https://www.openh264.org/BINARY_LICENSE.txt
[00:14:25 CEST] <JEEB> yup, that's just cisco's understanding
[00:14:39 CEST] <JEEB> most likely they talked with MPEG-LA and it will hold :P
[00:14:48 CEST] <faLUCE> ok, but it something more "official"
[00:14:53 CEST] <faLUCE> than the old thread
[00:15:00 CEST] <faLUCE> (it is)
[00:15:27 CEST] <faLUCE> JEEB: otherwise even CISCO would be in violation!
[00:15:42 CEST] <JEEB> of course, but they already pay the max :D
[00:15:55 CEST] <JEEB> that's why they are distributing it
[00:16:00 CEST] <JEEB> because it doesn't cost anything to them
[00:16:03 CEST] <JEEB> and it's good PR
[00:16:44 CEST] <JEEB> faLUCE: also was there a spot where it notes that the source code is not applicable? either in the txt or the site.
[00:16:51 CEST] <JEEB> we both know that's most likely what MPEG-LA thinks
[00:16:56 CEST] <JEEB> but yunno :P welcome to legal stuff
[00:16:59 CEST] <JEEB> it's all vague
[00:17:19 CEST] <faLUCE> JEEB: yes, there is that spot:
[00:17:21 CEST] <faLUCE> https://www.openh264.org/faq.html
[00:17:47 CEST] <faLUCE> "In order for Cisco to be responsible for the MPEG LA licensing royalties for the module, Cisco must provide the packaging and distribution of this code in a binary module format (think of it like a plug-in, but not using the same APIs as existing plugins), in addition to several other constraints. This gives the community the best of all worlds - a team can choose to use the source code, in which case the team is
[00:17:48 CEST] <faLUCE> responsible for paying all applicable license fees, or the team can use the binary module distributed by Cisco, in which case Cisco will cover the MPEG LA licensing fees."
[00:18:18 CEST] <faLUCE> ----> this means that the first one who makes binaries, has to pay MPEG-LA
[00:18:25 CEST] <JEEB> no
[00:18:26 CEST] <JEEB> not make
[00:18:27 CEST] <JEEB> distribute
[00:18:30 CEST] <faLUCE> yes
[00:18:32 CEST] <faLUCE> distribute
[00:18:45 CEST] <JEEB> I could make the binary to a company, and the company would distribute it
[00:19:15 CEST] <JEEB> according to the common understanding the company would thus be the end-user distributor
[00:20:15 CEST] <JEEB> faLUCE: also do note how that quote does not specifically say that the source code is not relevant to the MPEG-LA licensing. it's the common understanding, but it is /not/ mentioned there.
[00:20:23 CEST] <JEEB> not trying to be an asshole, just noting it :P
[00:22:01 CEST] <faLUCE> JEEB: it says: "a team can choose to use the source code, in which case the team is  responsible for paying all applicable license fees"
[00:22:16 CEST] <faLUCE> which means that "source code itself is not subject to fees"
[00:22:45 CEST] <damdai> anybody know what it means by "fragmented" and "cbcs" ?  http://download.opencontent.netflix.com/?prefix=AV1/Chimera/
[00:24:29 CEST] <JEEB> faLUCE: no it does not. it just says that if you don't use cisco's binary it is of course not licensed
[00:24:47 CEST] <JEEB> you can /read/ it like you do
[00:24:58 CEST] <JEEB> it doesn't /say/ it
[00:25:29 CEST] <faLUCE> JEEB: I'm talking about the non-binary case
[00:25:36 CEST] <faLUCE> for an user too
[00:25:59 CEST] <faLUCE> so, an user that uses cisco's code for making another non-binary library
[00:27:16 CEST] <JEEB> faLUCE: but see how it specifically say what you read out of it :P it does /not/ say "source code itself is not subject to fees". it /only/ says "if you choose to use the source code, you are responsible for handling any applicable license fees"
[00:27:33 CEST] <faLUCE> JEEB: ok, but it can be deduced
[00:27:46 CEST] <JEEB> no it cannot be. it is the common understanding though.
[00:27:53 CEST] <JEEB> that mpeg-la does not care about source
[00:28:05 CEST] <JEEB> anyways, good night :)
[00:28:12 CEST] <faLUCE> good night :-)
[00:29:40 CEST] <kevinnn> does anyone know why some routers block UDP traffic and others don't? Specifically for rtsp streams
[00:31:05 CEST] <faLUCE> kevinnn: by closing ports
[00:31:06 CEST] <faLUCE> ;-)
[00:31:54 CEST] <kevinnn> faLUCE: how do I get my routers not to do that? it is affecting my performance
[00:31:55 CEST] <kevinnn> also...
[00:32:01 CEST] <kevinnn> apple's facetime
[00:32:06 CEST] <kevinnn> that goes over rtp
[00:32:10 CEST] <kevinnn> which is UDP
[00:32:18 CEST] <kevinnn> but it doesn't seem to be blocked
[00:32:37 CEST] <faLUCE> kevin, be more precise
[00:32:40 CEST] <kevinnn> or do they also revert to TCP when UDP is blocked
[00:32:48 CEST] <faLUCE> yes, it can be
[00:32:58 CEST] <faLUCE> kevinnn: you should a test with netcat
[00:33:05 CEST] <kevinnn> but the other thing is I tested with netcat
[00:33:08 CEST] <kevinnn> yes exactly
[00:33:20 CEST] <faLUCE> and what netcat said?
[00:33:21 CEST] <kevinnn> and it seems like UDP traffic gets through just fine
[00:33:32 CEST] <kevinnn> but for some reason rtsp still defaults to tcp
[00:33:35 CEST] <kevinnn> as opposed to UDP
[00:33:41 CEST] <kevinnn> it is very confusing
[00:33:55 CEST] <faLUCE> kevinnn: it works in this way
[00:34:13 CEST] <faLUCE> the RECEIVER asks to the STREAMER a rtsp description
[00:34:38 CEST] <faLUCE> then the receiver allocates resources based to this description
[00:34:50 CEST] <faLUCE> then the streamer sends stuff over UDP
[00:35:24 CEST] <faLUCE> RTSP is TCP, while RTP is UDP
[00:35:54 CEST] <faLUCE> now, given that your player doesn't allow you to allocate a netcat receiver
[00:36:02 CEST] <faLUCE> use a SNIFFER
[00:36:04 CEST] <faLUCE> ok?
[00:36:29 CEST] <kevinnn> faLUCE: okay use a sniffer to determine if rtp is being passed via UDP or tcp?
[00:36:52 CEST] <faLUCE> yes
[00:37:07 CEST] <kevinnn> because on some routers (like mine) vlc produces this error when connecting to an rtsp stream: no data received in 10s. Switching to TCP
[00:37:17 CEST] <kevinnn> okay, let me download and install a sniffer
[00:38:10 CEST] <faLUCE> [00:37] <kevinnn> because on some routers (like mine) vlc produces this error when connecting to an rtsp stream: no data received in 10s. Switching to TCP  <--- then you should already know if it's TCP or UDP
[00:38:29 CEST] <kevinnn> right... but I don't want it to be
[00:38:37 CEST] <kevinnn> I want it to be UDP
[00:38:50 CEST] <kevinnn> and it is my router so I should be able to configure it like so
[00:39:03 CEST] <kevinnn> but there don't appear to be any settings that allow me to adjust that
[00:39:42 CEST] <faLUCE> kevinnn: this is another problem
[00:40:10 CEST] <kevinnn> faLUCE: I know... but for my specific use case TCP adds too much latency
[00:40:17 CEST] <kevinnn> the stream must be in UDP
[00:40:46 CEST] <faLUCE> kevinnn: first check if it's TCP instead of UDP
[00:41:04 CEST] <kevinnn> okay, I will check with wireshark, give me a sec
[00:41:25 CEST] <faLUCE> kevinnn: also note that TCP doesn't introduce significant latency
[00:41:33 CEST] <faLUCE> you can drop frames
[00:41:57 CEST] <kevinnn> for bi-directional streaming, every millisecond of latency counts
[00:42:28 CEST] <faLUCE> kevinnn: not if they have the same latency
[00:43:15 CEST] <kevinnn> TCP always higher latency than UDP does doesn't it?
[00:43:31 CEST] <kevinnn> I mean it physically feels slower when i use tcp vs udp
[00:43:55 CEST] <faLUCE> yes, but it doesn't mean that it's higher in a significant way for you
[01:48:01 CEST] <garyserj> I have this video that won't upload to youtube. I used ffmpeg to cut 1 second of it so I have a very small example file. That gives the same error from youtube  The video is available here http://ge.tt/1wtKlWv2   and is 98KB in size.   The story is I had a bunch of .3gp videos on my phone, I converted them to mp4. For some inexplicable reason a few 90 degrees rotated or needed a vertical flip,
[01:48:01 CEST] <garyserj> which I did with -vf "transpose=1"  or -vf "transpose=vflip".  One of the videos (I don't know if it's one that needed rotating or flipping), but it's not uploading.  mediainfo shows https://pastebin.com/raw/YXdiTxeu  e.g. mp4(AVC,AAC) so I am not sure the problem or  how  can change it so it uploads.
[02:38:23 CEST] <fella> garyserj: nothing that strikes my eye, but see for yourself: https://support.google.com/youtube/answer/1722171
[02:38:39 CEST] <fella> what's the err you get?
[03:08:25 CEST] <garyserj> fella: "The video has failed to process. Please make sure you are uploading a supported file type."
[03:37:49 CEST] <garyserj> fella: just tried importing to movie maker and saving, and it worked, and I made a cut of the working one. http://ge.tt/1jovrWv2    i'm interested if i can use ffmpeg to turn the bad one into the good one?
[08:44:23 CEST] <fella> garyserj: could you post the 'mediainfo' of the working one as well?
[12:12:16 CEST] <aleek> hello!
[12:14:07 CEST] <aleek> we are trying to transcode live stream of regular tv. We have mpeg2ts subtitles Elementary Stream, but usually it is empty, because there is no subtitles at the moment. I cannot start ffmpeg, until it finds at least one subtitle frame. Can we somehow overcome that?
[12:19:00 CEST] <JEEB> not sure the probe function would do that
[12:37:53 CEST] <DHE> failure to detect the input codec causing fail?
[12:38:34 CEST] <xerpi> Hi, I'm trying to use ffmpeg (libavcodec to do video decoding specifically) on bare-metal, without any OS beneath
[12:38:56 CEST] <xerpi> I can provide my own implementations of simple libc functions such as memcpy, strlen, etc
[12:39:45 CEST] <xerpi> But I see that for example file_open.c of libavutil, which uses open() and the like, is unconditionally compiled
[13:43:26 CEST] <dongs> JEEB: im sorry, im gulag'd in korea/taiwan so i probably missed your message. i saw a highlight here. but it scrolled off my screen
[14:42:09 CEST] <Numline1> Hey folks. I'm looking to implement this one thing in Go and I wonder if I could use ffmpeg for that. I need to process a video and extract one frame for each second of the video. This frame should be as stable as possible, meaning no blurry mid-frames and stuff. Would this be possible? Thanks!
[14:43:05 CEST] <BtbN> That depends on the material. If the source material uses tons of motion blur, you're gonna get motion blur.
[14:45:29 CEST] <BtbN> Extracting frames will not mix multiple together unless you explicitly use a filter for that.
[14:48:19 CEST] <Atlenohen> furq: Looks like it's the other way around /GS checks for buffer security, not disables it https://docs.microsoft.com/en-us/cpp/build/reference/compiler-options-listed-by-category?view=vs-2019
[14:49:05 CEST] <Atlenohen> unless, the minus after it
[14:49:44 CEST] <Atlenohen> it was typed -GS-  so I guess - at the end would mean to disable, I was wondering what does it mean I guess I figured it out
[14:52:28 CEST] <Numline1> BtbN yeah I was either hoping I could find the most "perfect" frame or mix multiple ones. Are there any resources on this online?
[14:55:23 CEST] <BtbN> mixing multiples ones will only make matters worse
[14:55:45 CEST] <BtbN> But generally, just set the fps to 1, and you will get, well, one frame per second.
[14:56:14 CEST] <Numline1> BtbN Fair enough. I found this - https://www.bugcodemaster.com/article/extract-images-frame-frame-video-file-using-ffmpeg which seems to be what I need. I was just hoping there's a filter that could look for stable frames
[14:59:29 CEST] <furq> Numline1: check the thumbnail filter
[14:59:38 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#thumbnail
[15:00:49 CEST] <Numline1> furq that'll probably do. Thanks! Are there any details on how that filter works?
[15:00:59 CEST] <Numline1> by that I mean what are the methods it uses under the hood
[15:01:43 CEST] <furq> https://notbrainsurgery.livejournal.com/29773.html
[15:01:46 CEST] <furq> apparently it's derived from this
[15:02:18 CEST] <Numline1> thank you :)
[15:02:27 CEST] <Numline1> also, what I like is that there's a filter named "ass"
[15:02:37 CEST] <Numline1> I'm going to use that even though I might not need it
[15:08:12 CEST] <iive> well, there is also file format .ass
[15:08:37 CEST] <iive> that this filter renders
[15:10:05 CEST] <Numline1> one could say it consumes ass
[15:55:15 CEST] <killown> hi there
[15:55:22 CEST] <killown> ffmpeg -y -ss 00:00:10.0 -i *BuUOl7HKM8U.mkv -strict -2 -c copy -t 00:00:13.0 ../videos/video.mp6
[15:55:25 CEST] <killown> mp4
[15:55:37 CEST] <killown> why ffmpeg is cuting random times?
[15:56:14 CEST] <killown> I want to cut 3 seconds from starting from second 10
[15:57:13 CEST] <ritsuka> you can't cut at a random point if the video format uses P and B frames
[15:57:23 CEST] <DHE> you must cut starting on a keyframe
[16:08:52 CEST] <killown> how is that?
[16:08:57 CEST] <killown> starting on a keyframe?
[16:09:12 CEST] <killown> cannot cut from 10 seconds?
[16:09:22 CEST] <furq> only if there's a keyframe at that timestamp
[16:09:27 CEST] <HickHoward> okay, so i did... this
[16:09:27 CEST] <HickHoward> ffmpeg.exe -i "H:\Tourist Trophy (Japan)\tt_vol\openingW.m2v" -vf "il=1:1:1,setfield=tff,separatefields" -vcodec rawvideo -pix_fmt bgr24 openingW.avi
[16:09:30 CEST] <furq> otherwise it'll seek to the (closest/next) one
[16:09:39 CEST] <HickHoward> now the final .avi video is looking all jittery every 2 frames
[16:10:09 CEST] <HickHoward> this is especially noticeable on parts of the video in which only a static image is displayed for like 5 seconds
[16:12:10 CEST] <killown> furq how do I check that?
[16:12:48 CEST] <furq> well there almost certainly isn't one there
[16:12:52 CEST] <furq> so you probably just want to reencode the video
[16:13:26 CEST] <furq> HickHoward: what's that even supposed to do
[16:13:49 CEST] <killown> furq could you help me reencode the vide to have this keyframe?
[16:14:02 CEST] <furq> killown: i don't mean reencode the entire video
[16:14:23 CEST] <furq> you can't cut at arbitrary points and use -c copy
[16:14:31 CEST] <furq> because the output video has to start on a keyframe
[16:14:31 CEST] <DHE> move your -ss to after the input file, and change -c to an actual codec. plus codec options suitable to your needs
[16:14:40 CEST] <furq> if you cut and reencode then it'll work fine
[16:19:42 CEST] <HickHoward> furq: not do this shit to me -> https://diff.pics/2vcOo4y6Z3ve/1
[16:31:05 CEST] <killown>  ffmpeg -y  -i *BuUOl7HKM8U.mkv -acodec aac -strict experimental   -ss 0:00:10 -t 0:00:13  ../videos/A...
[16:31:08 CEST] <killown> still not working
[16:35:03 CEST] <quakenet>  INVITED: Hello I invite you to a great chat service >< IRC quakenet  ORG <<< #philosophy  #DALnet , #NASA , #Freenode , #LOVE
[16:45:52 CEST] <zalaare> https://pastebin.com/raw/Gg8wUS8x  Can someone point out where the "other" filters exist that ffmpeg is complaining about here?  I only thought I was using 1 video and 1 audio.
[17:16:02 CEST] <Atlenohen> hey guys, VS2017 MSVC doesn't list any linker/compiler arguments for -L from --extra-ldflags=-L../build/lib
[17:17:04 CEST] <Atlenohen> I see where was -IInclude like if "Include" is an environmental variable would it for both included headers and libs then?
[17:47:06 CEST] <Atlenohen> pretty much the only thing left is ERROR: libx264 not found
[17:54:53 CEST] <Atlenohen> Ah it may just be a bash/VC env variable problem, i'll ask on ##linux
[17:56:37 CEST] <DHE> pkg-config? maybe you need to set PKG_CONFIG_PATH=/usr/local/lib/pkg-config if you used configure defaults on x264 when building from source
[17:56:48 CEST] <DHE> dunno how that works on VC...
[17:58:27 CEST] <Atlenohen> Well, it's a MINGW64 shell created from a VS2017 Developer Command prompt which sets environmental variables for that windows (without having to set them up in windows)
[17:59:07 CEST] <Atlenohen> I had manually created added more paths and it didn't work, looks like it ignores system ones, that or I need to reboot the machine for it to confirm
[17:59:48 CEST] <Atlenohen> I'll first reboot and then all I think I need to figure out is how to input new env into bash
[18:00:08 CEST] <Atlenohen> oh I don't have PKG at all, i'll add if it's needed, brb
[18:06:01 CEST] <Atlenohen> back
[18:06:40 CEST] <Atlenohen> Looks like the windows env won't work echo %comspec% or $comspec won't wokr
[18:07:11 CEST] <Atlenohen> well it does work now
[18:07:28 CEST] <Atlenohen> I guess it works for non %windowspercent% variables
[18:07:59 CEST] <Atlenohen> but that's base MINGW64, now I'll test the VC initialized one,
[18:14:09 CEST] <Atlenohen> Oh so if I specify pkconfig then I don't need to specify include and libs
[18:18:23 CEST] <Atlenohen> DHE: does this pkg_config_path need to be set under ./configure ?
[18:24:06 CEST] <furq> it's an environment variable
[18:31:25 CEST] <Atlenohen> Oh I need to use that to actually enter it, okay --pkgconfigdir
[18:38:18 CEST] <Atlenohen> furq: still won't work: https://pastebin.com/7fWqUKC3
[18:38:42 CEST] <furq> i've never used msvc so i can't help you
[18:39:28 CEST] <DHE> well x264.pc isn't a dir
[18:39:46 CEST] <DHE> normally I do "export PKG_CONFIG_PATH=/usr/local/lib/pkg-config" before running configure
[18:40:10 CEST] <furq> afaik you should still be able to link gcc dlls to an msvc host as long as you don't link any C++ libs to ffmpeg
[18:40:16 CEST] <furq> so it'll probably be easier to do that
[18:41:28 CEST] <DHE> it's not recommended, but if you specify enough options that compiling a test x264 application succeeds (eg: --extra-cflags=-I/path/to/x264 --extra-ldflags=-L/path/to/x264 substitute for what VC needs) it should work.
[18:42:06 CEST] <furq> if you're only linking one library then specifying paths manually will be simpler
[18:42:51 CEST] <furq> you also probably want to use relative paths
[18:43:16 CEST] <furq> i never quite figured out at what point msys bash transforms msys paths to actual windows ones but i do know it's pretty fragile
[18:43:34 CEST] <furq> so it's best to just use relative paths and avoid that entirely
[19:05:14 CEST] <Atlenohen> furq: oh oops, DHE: I'm not sure anymore what pkg-config were talking about, I was linking to the x264 one, not the one in the msys2 install directory
[19:05:58 CEST] <syadnom> I'm pulling my hair out trying to get ffmpeg to encode mpeg dash at all, with a goal of multiple streams.  can anyone point me to a comprehensive guide or look at what I've got right now that isn't working?
[19:11:08 CEST] <Atlenohen> DHE:okay but all the talk is about the env variable, what do you put into the --pkgconfigdir then ?
[19:13:28 CEST] <Atlenohen> for relative paths that means I make a copy of the x264/built folder over to the ffmpeg repo, make sense, okay
[19:13:50 CEST] <Atlenohen> indeed the whole linux paths inside windows is confusing
[19:17:33 CEST] <Atlenohen> DHE: then --extra-ldflags -L is not a MSVC option then, there is no -L anywhere in the C complier options, linker
[19:21:10 CEST] <Atlenohen> okay it didn't work, i'm out of ideas, I'll be back later after a big break
[19:21:39 CEST] <Atlenohen> this was the final command I tried:  CC=cl ./configure --toolchain=msvc --arch=x86_64 --pkgconfigdir=./x264/built/lib/pkgconfig --extra-cflags='-I./x264/built/include -O2 -MT -GS-' --extra-ldflags=-L./x264/built/lib --prefix=./BUILT/x64
[19:25:44 CEST] <Atlenohen> petty issues with licenses making this process 100x the hassle it could be
[19:26:04 CEST] <Atlenohen> it could have just been boundled with ffmpeg
[19:26:10 CEST] <Atlenohen> meh
[19:48:21 CEST] <cptnhmr42> How can I keep audio and video timestamps / idr alligned when transmuxing and transcoding from rtmp to mpeg-dash / hls?
[19:59:10 CEST] <cptnhmr42> If no one answers questions in here, should I ask my question on the mailing list?
[19:59:47 CEST] <Atlenohen> Okay the thing actually worked
[20:01:27 CEST] <Atlenohen> DHE: I used your suggestion to define PGK_CONFIG_PATH, and then I ommited any text from the -I and -L options, and it worked, but it is set for x86 instead of x64, I have installed both YASM and NASM in , I did not enable yasm in ffmpeg config, not sure if that has anything to do with x64 vs x86
[20:02:32 CEST] <Atlenohen> https://pastebin.com/ZTyWR59m
[20:07:01 CEST] <Atlenohen> oh, yasm stuff seems deprecated
[20:20:42 CEST] <Atlenohen> so wait, it is x86_64, it just gets reported as x86
[20:20:45 CEST] <Atlenohen> i[3-6]86*|i86pc|BePC|x86pc|x86_64|x86_32|amd64)
[20:20:45 CEST] <Atlenohen>         arch="x86"
[21:10:42 CEST] <lindylex> I am trying to concat video using this script https://pastebin.com/jjpkY7w5  I am not sure why it is turning my 4 and 2 minute in length videos in 4 second videos.
[21:10:58 CEST] <garyserj> What is this command for libx264? fmpeg -i input.mkv -c:v libx265 -b:v 4000k -an -x265-params no-info=1 output.hevc
[21:11:35 CEST] <garyserj> I tried ffmpeg -i c_bad.mp4 -vcodec libx264 -libx264-params no-info=1 -acodec copy d.mp4  and I got "Unrecognized option 'libx264-params'."
[21:12:55 CEST] <garyserj> ah, maybe x264-params
[21:13:36 CEST] <garyserj> it isn't removing all encodnig settings though. how would I do that?
[21:13:43 CEST] <lindylex> garysej: are you trying to use something like this?  -codec:v libx264
[21:14:07 CEST] <garyserj> I think -codec:v is the same as -vcodec copy so that's ok
[21:15:46 CEST] <Atlenohen> lindylex: because the scrip has trim, you don't need any trim if you want to combine full videos
[21:16:13 CEST] <lindylex> Ok thanks that was new to me.  Trimming videos this way.
[21:16:23 CEST] <garyserj> if you look here https://pastebin.com/raw/ZJHhHGnD  c_good.mp4 has no encoding settings  and no writing library listed,  whereas c_bad.mp4 has writing library listed, and encoding settings  e.g. "cabac=1 / ref=3 / deblock=1:0:0 "     I want to convert c_bad.mp4 to have no writing library and no encoding settings listed
[21:16:25 CEST] <Atlenohen> concat works differnt on different OS I believe, in windows I usually got it working with a list file, a text file that has the input filenames
[21:17:51 CEST] <Atlenohen> lindylex: that trim is probably used for a large batch, if you only have one or two videos to trim, you can do that manually with -ss and -to options,  like -ss 00:00:10  (start at 10 seconds) and -to 00:00:30 (end at 30)
[21:17:54 CEST] <lindylex> Atlnohen I use to use a text file with a list. I no longer.
[21:18:12 CEST] <lindylex> That is the way I use to do it.  -ss and -to
[21:18:14 CEST] <Atlenohen> As you wish if it works better for you
[21:18:50 CEST] <lindylex> I was just sharing.  And thanks for your input.  Trimming the way it is done in the script is new to me.
[21:19:26 CEST] <Atlenohen> You may need to customize the script to make trimming the way you need, if necessary, I never used that so I can't tell much from the first look.
[21:20:05 CEST] <garyserj> ah -to is useful, i've been using -t
[21:20:41 CEST] <lindylex> -to is my favorite.  Before that I used -t and I had to do to many calculations.
[21:20:49 CEST] <garyserj> heh, i wrote a program to do the calculations!
[21:21:04 CEST] <lindylex> What language?
[21:21:07 CEST] <Atlenohen> garyserj: the bad.mp4 may be corrupt so it may not show details, if you want to ommit details you may need to HEX edit the file, that's the only way I know right now, I don't know if encoders/muxers have the option to customize or omit the info
[21:21:35 CEST] <lindylex> I wrote a wrapper for special effects python calling ffmpeg
[21:22:40 CEST] <garyserj> lindylex: ruby
[21:23:01 CEST] <garyserj> How can I check if an mp4 is corrupt?
[21:23:16 CEST] <garyserj> and how can I fix it if it is corrupt?
[21:23:36 CEST] <garyserj> atlen: the bad one iis the one with the details
[21:23:56 CEST] <garyserj> Both play, but the bad one won't upload to youtube and the good one will
[21:24:29 CEST] <lindylex> Atlenohen : https://github.com/MotionDesignStudio/commandlinemediahelper
[21:24:48 CEST] <lindylex> sorry I meant garyserj : https://github.com/MotionDesignStudio/commandlinemediahelper
[21:24:56 CEST] <garyserj> thanks ,i'll check that
[21:25:00 CEST] <garyserj> bbiab like 3h but will check for any replies when I get back
[21:25:13 CEST] <lindylex> Send me yours please.
[21:25:39 CEST] <Atlenohen> furq, DHE: I managed to do something, but it produced .a files not .lib, they're much smaller than they suppose to be IMO as I have a lot more encoders/muxers enabled than the 5 year old versions have
[21:29:12 CEST] <garyserj> lindylex : I haven't used it in ages and there seems to be a little bug but anyhow   https://pastebin.com/raw/B0fTcJLt    you just entered the start and end time and it told you the -ss and -t
[21:29:25 CEST] <lindylex> Atlenohen I am having issues removing the trim and getting this ffmpeg command to run.
[21:29:46 CEST] <garyserj> lindylex: nothing special, like a 5min program!
[21:30:02 CEST] <lindylex> Thanks nice work.
[21:30:10 CEST] <garyserj> heh, gotta go though, bbaib 3h
[21:30:32 CEST] <garyserj> slight bug on gets.chomp but the calculation s fnie
[21:30:56 CEST] <garyserj> not seen that bug before but it says bug on line 18. should be quick to fix but i gtg.
[23:07:13 CEST] <lindylex> I am trying to remove the trim and the audio removal from the ffmpeg command without success.  I tried sever variations with no success.  https://pastebin.com/Psn7cp0H
[23:23:00 CEST] <Ariyasu> maybe try concat the source files
[23:23:11 CEST] <Ariyasu> so your working with less inputs
[23:23:23 CEST] <Ariyasu> might make it easier to work out where things are goign wrong
[23:24:18 CEST] <lindylex> The source files need to have the transition between them.  This is why I can not concat before applying the filter.
[23:56:12 CEST] <faLUCE> Hello all. JEEB: I just got a confirmation from MPEG LA :-)
[00:00:00 CEST] --- Sat Apr 13 2019


More information about the Ffmpeg-devel-irc mailing list