[Ffmpeg-devel-irc] ffmpeg.log.20191204

[burek]

[00:15:35 CET] <rafael2k> cehoyos: errr, because AOT 42 is USAC?
[00:15:57 CET] <rafael2k> cehoyos: and in the description of the stream says "xHE-AAC"
[00:16:25 CET] <rafael2k> cehoyos: and I see DTV here in Brazil for ages, and the mux is LATM?
[00:16:30 CET] <rafael2k> :P
[00:21:02 CET] <rafael2k> ffmpeg does have a latm parser.
[00:27:50 CET] <cehoyos> Yes, in the FFmpeg-sense, it has a "latm parser", but it does not contain the necessary parser ("bitstream filter") to convert latm into aac
[00:28:12 CET] <cehoyos> (Current bitstream filters do not allow to change codec iirc)
[00:29:34 CET] <rafael2k> right, I'll investigate a bit more, but here is the xHE-AAC sample, anyway:
[00:29:44 CET] <rafael2k> wget http://stream.radioh.no:443/rh.x16
[00:36:07 CET] <cehoyos> Yes, you posted it already (thank you!)
[00:38:31 CET] <rafael2k> I was just confirming it is indeed a USAC sample, after you questioned if it really was it... tks.
[00:41:45 CET] <cehoyos> How were you able to confirm?
[00:41:55 CET] <cehoyos> (Is there software that allows decoding?)
[00:48:32 CET] <JEEB> cehoyos: fdk-aac's 2018 v2.x release apparently does xhe-aac?
[00:50:31 CET] <cehoyos> Indeed, I have libfdk 1,6 installed
[00:53:25 CET] <JEEB> libMpegTPDec/src/tpdec_asc.cpp apparently mentions it a lot?
[00:54:24 CET] <JEEB> git grep -i "usac" gave possibly a bit better results :)
[00:54:43 CET] Action: JEEB goes hit the sack
[01:03:26 CET] <rafael2k> JEEB: ; ))
[01:03:47 CET] <rafael2k> cehoyos: ffmpeg latm aac decoder can read the AOT...
[01:04:17 CET] <cehoyos> It fails here for the sample
[01:04:35 CET] <rafael2k> it fails as it does not implement USAC
[01:04:52 CET] <cehoyos> Yes.
[01:04:58 CET] <cehoyos> What were you trying to say before?
[01:06:54 CET] <rafael2k> If I could decode it I was not asking here
[01:07:33 CET] <rafael2k> I have fdk-aac v2 here
[01:07:49 CET] <rafael2k> yes, it should support xHE-AAC... the library
[01:09:19 CET] <rafael2k> <cehoyos> (Is there software that allows decoding?)
[01:10:04 CET] <rafael2k> cehoyos: Yes, I use Fraunhofer Multimedia Player, but it is not free software.
[01:10:35 CET] <cehoyos> The library may support it but as I explained before, FFmpeg does not contain the necessary bitstream filter ("parser") afaict.
[01:10:50 CET] <rafael2k> I understand now, tks cehoyos, and sorry to bother
[01:10:53 CET] <cehoyos> Unless the library accepts latm which I cannot test because I don't have v2 here.
[01:11:17 CET] <cehoyos> Where can I find the Fraunhofer Multimedia Player?
[01:12:24 CET] <rafael2k> it is not free...
[01:12:43 CET] <cehoyos> That means I cannot find it?
[01:15:02 CET] <rafael2k> I'm not stopping you from nothing
[12:43:36 CET] <Ducky^> ffmpeg -y -i WAVs/DDC-Logo_f_DCDM_1-0.wav -i WAVs/DDC-Logo_f_DCDM_2-0.wav -i WAVs/DDC-Logo_f_DCDM_3-0.wav -i WAVs/DDC-Logo_f_DCDM_4-0.wav -i WAVs/DDC-Logo_f_DCDM_5-0.wav -i WAVs/DDC-Logo_f_DCDM_6-0.wav -filter_complex
[12:43:37 CET] <Ducky^> "[0:a]apad=whole_dur=3600[merge];[1:a]apad=whole_dur=3600[merge];[2:a]apad=whole_dur=3600[merge];[3:a]apad=whole_dur=3600[merge];[4:a]apad=whole_dur=3600[merge];[5:a]apad=whole_dur=3600[merge];[merge]amerge=inputs=6[out]" -map "[out]" -c:a pcm_s16le -ar 48000 -rf64 auto -t 3600 5.1_Test.wav
[12:43:47 CET] <Ducky^> with this command I get the error Cannot find a matching stream for unlabeled input pad 1 on filter Parsed_amerge_6
[12:44:13 CET] <Ducky^> does anyone know where the command is tripping up?
[12:45:03 CET] <durandal_1707> Ducky^: because you use output pad merge several times
[12:45:35 CET] <durandal_1707> [a][b][c]..amerge[out]
[12:45:41 CET] <durandal_1707> is proper way
[12:45:51 CET] <Ducky^> ahh, I see
[12:45:55 CET] <Ducky^> thanks durandal_1707 I'll try it
[12:48:15 CET] <Ducky^> it worked!
[12:57:43 CET] <Ducky^> simplified it a little by only having apad instead of the whole_dur part
[16:18:26 CET] <PeRy_SoY> Hello! I have tvheadend server and ffmpeg to stream my local videos and works really nice but i cant forward or rewind videos, my ffmpeg: <    pipe://ffmpeg -loglevel fatal -i /home/pery/Escritorio/WindowsShare/it2.mkv -g 250 -vcodec copy -acodec copy -metadata service_provider="provider" -metadata service_name=" Name" -f mpegts -tune zerolatency  pipe:1    > any suggest? thanks in advance and sorry for my english !
[16:18:59 CET] <pink_mist> you're using pipes
[16:19:15 CET] <pink_mist> pipes don't support rewinding
[16:19:42 CET] <PeRy_SoY> oh! :(
[16:19:46 CET] <PeRy_SoY> thanks :)
[16:22:01 CET] <ButtDog> Is it possible to obtain video duration, bitrate, etc in the browser front end, from a file input without actually uploading the video?
[16:45:53 CET] <another> you could write something in js that does that
[17:09:57 CET] <ButtDog> another: I found something actually. Looking at implementing it: https://mediaarea.net/en/MediaInfo/Download/JavaScript
[17:10:03 CET] <ButtDog> Not sure if there's a more simple way
[18:05:47 CET] <capin> does ffmpeg support providing an input file for peformming a series of `-t` `-ss` operations on a media file?
[18:08:27 CET] <kepstin> capin: the "concat" demuxer: https://ffmpeg.org/ffmpeg-formats.html#concat takes a playlist file as input, which can specify "inpoint" and "outpoint", but note that as this is the demuxer level, it can only start at keyframes (no accurate seek)
[18:11:32 CET] <capin> kepstin: was looking for something like this, https://gist.github.com/georgechalhoub/e9c1c50507f651c8af90c5f40e8376c7#file-ffmpegtrim-py
[18:13:02 CET] <kepstin> not sure what your final goal is, but it seems you already have a working solution?
[18:14:32 CET] <kepstin> if you plan to concatenate all those pieces after splitting them, then you could certainly do it all in one ffmpeg command (build the command line with a script)
[18:14:33 CET] <capin> trying to edit an input file using the `-ss` and `t` flags to remove the segments of input file i'd like to keep and then later produce a single file from the segments
[18:15:37 CET] <capin> and was wondering if ffmpeg provides a option for supplying an input file to the `-ss` and `t` flags, so i don't end up with some super gnarly command line command
[18:16:09 CET] <kepstin> it doesn't really make sense, since you have to specify the input file multiple times anyways, and those options are separate per input
[18:16:39 CET] <capin> it'd be a lot easier to edit an input file rather than having to navigate the cmd line to edit the `-ss` and `-t` flags for where i want trim the input file
[18:16:45 CET] <kepstin> the concat demuxer that I already told you about would work, but it would have the same issue as your current script, which is that your seek points are rounded to the nearest keyframe
[18:18:07 CET] <kepstin> I'd actually suggest using a -filter_complex_script option, use a "movie" source filter for each input segment, and then the concat filter to combine them.
[18:18:25 CET] <kepstin> but that would still be sufficiently complicated that it would be best to generate with a script
[18:20:01 CET] <capin> are there any GUIs / frontends for ffmpeg that could provide such editing that i seek, (no pun there)
[18:22:16 CET] <kepstin> you probably should just go straight to using a proper NLE. Maybe kdenlive or something?
[18:30:58 CET] <capin> kdenlive = total PITA on macOS just to get setup
[18:31:18 CET] <capin> i'll just end up cobbling together a py or shell script
[18:31:39 CET] <kepstin> well, on mac os there's lots of well known native video editing applications...
[18:31:46 CET] <capin> and i use shotcut for NLE
[18:32:03 CET] <kepstin> like you could probably just straight up use imovie
[18:35:01 CET] <kepstin> the ffmpeg cli is a multipurpose tool that can do many things, but it's not necessarily the easiest or best tool for any specific job. It can certainly do this, but writing a script to generate the filter chain and cli options would be a good idea.
[18:38:21 CET] <capin> yeah the issue with using apps such as `imovie` or most other NLA editors is they reencode the input stream after editing, and i'm specifically trying to avoid reencoding the input stream thus why i'm relying on using ffmpeg
[18:38:47 CET] <kepstin> the issue with not re-encoding is that your seek points will not be exactly where you want them
[18:38:53 CET] <kepstin> since it can only start at keyframes
[18:39:12 CET] <kepstin> (how far off they will be depends on the settings the source file was encoded with)
[18:43:34 CET] <kepstin> if you're sure you don't want to re-encode, my initial suggestion of using the concat demuxer's playlist format is probably the simplest way forwards
[20:54:48 CET] <void09> any flags i can pass to ffmpeg so that it writes proper mkv metadata readable by mediainfo ? for example i am getting no audio info from mediainfo after writing an mkv with ffmpeg
[21:02:09 CET] <pbd> Is anyone able to stream live webm with ffmpeg 4.1 and 4.2? It looks like the EBML header is kinda broken.
[21:03:24 CET] <JEEB> void09: you can forcibly write that metadata field but the general problem is that mediainfo does not want to calculate that value by itself :P
[21:03:51 CET] <void09> reading a value is simpler than calculating it :P
[21:04:07 CET] <JEEB> sure, but mediainfo is already parsing the matroska file
[21:05:00 CET] <JEEB> pbd: there have been improvements to the matroska muxer recently, so I'd check with master. if master is still broken, report the issue on trac.ffmpeg.org , and if not then there's just some fixes not applied onto the "stable" branches
[21:05:44 CET] <JEEB> void09: the problem with metadata fields that are metadata is that they get copied by many tools without a further thought even if stuff gets re-encoded :P
[21:06:05 CET] <JEEB> but hey, you can get your "this is totally 32kbps" or "this is totally 1280kbps"
[21:06:32 CET] <void09> ok, any flag for ffmpeg to automagically write everything ?
[21:08:52 CET] <JEEB> there is no automatic thing since they are not standardized?
[21:09:17 CET] <pbd> JEEB: yes, master is still broken.
[21:09:27 CET] <JEEB> pbd: cool. feel free to report it then
[21:09:37 CET] <JEEB> and if possible, with how to replicate it
[21:09:48 CET] <JEEB> example command line with ffmpeg.c etc, if possible
[21:10:13 CET] <pbd> JEEB: yes, I had already planned to open a ticket but it was very complicated to get into the trac.
[21:10:36 CET] <pbd> First, it says that I am a spammer based on shitty blacklists, then I was only able to get the verification mail yesterday.
[21:10:40 CET] <pbd> That was quite annoying.
[21:10:42 CET] <JEEB> ouch
[21:10:57 CET] <pbd> And the worst was to complete a google captcha.
[21:11:16 CET] <JEEB> but I must guess the person who set that up had the best intentions in mind
[21:11:24 CET] <JEEB> just that it unfortunately hit you
[21:11:46 CET] <pbd> yes usually they have good intentions but people should be aware of the consequences of creating a relliance on third parties.
[21:12:18 CET] <JEEB> funny enough I do remember configuring some systems myself back a few years ago
[21:12:36 CET] <JEEB> and then there was some standard captcha thing
[21:12:51 CET] <JEEB> and yea, that let a lot of stuff through :/
[21:13:02 CET] <JEEB> I think the best stopper were custom questions
[21:13:26 CET] <JEEB> which someone having the in-theme knowledge would be able to complete
[21:13:40 CET] <pbd> When it's on the trac, it's not too bad because you are told what the problem is, but when it's used on mail servers that's awful.
[21:13:52 CET] <pbd> It allows some shady organizations to blackmail ISP.
[21:14:12 CET] <pbd> So whenever I see this I get frustrated.
[21:14:49 CET] <JEEB> but yea, I think I was able to come up with like two or three questions when I was maintaining the thing I used to maintain
[21:15:08 CET] <JEEB> and some spam did still come through, but at least the numbers went a bit down
[21:15:33 CET] <JEEB> so I can see why going for the google recaptcha seemed like a GoodIdea
[21:15:53 CET] <JEEB> anyways, not even sure who maintains the trac at this point :/
[21:35:44 CET] <pbd> JEEB: it would be best to have some bayes, because if the spam still comes through it probably mean that it's not fully automated.
[21:54:12 CET] <JEEB> &30
[22:57:18 CET] <ncouloute> I'm trying to concatenate some video files from a Drone together and I'm getting this error message: h264_mp4toannexb filter failed to receive output packet on the stitch of Clip12+Clip13. Is there a way around this? Apparently it used to work a long time ago. As I was able to concat with a old version of ffmpeg from 2013.
[00:00:00 CET] --- Thu Dec  5 2019