[Ffmpeg-devel-irc] ffmpeg.log.20161204

[burek]

[00:28:05 CET] <mantas322> everything worked out great.  Thank you for the help c_14.
[00:38:14 CET] <Threads> lets say i have a video and its 15mins long and i want to cut out the 5mins point B and be left with 5mins from point A and 5mins from point C how would i do that i want to be left with a 1 10min file with point B remved and point A-C left
[00:47:14 CET] <kyleogrg> Does ffmpeg's VP9 support GPU encoding?
[00:52:07 CET] <waster> hey all, how can i precisely cut ts file with audio only with milliseconds?
[00:52:22 CET] <waster> aand I need to preserve timestamps
[00:52:55 CET] <waster> input seeking does not have needed accuracy
[00:53:20 CET] <waster> and output seeking reset timestamps to 0 even with copyts option
[00:53:57 CET] <waster> for example
[00:55:14 CET] <waster> ffmpeg -i out2020.ts -ss 2:48:23.423633 -y -c copy -c copyts -muxdelay 0 test.ts
[00:55:46 CET] <waster> initial timestamp is considered with ss
[00:56:30 CET] <waster> but output test.ts has timestamps reset
[01:21:25 CET] <DHE> kyleogrg: ffmpeg just uses libvpx, so as soon as that gets GPU acceleration ffmpeg will as well
[01:21:53 CET] <kyleogrg> ah, so we don't know when libvpx would get it?
[01:23:30 CET] <DHE> not like we have any say in it
[01:25:22 CET] <kyleogrg> DHE: ok, so what about x265 gpu encoding?
[01:27:34 CET] <jkqxz> libvpx is a software encoder, that's very unlikely to happen.
[01:28:04 CET] <jkqxz> There are patches for VP9 encode on Intel GPUs on the libav list, though (via VAAPI).
[01:28:58 CET] <kyleogrg> ok, thanks for the info.  be back later probably
[01:29:10 CET] <DHE> well, x264 got partial GPU offload support... not that it's the most brilliant thing, but still...
[01:41:52 CET] <MrMonkey31> is it possible to set up the fade=out filter to automatically do the end of the input stream, where the duration is specified?  like when it's set with -t ?
[01:42:26 CET] <MrMonkey31> *the duration of the fade is configured, AND the whole video has a given duration
[02:25:44 CET] <TikityTik> subprocess.check_call(['xdg-open', '--', '/home/user') is giving me an exit code of 1, why is that?
[02:26:31 CET] <TikityTik> apparently I had to take out the '--' argument
[06:30:41 CET] <ferdna> what is the average of delay of you ipcamera video that you guys are getting?
[06:30:46 CET] <ferdna> i am getting about 8 seconds
[06:33:48 CET] <ferdna> also, what are your ip camera parameters?
[08:13:51 CET] <tomed> anybody know a good GUI video/audio player
[08:14:00 CET] <tomed> mplayer sucks
[08:23:57 CET] <ferdna> tomed, mpv and vlc are my favs
[08:24:04 CET] <tomed> mpv ?
[08:25:37 CET] <tomed> ferdna anyways, thank you
[08:26:01 CET] <ferdna> tomed, yes mpv... that is the name of the player
[08:26:28 CET] <tomed> i see
[08:26:30 CET] <tomed> i see 2 builds
[08:26:39 CET] <tomed> Windows builds by lachs0r 	https://mpv.srsfckn.biz
[08:26:39 CET] <tomed> Windows builds by shinchiro 	https://sourceforge.net/projects/mpv-player-windows/files
[08:26:44 CET] <tomed> is there any difference?
[08:28:19 CET] <ferdna> tomed, this one is the official one: https://mpv.io/
[08:28:43 CET] <ferdna> so yeah... any of those two links will work
[08:29:37 CET] <ferdna> tomed, i would go with lanchs0rs builds
[08:29:43 CET] <ferdna> seem cleaner
[08:29:49 CET] <tomed> i see
[08:29:56 CET] <ferdna> tomed, i dont use windows so i wouldnt know...
[08:30:00 CET] <ferdna> ive never tried them
[08:30:11 CET] <ferdna> tomed, type my nick... if you expect a fast response..
[08:30:33 CET] <tomed> ferdna i see, sorry,  same for me
[08:30:42 CET] <ferdna> tomed, yup =)
[08:31:05 CET] <tomed> ferdna mpv sucks though
[08:31:11 CET] <tomed> i have to drag a file
[08:31:38 CET] <ferdna> tomed, probably but i like it very much... it has worked wonders for me...
[08:32:36 CET] <tomed> ferdna good thing is: it has #mpv channel
[08:32:45 CET] <ferdna> tomed, yup
[08:35:38 CET] <tomed> ferdna this whole drag and drop method is annoying
[08:35:59 CET] <ferdna> tomed, i dont know what you mean... but it works fine for me
[08:36:10 CET] <ferdna> you need to associate your .avi .mpg to that program
[08:36:16 CET] <ferdna> and it will just open your files
[08:36:51 CET] <tomed> and how do i open  dvd-source
[09:13:05 CET] <ferdna> tomed, just double click the dvd files
[09:13:17 CET] <tomed> ferdna  it doesn't work taht way with dvd
[09:13:25 CET] <ferdna> yes it does
[09:14:11 CET] <tomed> ferdna it doesn't
[09:24:40 CET] <thebombzen> tomed: ferdna: mpv is primarily a CLI player
[09:24:48 CET] <tomed> thebombzen okay
[09:24:58 CET] <tomed> thebombzen but i should able to play dvd source
[09:25:55 CET] <thebombzen> run from the command line: mpv dvd://
[09:26:20 CET] <thebombzen> you can. but MPV is a command-line player. It opens a gui that does essentially nothing but display the video and allow you to seek.
[09:26:51 CET] <thebombzen> the drag-n-drop is just an approximation for a GUI. It's not really a gui player
[09:27:14 CET] <thebombzen> if you're looking for a fully-functional GUI player that's based on ffmpeg, use VLC: http://videolan.org/
[09:27:52 CET] <thebombzen> in VLC, to play a DVD you can use File -> Open Disc and it's intuitive
[09:28:02 CET] <thebombzen> sorry. Media -> Open Disc
[09:29:25 CET] <thebombzen> ferdna: that won't work
[09:30:02 CET] <thebombzen> on the dvd filesystem, videos are stored in an encrypted program stream and you cannot decrypt them without reading the DVD as a block device
[09:30:29 CET] <thebombzen> if you open a the disc as a whole in a DVD player, it can decrypt it.
[09:33:19 CET] <ferdna> thebombzen, yes it will. it will decrypt the files if they are encrypted
[09:33:41 CET] <thebombzen> no, dvdcss is not able to decrypt the individual VOBs
[09:34:17 CET] <thebombzen> either way, "double click it" is bad advice, becuase the .vob extension on windows by default opens with Windows Media Player
[09:34:36 CET] <Ella_hpost> Can anyone help me with enable drefs please?
[09:34:58 CET] <tomed> thebombzen thank you
[09:35:56 CET] <Ella_hpost> I can't convert QuickTime reference to Mxf,  how do I enable drefs?
[09:49:55 CET] <ferdna> tomed, thebombzen, http://unix.stackexchange.com/questions/265874/using-mpv-to-play-dvd-movies
[09:50:26 CET] <ferdna> this one tells you what you need to decrypt dvds:
[09:50:27 CET] <ferdna> http://www.linuxquestions.org/questions/slackware-14/unable-to-play-encrypted-dvds-properly-with-mpv-with-libdvdcss-installed-4175513902/
[09:50:57 CET] <ferdna> good night
[10:21:36 CET] <kerio> why would anyone want to play a dvd in 2016
[11:01:45 CET] <Tushar> i have having problem while installing ffmpeg
[11:11:04 CET] <durandal_170> Tushar: what kind of problem?
[11:11:45 CET] <Tushar> yasm/nasm not found or too old. Use --disable-yasm for a crippled build.
[11:11:46 CET] <Tushar> If you think configure made a mistake, make sure you are using the latest
[11:11:46 CET] <Tushar> version from Git.  If the latest version fails, report the problem to the
[11:11:46 CET] <Tushar> ffmpeg-user at ffmpeg.org mailing list or IRC #ffmpeg on irc.freenode.net.
[11:11:46 CET] <Tushar> Include the log file "config.log" produced by configure as this will help
[11:11:46 CET] <Tushar> solve the problem.
[11:11:57 CET] <Tushar>               while installing
[11:25:20 CET] <durandal_170> Tushar: so install or update your Yasm or nasm
[12:25:08 CET] <beauty> when to use av_free_packet and av_packe_unref
[12:25:29 CET] <beauty> what's the different between the two function?
[12:33:17 CET] <jkqxz> They both decrease the reference count on the buffer referred to by the packet, and free it if that reaches zero.  Then, av_packet_unref() clears the AVPacket structure to a blank one (so it can be reused with av_packet_ref() or av_new_packet(), say), while av_packet_free() actually frees the memory and so it can't be reused.
[13:30:15 CET] <beauty`> thanks
[15:01:42 CET] <waster> hello, hwo can I cut ts file with ms accuracy?
[15:01:56 CET] <waster> I need to preserve timestamps
[15:02:31 CET] <waster> output seeking doesn not work in this case and even with copyts resets timestamps
[15:04:00 CET] <DHE> well, if you're okay with trial and error, mpgets is relatively friendly to 'dd' abuse. you might not get millisecond resolution but maybe close enough for your needs
[15:17:04 CET] <waster> the problem is that I get needed accuracy with output seekingm bu I need timestamps preserved
[15:19:32 CET] <waster> and I don't understand when timestamps are reset even with copyts option in case of output seeking
[16:09:56 CET] <waster> it seems that the only way to preserv timestamps in this case is to use -itsoffset and -output_ts_offset
[16:10:38 CET] <waster> e.g. ffmpeg -itsoffset -10101.4069 -i test.ts -ss 2.78 -y -c copy -copyts -muxdelay 0 -output_ts_offset 10104.1869  test1.ts
[16:38:31 CET] <jaeckyl> How can I specific the channel layout explictly on the input?  ffmpeg is guessing incorrectly
[16:38:31 CET] <jaeckyl> Guessed Channel Layout for  Input Stream #0.1 : 5.1
[16:47:38 CET] <durandal_170> jaeckyl: what format?
[16:48:54 CET] <jaeckyl> pcm_s16le
[16:49:46 CET] <jaeckyl> Stream #0:1: Audio: pcm_s16le, 48000 Hz, 6 channels, s16, 4608 kb/s
[16:51:25 CET] <durandal_170> that's codec
[16:53:04 CET] <jaeckyl> yes, thats the input, capturing to an Mp4 container with aac using:
[16:53:06 CET] <jaeckyl> codec:a aac -ac 6 -ar 48000 -ab 384k
[16:53:38 CET] <jaeckyl> sort of a newb, so please be patient ;)
[16:54:54 CET] <jaeckyl> the problem is that the center channel is getting mapped to LFE
[16:58:53 CET] <durandal_170> jaeckyl: you will need to use channelmap filter, see fine documentation
[17:07:24 CET] <jaeckyl> yea, I tried -filter 'channelmap=1|2|0|5|3|4:5.1' --> but unsure how to select the audio stream
[17:11:12 CET] <jaeckyl> error: Cannot connect audio filter to non audio input
[17:11:43 CET] <jaeckyl> So I tried: -filter "channelmap=0:1:1|0:1:2|0:1:0|0:1:5|0:1:3|0:1:4:5.1"
[17:12:19 CET] <jaeckyl> which errored out with: No option name near '1|0:1:2|0:1:0|0:1:5|0:1:3|0:1:4:5.1'
[17:12:45 CET] <jaeckyl> So I know the problem is my understanding of filters and stream selection
[17:41:43 CET] <durandal_170> jaeckyl: -af channnelmap=...
[17:42:14 CET] <jaeckyl> thanks for the help, the solution is: -af "channelmap=1|2|0|5|3|4:5.1" as shown in the documentation
[17:42:37 CET] <jaeckyl> another RTFML moment
[17:42:47 CET] <jaeckyl> thanks for the pointer :)
[17:44:39 CET] <durandal_170> you select audio stream with [1:a] for example and put that before channelmap
[19:07:19 CET] <MrMonkey31> heiii
[19:10:12 CET] <slimfast> hi.  i built mpv from github and playing a youtube url gives me
[19:10:14 CET] <slimfast> [ffmpeg] tls: A TLS packet with unexpected length was received.
[20:04:54 CET] <Ella_hpost> Hi,  can anyone please help me converting qt ref to Mxf?  I don't know how to enable drefs
[20:21:34 CET] <MrMonkey311> compared to xvid and such, how forgiving is h264 on re-encodes?
[20:53:11 CET] Action: MrMonkey311 rages
[21:14:43 CET] <slimfast> hooray mpv works with --ytdl
[21:14:48 CET] <slimfast> cheers
[21:37:40 CET] <superguest> Hello.
[21:39:13 CET] <superguest> the ffmpeg cmdline in the first answer give me a countdown timer, http://video.stackexchange.com/questions/15556/ffmpeg-video-with-timer
[21:41:13 CET] <superguest> Is there a way to control when that countdown timer (via the 'drawtext' filter) appears, i.e., control when in the video the timecode gets burnt-in
[21:42:13 CET] <superguest> For example, I want the countdown timer to appear in the last 10 secs of the video.
[21:43:09 CET] <superguest> 10, 9, 8, .... 0
[21:44:52 CET] <superguest> as opposed to "(video.length - 0), (video.length - 1), (video.length - 2) ... 0"
[21:45:00 CET] <superguest> I think you get the idea.
[22:32:11 CET] <daemon> hey all I am trying to concat a few videos but keep hitting:
[22:32:11 CET] <daemon> [mp4 @ 0x80b441000] Could not find tag for codec vp9 in stream #0, codec not currently supported in container
[22:32:21 CET] <daemon> but I have VPx support compiled in
[22:32:33 CET] <daemon> am I missing something I am using: ffmpeg -f concat -safe 0  -i /tmp/merge.list  -c copy S01E02.mp4
[22:32:59 CET] <daemon> http://paste.ee/r/XDsdn
[22:33:02 CET] <daemon> is what console says
[22:33:51 CET] <jkqxz> Yes; you are missing that VP9 cannot be muxed into mp4 (as the error message helpfully informs you, in fact).  Use a different container.
[22:34:13 CET] <daemon> can I get it to cross encode to mp4
[22:34:36 CET] <daemon> did not really want to have multiple different file types
[22:35:57 CET] <jkqxz> Drop the "-c copy" and give it some encoding options instead and it will transcode; probably you want to use libx264.
[22:36:31 CET] <daemon> sounds sane, thank you
[22:36:58 CET] <daemon> the last concat vio was
[22:36:58 CET] <daemon>     Stream #0:0(und): Video: h264 (Main) (avc1 / 0x31637661), yuv420p, 292x234, 242 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
[22:37:02 CET] <daemon>     Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 125 kb/s
[22:37:11 CET] <daemon> so I should be able to use that
[22:40:41 CET] <kyleogrg> How do I specify two input files (one only video and one only audio) to FFplay?
[22:43:08 CET] <daemon> hmm
[22:43:20 CET] <daemon> jkqxz, is there something special about vp9, I am concatting 3 files to one
[22:43:25 CET] <daemon> it works for the first 10 minutes
[22:43:40 CET] <daemon> then I get a STREAM of
[22:43:41 CET] <daemon> [vp9 @ 0x80b468f00] Invalid frame marker
[22:43:42 CET] <daemon> Error while decoding stream #0:0: Invalid data found when processing input
[22:43:45 CET] <daemon> literally thousands of them
[22:43:49 CET] <daemon> just a shitty codec?
[22:43:55 CET] <daemon> because the input video plays fine
[22:44:28 CET] <iive> depends how you concat them
[22:44:35 CET] <daemon> $ ffmpeg -f concat -safe 0  -i /tmp/merge.list -vcodec h264 -acodec libfaac S01E02.mp4
[22:44:42 CET] <daemon> with a list of files in /tmp/merge.list
[22:44:48 CET] <daemon> it looks like the first video is being transcoded fine
[22:44:54 CET] <daemon> the header error flod starts with video 2
[22:44:57 CET] <iive> well, you may have to use the filter
[22:45:21 CET] <iive> i think that the above method works for mpeg files
[22:46:12 CET] <daemon> worked fine for the first set of videos
[22:46:16 CET] <daemon> being a pain in the ass with the second set
[22:46:32 CET] <daemon> sod it lets just let it run with the errors see what happens at the end
[22:46:54 CET] <iive> there are 3 concat methods
[22:47:37 CET] <iive> 1. is simple file concat, if you append files to one onother you'd get the same result. works for mpeg-ts/es etc..
[22:47:52 CET] <daemon> thats what I used with the first set
[22:47:56 CET] <daemon> <daemon>     Stream #0:0(und): Video: h264 (Main) (avc1 / 0x31637661), yuv420p, 292x234, 242 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
[22:47:57 CET] <daemon> <daemon>     Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 125 kb/s
[22:47:58 CET] <daemon> worked a treat
[22:48:13 CET] <daemon> but this second set is using vp9 for video
[22:48:53 CET] <iive> 2. a packet level concat, you need to have same extradata for all files in order for this one to work, because the decoders neer know you started a second file.
[22:49:16 CET] <iive> basically it works if you split a file on smaller files then trying to merge them
[22:49:36 CET] <daemon> it appears if I seperately transcode the file parts as in one by one, into h264|aac then they will concat together, might be a long way around though
[22:49:54 CET] <iive> 3. filter one. you decode the first file pass images to the filter chain, when the file ends second one starts decoding after the first one.
[22:50:22 CET] <iive> it works with all types because it uses separate decoders for each file and you merge the decoded pictures.
[22:50:33 CET] <daemon> gotcha
[22:50:45 CET] <daemon> ffmpeg is incredibly flexible
[22:50:49 CET] <daemon> quite an impressive program
[22:51:13 CET] <iive> yeh, but it is also quite complex.
[22:52:08 CET] <superguest> Hey guys. Can I get help too?
[22:52:25 CET] <daemon> superguest, can try, whats up :)
[22:52:55 CET] <superguest> the ffmpeg cmdline in the first answer give me a countdown timer, http://video.stackexchange.com/questions/15556/ffmpeg-video-with-timer
[22:53:46 CET] <superguest> Is there a way to control when that countdown timer (via the 'drawtext' filter) appears, i.e., control the point (in the video) at which the timecode gets burnt-in
[22:53:57 CET] <superguest> For example, I want the countdown timer to appear in the last 10 secs of the video.
[22:53:57 CET] <superguest>  10, 9, 8, .... 0
[22:54:09 CET] <superguest> as opposed to "(video.length - 0), (video.length - 1), (video.length - 2) ... 0"
[22:54:24 CET] <superguest> I think you get the idea.
[22:55:53 CET] <daemon> not a bloody clue did not think ffmpeg could even do that, would assume you would have to insert frames into the video its self
[22:56:03 CET] <daemon> I am just trying to recover some old videos from file parts :P
[23:15:57 CET] <superguest> anyone wants to chime in? ^^
[23:21:36 CET] <durandal_170> superguest: enable option within filter?
[23:23:29 CET] <superguest> durandal_170, I don't quite follow what you mean.
[23:23:34 CET] <superguest> I am sorry.
[23:24:29 CET] <durandal_170> read documentation
[23:25:00 CET] <durandal_170> you need to know how long is video
[23:25:43 CET] <superguest> durandal_170, I know how to do that, as the SoF answer provides me with a solution using ffprobe.
[23:26:09 CET] <superguest> I mean *stackexchange*
[23:27:27 CET] <superguest> It'd be helpful if you can tell me what I should look for in the doc.
[23:27:56 CET] <superguest> as far as the options and associated parameters for drawtext, I am familiar with.
[23:29:10 CET] <superguest> I want the count down to appear in the last 10 secs of the video, as opposed to appearing through the entire video.
[23:30:37 CET] <superguest> s/through/throughout/
[00:00:00 CET] --- Mon Dec  5 2016