[Ffmpeg-devel-irc] ffmpeg.log.20190911

[burek]

[00:00:01 CEST] <Diag> I had put the prefix in
[00:00:22 CEST] <BtbN> "make install"'ing ffmpeg usually causes a mess because it clashes with distro librariess
[00:00:28 CEST] <Diag> ah
[00:00:48 CEST] <Diag> just for reference im doing this on the ps3~
[00:02:00 CEST] <qmr> how can I process only a small section of a clip?  like to see results before running the whole thing
[00:02:13 CEST] <qmr> just got hit on my bike, trying to get this woman's face from mirror
[00:02:59 CEST] <Diag> -t should set the encode length
[00:03:07 CEST] <qmr> -t <seconds> ?
[00:03:14 CEST] <Diag> sec ill getcha the page
[00:04:11 CEST] <Diag> qmr: https://trac.ffmpeg.org/wiki/Seeking
[00:04:40 CEST] <Diag> make sure you set the time *BEFORE* the input *AND* after it
[00:05:36 CEST] <Diag> err sorry
[00:05:45 CEST] <Diag> Seek before, use -t after
[00:05:57 CEST] <qmr> should I be playing with gamma, or something else?   gamma=3 brightened her face, but it's ... too bright.  like it needs contrast adjusted too
[00:06:34 CEST] <Diag> if you just need a few frames out of a video and the video isnt a crazy size, id be happy to cut some stuff out of it for you with actual editing software
[00:07:34 CEST] <Diag> youre sorta doing it in a really roundabout way
[00:08:39 CEST] <qmr> well I tried in imovie, it didn't go high enough though.  actually just got a good face shot from clip where she hit me, idk why I didn't look there first
[00:10:41 CEST] <Diag> oh jesus its building
[00:11:41 CEST] <Diag> BtbN: http://tyronesbeefarm.com/images/201970eb118f-112d-4929-bba5-7a10a4807de4.png XD
[00:12:14 CEST] <Diag> its running off a flashdrive. im comitting so many sins right now
[01:15:36 CEST] <chinknot> How do I solved this error? I happened today for the first time: "Too many packets buffered for output stream 0:1."
[01:15:40 CEST] <chinknot> solve*
[01:15:49 CEST] <chinknot> ah can't type tonight
[01:29:24 CEST] <Diag> still compiling
[01:29:40 CEST] <Diag> its not very fast when it has to keep jumping back and forth to the flashdrive X_X
[01:29:58 CEST] <Diag> just passing through libav hevc now
[01:48:32 CEST] <quint> Is it possible to do 2-pass encoding while resizing a video?
[01:55:23 CEST] <Diag> would you not just put the resize filter before the encode options
[01:58:39 CEST] <quint> Diag: I actually just realized one of my options was wrong.
[01:58:42 CEST] <quint> I've got it working
[01:58:43 CEST] <Diag> :V
[01:58:46 CEST] <Diag> far out
[01:59:32 CEST] <quint> The first pass worked, but the second didn't.. Sure enough it was the audio
[02:00:18 CEST] <quint> All g.
[02:01:55 CEST] <Diag> nice
[02:52:25 CEST] <Diag> the build failed
[06:25:31 CEST] <markfilipak> Seeking function that will alternate pixels of field x.top with pixels of field x+1.top (i.e., blend two fields). I've searched but not found joy. Any way using ffmpeg?
[08:19:10 CEST] <markfilipak> Troubleshooting... Am I actually on #ffmpeg?
[08:25:38 CEST] <markfilipak> If HexChat is designed to work with IRC, then why doesn't HexChat help, like asking for me to make a password and then stuffing it into the right places? Instead, it leaves it up to me to figure out.
[08:26:54 CEST] <markfilipak> It's not a mountain to climb, it's a mountain with an initial cliff.
[08:28:29 CEST] <snooky> moin
[08:29:16 CEST] <markfilipak> Sorry. I'm a little angry about the days of my life spent trying to discover why HexChat doesn't work as advertised ...and it turns out to be the most basic thing that should be documented but isn't.
[08:30:12 CEST] <markfilipak> Oops. Sorry. I'm not on #hexchat. Leaving...
[09:28:30 CEST] <markfilipak> Seeking function that will alternate pixels of field x.top with pixels of field x+1.top (i.e., blend two fields). I've searched but not found joy. Any way using ffmpeg?
[09:30:50 CEST] <markfilipak> In other words, if the pixels of field x.top are abcd... and the pixels of field x+1.top are mnop... the resulting field would be amcodp...
[09:31:23 CEST] <markfilipak> Make that amcoeq...
[09:39:23 CEST] <Radiator> Is anyone aware if the swscale function is broken ?
[10:08:48 CEST] <Radiator> Does anyone knows why sws_scale gives me wrong output from a YUVJ420P to RGB24 ? I look like a smurf after the conversion...
[10:09:03 CEST] <Radiator> I typically follozed this example : https://gist.github.com/nakaly/11eb992ebd134ee08b75e4c67afb5703
[10:26:51 CEST] <JEEB> Radiator: if you are using ffmpeg for decoding i recommend trying out the libavfilter scale filter
[10:27:01 CEST] <JEEB> takes avframes
[10:27:38 CEST] <Radiator> Ok, from what I've read, the sws_scale function isn't really reliable ?
[10:27:54 CEST] <JEEB> so less chances of getting it wrong
[10:28:06 CEST] <JEEB> Radiator: scale filter uses swscale in the background
[10:28:23 CEST] <Radiator> Oh ok, I'll give it a look
[10:28:39 CEST] <JEEB> but you don't have to do the swscale stuff yourself in other words
[10:28:54 CEST] <JEEB> and it makes adding stuff like deinterlacing simplet
[10:29:12 CEST] <JEEB> since you csn just plug filters into your filter chain
[10:29:24 CEST] <Radiator> Well, if there's functions that ease my job let's use them
[10:31:29 CEST] <JEEB> and you can also utilize zimg library through the zscale filter which supports only the most common pixel formats but generally does stuff like gamma correct scaling
[10:31:45 CEST] <JEEB> if you built ffmpeg with that library of course
[10:31:57 CEST] <JEEB> but the scale filter is what you can start with :)
[10:32:50 CEST] <Radiator> I don't think I'll have time to implement the zimg usage but I will definitely look at it
[16:04:32 CEST] <chinknot> Applying a fade out filter to an audio track causes an "Too many packets buffered for output stream 0:1" error and if I try to use this source in a loop with a still image and use the -max_muxing_queue_size workaround the output file completely ignores the image input. I tried multiple times and this happens only when I use an audio filter (fade
[16:04:32 CEST] <chinknot> out in this case).
[16:06:31 CEST] <chinknot> Applying the filter itself doesn't output errors btw. I get that error only if I try to use the faded audio track as source again
[16:43:14 CEST] <lyncher> hi. I'm using the filter showinfo to display the input PTS from a file recorded from a DVB broadcast
[16:43:41 CEST] <lyncher> the problem that I've is that when the PTS is reset, the showinfo output keeps showing an incremental PTS
[16:44:01 CEST] <lyncher> not related with the input PTS (I have the -copyts flag)
[16:44:19 CEST] <lyncher> the input is being filtered with a filtercomplex
[16:45:21 CEST] <lyncher> is there any way to get the input PTS after a filtecomplex+showinfo?
[18:24:02 CEST] <chinknot> after some testing I found out I don't need the -max_muxing_queue_size flag at all if I remember to add the -vn flag when converting audio files. That alone solved the packets issue and fixed the image input being ignored in the final remux
[19:49:54 CEST] <ncouloute> When I use the concat demuxer it will change the duration of my original clips. Is there any way to prevent that or get the new start and stop times? (using ffmpeg.c) I know there is a inpoint outpoint and duration but it seems add frames before and after due to non-intra frame codec. (h264) Maybe I can cut the video at the nearest key frame?
[21:57:29 CEST] <Sirisian|Work> When compiling on windows how do I generate ffmpeg.dll that's used by some programs? I thought --enable-shared did that, but I guess it only builds the other dlls.
[22:03:44 CEST] <Sirisian|Work> oh nvm I found a tutorial in the docs for another app for doing it.
[22:11:40 CEST] <puchu> hi
[22:12:20 CEST] <puchu> how can i copy the audio+video of the second track of an dvd to an file?
[22:26:29 CEST] <cehoyos> puchu: What did you try? Use the map option
[22:28:43 CEST] <kepstin> puchu: note that ffmpeg doesn't support reading dvd structures directly. depending on how the dvd is laid out you might be able to find the right vob files, concatenate them, and use them as ffmpeg input - but that won't always work
[22:29:29 CEST] <kepstin> i sometimes use `dvdbackup` to pull specific titles / chapter ranges / angles out of a dvd to use with ffmpeg
[22:34:51 CEST] <AlexVestin> is there a way to tell the decoder that the frame_num will reset after an IDR frame? I'm having issues using the hwenc that raspberry pi uses
[22:36:05 CEST] <KodiakIT[m]> Is there a particular opt I should be looking for for stripping out any id3v2_priv.<> tags? (and/or excluding them from `ffprobe -show_format -show_streams`?)
[22:55:23 CEST] <KodiakIT[m]> Also, on that note, what's the proper syntax for compact json? (or passing any other opts to any other format specified by `-of` for that matter?)
[23:09:27 CEST] <puchu> cehoyos which map option - some examples?
[23:09:32 CEST] <puchu> kepstin: :(
[23:09:58 CEST] <puchu> cehoyos kepstin: i only have an iso file of a dvd
[23:10:21 CEST] <puchu> mpv plays it nicly and i'd like to extract an track
[23:10:41 CEST] <cehoyos> ffmpeg cannot deal with iso files
[23:11:25 CEST] <cehoyos> mplayer -dumpstream should produce an mpeg stream that ffmpeg can read
[23:11:26 CEST] <KodiakIT[m]> puchu, you can use dvdbackup with an iso file too. Handbrake can also handle (some of) them
[00:00:00 CEST] --- Thu Sep 12 2019