[Ffmpeg-devel-irc] ffmpeg.log.20160924

[burek]

[00:01:14 CEST] <garth> only processed the first 20 seconds (needed a quick run to test the image only), got Avg QP:20.83
[00:03:07 CEST] <garth> iive: do the loss of quality looks  normal to you (just making sure before I try a full encode with the final options I plan on using)?
[00:05:13 CEST] <iive> you should not be able to see quality loss at qp=20
[00:06:25 CEST] <garth> so you feel like something else is wrong?
[00:09:11 CEST] <iive> no idea
[00:19:43 CEST] <ChefBrad> could someone help me here: ffmpeg -i in.mp4 -framerate 60 -pattern_type glob -i "nuke.png" -filter_complex "[1:v][0:v]scale2ref=iw:ih[ovr][base];  [ovr]colorchannelmixer=aa=1.0[ovrl]; [base][ovrl]overlay[v]" -map [v] out.mp4
[00:19:57 CEST] <ChefBrad> I have complete loss of audio from the out.mp4 file
[00:25:08 CEST] <c_14> because you're not mapping it
[00:25:10 CEST] <c_14> add -map 0:a
[00:25:26 CEST] <c_14> As soon as you explicitly map something, ffmpeg discards all the implicit maps
[00:25:37 CEST] <ChefBrad> Where should I stick -map at?
[00:26:00 CEST] <c_14> somewhere before out.mp4 and after nuke.png
[00:26:11 CEST] <ChefBrad> cool beans brother I'll give this a go
[00:26:13 CEST] <c_14> same general location as your other map
[00:26:41 CEST] <ChefBrad> I already have -map [v] though is it oke do another -map ?
[00:27:04 CEST] <c_14> you can have as many maps as you can fit on your command line
[00:27:08 CEST] <c_14> (and output container)
[00:32:46 CEST] <ChefBrad> I'll grab a new file and see how she works
[00:44:09 CEST] <ChefBrad> c_14: I wanna have your children. Thank you so much
[06:00:44 CEST] <Spring> odd. I go to encode a video, ffmpeg errors stating 'invalid data found when processing input'. I google it. Coming back I then trim it 1s in, it then encodes fine. Then I encode it without trim again and it's fine all of a sudden.
[06:05:21 CEST] <blb> Spring: nice
[06:05:38 CEST] <furq> good old heisenbugs
[11:25:28 CEST] <jimgray> ffmpeg has native RTMP. Should I build ffmpeg with librtmp?
[11:25:35 CEST] <jimgray> librtmp is better?
[14:45:57 CEST] <Threads> any reason why ffmpeg might be failing to compile on $(eval $(RULES))
[14:46:28 CEST] <Threads> common.mak and library.mak i commented them out with # so hope it works now
[16:32:11 CEST] <jacobwg_> Any ideas why the following would attempt to use "none" as the pixel format?  ffmpeg -fix_sub_duration -i input.mkv -vcodec h264_nvenc -map 0:0 -vb 835k -c:a:0 aac -map 0:1 -disposition:a:0 default -ac:a:0 2 -b:a:0 256k -metadata:s:a:0 language=eng -strict experimental -c:a:1 ac3 -map 0:1 -disposition:a:1 none -ac:a:1 6 -b:a:1 1536k -metadata:s:a:1
[16:32:11 CEST] <jacobwg_> language=eng -f mp4 -threads auto -y output.mp4
[16:32:56 CEST] <jacobwg_> It works if I explicitly specify the pixel format, but somehow it tries "none" as the output pixel format if no explicit format is specified
[17:38:07 CEST] <TwinTailed> What does "[mp4 @ 00710940] track 1: codec frame size is not set" mean?
[17:55:39 CEST] <TwinTailed> is there a way to change major brand from iso to isomp42?
[19:08:54 CEST] <Spring> ffmpeg -i input.mp4 -c:a aac -filter_complex "[0:v]scale=4:-1[v];[0:a]anull[a]" -map "[v]" -map "[a]" -vframes 1 -loglevel quiet -y -f mp4 NUL
[19:09:13 CEST] <Spring> ^ my hacky method of checking for audio stream using ffmpeg only. Any faster method?
[19:09:41 CEST] <Spring> apart from ffprobe
[19:10:16 CEST] <furq> what's wrong with ffprobe
[19:11:15 CEST] <Spring> nothing per se, it's just the idea was to have the minimum requirement as ffmpeg.exe + a script.
[19:11:46 CEST] <Spring> if there's no faster method I could instead check for ffprobe's presence and use that if it's faster
[19:13:43 CEST] <furq> i guess -vn would be faster than -vframes 1
[19:15:11 CEST] <Spring> thanks, it is
[19:17:26 CEST] <furq> ffmpeg -i in.mkv -c:a copy -vn -frames:a 1 -f null -
[19:17:29 CEST] <furq> that works for me
[19:33:59 CEST] <Spring> nice, quite a bit shorter than what I was using before I saw your message
[19:34:25 CEST] <Spring> also ffprobe's -loglevel quiet still outputs results
[19:35:12 CEST] <Spring> trying to make it print nothing but still be able to check for errors using the script
[19:37:03 CEST] <furq> -v quiet works for me
[19:38:32 CEST] <Spring> this for example prints the audio stream info in the CLI, ffprobe -i input.mp4 -show_streams -select_streams a -v quiet
[19:38:44 CEST] <furq> oh right, ffprobe
[19:39:47 CEST] <furq> well that returns 0 for files with no audio so it's probably not that helpful
[19:42:07 CEST] <furq> if you want to shut it up then use -v quiet 1> nul
[19:44:09 CEST] <Spring> using ffprobe there's only a fraction of a second saved, not sure it's worth it tbh
[19:44:28 CEST] <Spring> though, perhaps on slower systems it may be noticeably faster, not sure
[19:45:51 CEST] <Spring> I considered adding the command after a user input prompt so the command seems 'invisible' but each prompt can be hidden so there's no certain place to put it.
[19:57:25 CEST] <mp1> Hi
[19:57:50 CEST] <mp1> How stop recording ffmpeg with a command in batch ? :)
[20:04:23 CEST] <Spring> when I use 1> nul and display -v error it states 'Argument '1> nul' provided as input filename, but "....input.mp4" was already specified'
[20:05:46 CEST] <furq> put it at the end
[20:05:54 CEST] <Spring> I am
[20:07:10 CEST] <furq> that works for me with ffprobe
[20:07:28 CEST] <Spring> using -v error as well?
[20:07:36 CEST] <furq> with -v quiet
[20:07:43 CEST] <furq> either works
[20:08:20 CEST] <furq> ffprobe -v error -i foo.mkv -show_streams -select_streams v 1> nul
[20:08:56 CEST] <furq> if this is in a batch file then maybe that's screwing with it
[21:22:44 CEST] <TwinTailed> I am having a weird problem with my output h264 video. The last 5 seconds of the video are missing, but the audio continues till the end
[21:23:44 CEST] <TwinTailed> i am talking about LibAV btw not ffmpeg cli
[00:00:00 CEST] --- Sun Sep 25 2016