[Ffmpeg-devel-irc] ffmpeg.log.20171201

[burek]

[00:00:57 CET] <SortaCore> play it in public and see if people give you odd looks
[00:01:15 CET] <durandal_1707> TheRock: yes, first you need to teach neural networks by giving it to them
[00:02:09 CET] <zash> neural networks solve all the problems!
[00:03:03 CET] <alexpigment> -vf sex_video_or_nah
[00:42:54 CET] <pos> An application crashes when using -vcodec copy:
[00:42:55 CET] <pos> 2017-12-01 04:41:27.765298	zmc_m1		5442	WAR	/srv/build/ZoneMinder2/src/zm_videostore.cpp:708: Mangled AVPacket: discarding frame	zm_videostore.cpp	708
[00:42:55 CET] <pos> 2017-12-01 04:41:27.551439	zmc_m1		5442	WAR	Unable to set movflags to frag_custom+dash+delay_moov
[00:43:40 CET] <pos> any ideas whats up?
[00:45:06 CET] <klaxa> that doesn't look like ffmpeg... zm_videostore.cpp?
[00:45:23 CET] <pos> no, it's calling avconv
[00:45:35 CET] <klaxa> it seems to use libav* (AVPacket) but then the call would be useufl
[00:45:38 CET] <klaxa> *useful
[00:45:42 CET] <klaxa> also avconv != ffmpeg
[00:47:32 CET] <pos> apparently the avconv job tasked with writing the video file to disk is unable to set some flags?
[00:47:54 CET] <klaxa> if it is calling avconv ask in #avconv
[00:56:48 CET] <c_14> you mean #libav (I believe)
[01:00:06 CET] <pos> i know this is #ffmpeg. i've run avconv -i "rtsp://source" -vcodec copy -acodec copy a.mp4 which ofc works
[01:01:11 CET] <pos> avconv -i "rtsp://source" -vcodec copy -acodec copy -map 0 -movflags frag_custom+dash+delay_moov a.mp4 ofc fails
[01:01:28 CET] <pos> are these extra movflags part of a separate package
[01:01:45 CET] <SortaCore> see fflogger's message, pos
[04:24:42 CET] <dystopia_> hello all
[04:24:53 CET] <dystopia_> trying to cut 60s from an m2ts file
[04:24:57 CET] <dystopia_> ffmpeg -i 00001.m2ts -t 60 -c:v copy -c:a pcm_s24le sample.m2ts
[04:24:57 CET] <dystopia_> ffmpeg -i 00001.m2ts -t 60 -c copy sample.m2ts
[04:25:10 CET] <dystopia_> ^both commands create an output without any audio
[04:25:14 CET] <dystopia_> anyone got any ideas
[04:25:53 CET] <TheRock> yes, is your windows volume maybe too small ?
[04:26:42 CET] <dystopia_> there is audio in the source file TheRock which plays fine, but not in the copied sample using ffmpeg
[04:26:53 CET] <dystopia_> so it's not a volume issue
[04:27:28 CET] <dystopia_> no audio in the output when checking with media info/ffprobe
[05:11:52 CET] <jaggz> how would I go about doing a live-preview of the output of ffmpeg, while adjusting settings (like bitrate)?
[06:08:23 CET] <FunkyBob> hey all... trying to convert a video from my phone, but the frame rate is munted... it plays way too fast
[06:09:05 CET] <FunkyBob> oh, looks like -r will do it
[06:10:03 CET] <FunkyBob> or not :/
[06:14:42 CET] <TheRock> if you are unsure
[06:14:49 CET] <TheRock> you can try all parameters from a..to z
[06:17:04 CET] <FunkyBob> ...
[10:19:49 CET] <Timmey> Hey guys, quick question: it seems I can't get the srt encoder to be included in ffmpeg on my raspberry pi 2 (raspbian). My ffmpeg installation on mac does have it out of the box.
[10:20:14 CET] <Timmey> Anyone know if I can install it manually
[12:53:52 CET] <termos> any way to set ffmpeg log level per module, like getting debug info from demuxer but not from muxers?
[14:36:48 CET] <jfmcarreira> heyyy guys
[14:37:45 CET] <jfmcarreira> i am using ffmpeg to demux and decode a stream, but av_read_frame returns EOF on the first call. that is quiet strange.
[14:58:05 CET] <pihpah> Is it a good bitrate for a 720p video ? 764 kb/s, 23.98 fps
[14:58:58 CET] <furq> depends but probably not
[14:59:43 CET] <pihpah> furq: you mean such a bitrate is too great for that kind of videos?
[15:00:25 CET] <pihpah> YouTube recommends on 5mbps for 30 fps videos.
[15:00:41 CET] <JEEB> there is no magic bit rate
[15:00:54 CET] <furq> i don't know why they'd recommend that considering they recompress 720p30 to 2.5mbit
[15:00:57 CET] <JEEB> if you are using x264,5 use CRF
[15:01:19 CET] <pihpah> Well, I do that with -crf 28
[15:01:44 CET] <JEEB> for x264 that's puh-ret-teh high of a value, although it's possible that it looks OK to you
[15:01:55 CET] <JEEB> and if uploading to a service just encode in as high possible quality as possible if you know the service will re-encode your stuff anyways
[15:02:07 CET] <furq> crf 28 will look "OK"
[15:02:14 CET] <furq> it's far from archival quality though
[15:07:14 CET] <pihpah> Yeah, I know that. I am using those videos for my own streaming service. Just started to think lately whether I overuse my resources by improper video encoding.
[16:15:08 CET] <SortaCore> when transcoding, do you use the same hw_device_ctx for decoding and encoding?
[16:15:19 CET] <SortaCore> rip buried
[16:22:35 CET] <traw> what is the purpose of AVIOInterruptCB in avio_open2 function?
[16:30:11 CET] <DHE> traw: some IO operations are effectively blocking, most notably network IO. this lets you provide a function that has the opportunity to veto waiting on the network and return an error code into the system
[16:34:14 CET] <traw> okay thanks DHE
[17:34:34 CET] <gon_> hello, came in a few days ago asking about hstack and aspect ratio. As long as the two inputs are the same height I have pretty good luck with the final output but I'm having trouble with a couple. setsar was recommended but I can't seem to use it with filter_complex https://pastebin.com/AQ3aQBsg
[17:54:21 CET] <DHE> gon_: you'll have to put it all into a single -filter_compltex
[18:08:43 CET] <gon_> So I'd just append that to the end of filter_complex, DHE ? Input 1 has the correct aspect ratio but for some reason input 2 is being stretched. A couple other clips I've put together have come out correctly but I just have a couple that have this issue
[18:12:33 CET] <DHE> ffmpeg -i ... -i ... -filter_complex "hstack,scale=...,setsar=...[output]" -map [output] -map 0:a $CODEC_OPTIONS outputtest.mp4
[18:12:56 CET] <gon_> Thanks DHE, I'll give that a try
[18:13:10 CET] <durandal_1707> gon_: if you are attempting to combine videos with different aspects you need to pad before combining
[18:13:42 CET] <DHE> I made that up on the spot so it might need fixing
[18:15:26 CET] <gon_> I'll look up padding since I haven't used it before. New to filters in ffmpeg
[18:16:54 CET] <gon_> The heights of both inputs are the same. Some successful clips come out with the width of input1+input2 while two I've tried so far come out with input2 stretched.
[19:47:41 CET] <sledg_000> ffmpeg -y -r 10 -i downloads/export_palette/001export.png -i downloads/export_palette/002export.png -i out.mp3 -filter_complex "nullsrc=size=640x480 [base]; [0:v]setpts=PTS-STARTPTS,scale=320x240[img1];[1:v]setpts=PTS-STARTPTS,scale=320x240[img2];[base][img1] overlay=shortest=1,[tmp1]; [tmp1][img2] overlay=shortest=1:x=320" -c:v libx264  -vframes 200 out.mp4
[19:48:26 CET] <sledg_000> i am trying to delay the first image
[19:50:22 CET] <sledg_000> i want the first image to show for 10seconds
[19:50:46 CET] <sledg_000> the second image to show after 10sec
[19:52:06 CET] <sledg_000> ffmpeg -y -r 10 -i downloads/export_palette/001export.png -i downloads/export_palette/002export.png -i out.mp3 -filter_complex "nullsrc=size=640x480 [base]; [0:v]setpts=PTS-STARTPTS,scale=320x240[img1];[1:v]setpts=PTS-STARTPTS,scale=320x240[img2];[base][img1] overlay=shortest=1,XXXXXXXXXXX[tmp1]; [tmp1][img2] overlay=shortest=1:x=320" -c:v libx264  -vframes 200 out.mp4
[19:52:24 CET] <sledg_000> i feel the filter i need should where i put XXXXXXXXXXX
[19:53:48 CET] <samoo8> How can I automatically rename output file like video.mkv to video-copy.mkv ? I have this bat script atm: ffmpeg -i %1 -c:v copy -c:s copy -c:a mp3 %1
[19:57:01 CET] <samoo8> i might convert mp4s also so it needs to do video.mp4 --> video-copy.mp4 also
[20:25:36 CET] <dystopia_> samoo8 use an if exist before the ffmpeg command
[20:28:08 CET] <dystopia_> like this
[20:28:13 CET] <dystopia_> if exist video.mkv (set output=video-copy.mkv) else (set output=video.mkv)
[20:28:13 CET] <dystopia_> ffmpeg -i %1 -c:v copy -c:s copy -c:a mp3 %output%
[20:58:09 CET] <samoo8> dystopia_: thx mate
[20:58:36 CET] <dystopia_> np
[21:00:02 CET] <sledg_000>  i am trying to delay the first image
[21:00:03 CET] <sledg_000> <sledg_000> i want the first image to show for 10seconds
[21:00:03 CET] <sledg_000> <sledg_000> the second image to show after 10sec
[21:00:03 CET] <sledg_000> <sledg_000> ffmpeg -y -r 10 -i downloads/export_palette/001export.png -i downloads/export_palette/002export.png -i out.mp3 -filter_complex "nullsrc=size=640x480 [base]; [0:v]setpts=PTS-STARTPTS,scale=320x240[img1];[1:v]setpts=PTS-STARTPTS,scale=320x240[img2];[base][img1] overlay=shortest=1,XXXXXXXXXXX[tmp1]; [tmp1][img2] overlay=shortest=1:x=320" -c:v libx264  -vframes 200 out.mp4
[21:00:13 CET] <sledg_000> anyhelp
[21:00:54 CET] <sledg_000> basically what i am trying to achieve is
[21:01:08 CET] <sledg_000> i have img1, img2, img3
[21:01:29 CET] <sledg_000> i want to create a video to show img1 for 5 sec
[21:01:45 CET] <sledg_000> then switch to img2 for 10sec
[21:01:56 CET] <alexp> https://trac.ffmpeg.org/wiki/Slideshow
[21:02:16 CET] <alexpigment> see the "sequential" section
[21:02:22 CET] <sledg_000> yes
[21:02:32 CET] <sledg_000> i used that but it does not show
[21:03:00 CET] <sledg_000> i also have a background mp3 that must play at the background throughout
[21:03:06 CET] <alexpigment> right
[21:03:19 CET] <alexpigment> so it sounds like you want the -r to be 1/10
[21:03:33 CET] <alexpigment> and then you can define the frame rate as whatever after the -i has been specified
[21:03:34 CET] <sledg_000> what happens is
[21:04:11 CET] <sledg_000> the output video show img1, img2 and img3 on first frame of video
[21:04:28 CET] <sledg_000> then img3 just sits all through the video
[21:04:49 CET] <alexpigment> did you try -r 1/10?
[21:04:50 CET] <sledg_000> i need a way to tell it to show img1 for 5 sec before switching to img2
[21:04:55 CET] <alexpigment> rather
[21:05:23 CET] <alexpigment> try ffmpeg -framerate 1/10 -i [inputs] -r 24 etc
[21:05:35 CET] <sledg_000> let me retry, i believe i tried every possible approach, it did not work
[21:05:40 CET] <alexpigment> ok
[21:05:41 CET] <sledg_000> i will retry this
[21:05:55 CET] <alexpigment> i'll follow along
[21:06:01 CET] <sledg_000> thanks buddy
[21:08:27 CET] <alexpigment> ok, i can see what you're talking about
[21:08:37 CET] <alexpigment> maybe you need to use the concat demuxer
[21:23:00 CET] <alexpigment> sledg_000: https://superuser.com/questions/1049452/ffmpeg-create-a-slideshow-video-from-images-in-a-directory-sorted-alphabetica
[21:23:08 CET] <alexpigment> see the answer with the green check mark
[21:23:20 CET] <alexpigment> as I mentioned above, you have to use the concat demuxer
[21:27:07 CET] <sledg_000> thanks
[21:27:28 CET] <sledg_000> i think the problem i have with the playback is from the player
[21:27:50 CET] <sledg_000> vlc changes the images
[21:28:18 CET] <sledg_000> but window media player just went black
[21:32:18 CET] <sledg_000> thanks
[21:32:32 CET] <sledg_000> seem like i am getting a way
[21:33:08 CET] <JEEB> -41
[21:33:10 CET] <sledg_000> still not there but this is extremely helpful
[21:35:25 CET] <alexpigment> sledg_000: when i followed that answer, i got WMP to play it correctly
[21:35:44 CET] <sledg_000> now i think i need a way to access the audio length
[21:35:55 CET] <alexpigment> so that it ends at the end of the track?
[21:35:59 CET] <alexpigment> if so, try -shortest
[21:36:27 CET] <sledg_000> do i can do an audiolength/number_of_images
[21:36:37 CET] <sledg_000> to evenly distribute images
[21:36:43 CET] <alexpigment> do you need to see all of the images?
[21:36:47 CET] <alexpigment> or are they just kind of filler?
[21:36:56 CET] <sledg_000> while ensuring the song plays through
[21:37:20 CET] <sledg_000> the audio take highest priority
[21:37:25 CET] <alexpigment> then use -shortest
[21:37:39 CET] <alexpigment> it'll stop encoding frames at the end of the audio
[21:37:46 CET] <alexpigment> put that after the inputs, btw
[21:38:05 CET] <sledg_000> shortest stops at the end of the image slide
[21:38:13 CET] <sledg_000> so audio never finishes
[21:38:20 CET] <sledg_000> let me play around a bit
[21:38:26 CET] <alexpigment> ok, so you don't have enough images then
[21:38:33 CET] <alexpigment> anyway, good luck
[21:38:43 CET] <sledg_000> i am guessing i need longest
[21:38:48 CET] <sledg_000> if it even exists
[21:39:06 CET] <sledg_000> if the audio is a 60sec audio
[21:39:08 CET] <alexpigment> honestly, i've never heard of longest
[21:39:15 CET] <alexpigment> it may exist, but i don't know about i
[21:39:16 CET] <alexpigment> *it
[21:39:32 CET] <sledg_000> then the 3 images gets splitted
[21:39:38 CET] <sledg_000> thanks
[21:58:36 CET] <sledg_000> -vf fps=25 option saved the day for me
[22:06:22 CET] <alexpigment> sledq_000 are you in a pal country?
[22:07:07 CET] <alexpigment> (i personally think pal frame rates are pointless in 2017, unless your source material is at that same frame rate)
[22:07:16 CET] <alexpigment> either way, you're doing stills, so it doesn't matter really
[22:08:56 CET] <furq> you know we still broadcast at 25p/50i right
[22:09:26 CET] <alexpigment> furq: yep. it's just that unless your source content is 25p inherently, there's no advantage to pal frame rates
[22:09:49 CET] <alexpigment> and I've yet to see a pal TV or player that isn't internally NTSC with some conversion hardware
[22:10:07 CET] <alexpigment> which is why all "multi-region" blu-ray players are just PAL players
[22:10:25 CET] <furq> this may be the case but it's still better than ffmpeg defaulting to 29.97
[22:10:45 CET] <alexpigment> furq: i'm not terribly concerned with the defaults
[22:11:06 CET] <alexpigment> more just that it doesn't make sense to use 25fps for purely synthetic content
[22:11:16 CET] <alexpigment> especially considering that it will never look smooth on a computer monitor
[22:11:31 CET] <therage3> i think what alexpigment is saying that unless the origin is a PAL broadcast, 25fps doesn't make much sense
[22:11:35 CET] <therage3> if i'm not mistaken
[22:12:19 CET] <alexpigment> therage3: yeah, that's exactly what I was saying. maybe I jumbled some words or didn't say it eloquently enough above ;)
[22:13:44 CET] <alexpigment> but yeah, i've talked to lots of pal people who are like "this pan isn't smooth". and it's because 25p or 50i just doesn't divide evenly into 60hz
[22:14:32 CET] <therage3> framerates always bother me
[22:14:51 CET] <alexpigment> therage3: i like the simplicity and complete ambiguity of that statement :)
[22:15:52 CET] <therage3> heh
[22:15:52 CET] <alexpigment> framerates bother me too, but it's usually because content providers don't care about frame rates and I have to look at judder on services that aren't worthy of paying for
[22:16:07 CET] <therage3> well, if you remember yesterday, there was this NTSC DVD i was trying to detelecine
[22:16:16 CET] <alexpigment> oh right
[22:16:19 CET] <therage3> and it turned out the detelecine was applied in not a specified pattern
[22:16:45 CET] <therage3> (the framerate comment here applies because it has to convert 29.97 to 24 after decimating duplicates)
[22:17:30 CET] <alexpigment> yep
[22:17:35 CET] <alexpigment> it's all kinda dumb
[22:17:51 CET] <alexpigment> NTSC does the 3/2 pulldown
[22:17:59 CET] <alexpigment> PAL usually does a 4% speedup
[22:18:11 CET] <alexpigment> I personally prefer the former, but both are definitely compromises
[22:19:24 CET] <alexpigment> what's awful though is that there are many paid tv streaming services now
[22:19:46 CET] <alexpigment> DirecTV Now, Playstation Vue, YouTube TV, etc
[22:20:11 CET] <alexpigment> and they take 1080i broadcasts and convert them to 720p30 or 1080p30 (and sometimes 720p60)
[22:20:21 CET] <therage3> some people do notice that the speedup introduces the "chipmunk effect" into people's voices
[22:20:31 CET] <therage3> and this is especially bad for musicals, where pitch perfection is required
[22:21:06 CET] <alexpigment> you completely lose the 3/2 pulldown when you convert from 60i to 30p. and so judder is just awful, and unaccepatable on a paid service. but people just have very low standards and so there's not a huge uproar that forces changes to be made
[22:21:34 CET] <alexpigment> yeah, the speedup is much more noticeable to me than the judder from a proper 3/2 pulldown
[22:23:12 CET] <therage3> yeah
[23:06:22 CET] <durandal_170> alexpigment: i pushed fillborders filter
[23:08:49 CET] <alexpigment> oh nice
[23:08:59 CET] <alexpigment> that'll be really handy in the future
[23:09:25 CET] <alexpigment> I ended up using crop+pad yesterday, although getting the parameters right was confusing because crop and pad don't work like you'd expect
[23:10:27 CET] <alexpigment> at least, crop doesn't work like you'd expect (you specify the uncropped area rather than the cropped area)
[23:11:36 CET] <alexpigment> lemme rephrase: it didn't work like i'd expect, but this is because I only need to crop on one side
[23:11:51 CET] <alexpigment> anyway, looking forward to using fillborders :)
[23:43:11 CET] <teratorn> what's a good way to create mp4 files that use data referencing without copying frame data from a source
[00:00:00 CET] --- Sat Dec  2 2017