[Ffmpeg-devel-irc] ffmpeg.log.20161221

[burek]

[00:25:40 CET] <sib> anyone here know how to run ffserver with cuda acceleration?
[00:26:55 CET] <llogan> unfortunately "anyone here know how to run ffserver?" would get the same silent response
[00:26:55 CET] <furq> you probably shouldn't run ffserver at all
[00:27:27 CET] <llogan> the only people that use it are people who try to use it and then give up
[00:27:38 CET] <furq> i thought it had already been removed tbh
[00:28:17 CET] <llogan> no. a few noisy developers squawked at the last minute and turned into a Bikeshedfest 2016.
[00:28:34 CET] <furq> lol
[00:28:52 CET] <llogan> or at least that was the status when i stopped caring about it
[00:28:53 CET] <furq> five commits between january and october, then a bunch of activity in the last two months
[00:29:33 CET] <llogan> more importantly right now is which coworkers office garbage smells like old milk?
[00:29:38 CET] <furq> i mean i guess i should be glad that someone's fixing it but i do feel like they should just throw it in the bin
[00:29:57 CET] <furq> s/fixing it/trying to fix it/
[00:30:00 CET] <furq> important distinction
[00:30:26 CET] <llogan> pretty much the sentiment of the silent majority
[00:31:17 CET] <BtbN> it's not even an attempt to "fix" it
[00:31:35 CET] <BtbN> it's just to make it use newer API, so it won't be dropped when that deprecated API does
[00:32:02 CET] <llogan> FFhoarding
[00:32:07 CET] <BtbN> I'd be surprised if it wouldn't go back into its forgotten state after that is achived.
[00:32:23 CET] <llogan> it will become a "collector's edition".
[00:32:43 CET] <furq> yeah i got that impression from the commits
[00:32:50 CET] <furq> it's all just "update this broken-ass shit to the new api"
[00:33:40 CET] <furq> does it actually depend on ffmpeg internals
[00:33:48 CET] <furq> i'm guessing they can't just split it into its own project or something
[00:34:07 CET] <BtbN> part of it, like the ffm thing, are in libav*
[00:34:29 CET] <furq> by internals i mean private APIs etc
[00:35:33 CET] <furq> oh nvm i read that wrong
[00:42:59 CET] <sib> if not ffserver, then what would be a good alternative to stream h264 through html5
[00:43:37 CET] <BtbN> "through html5"?
[00:44:15 CET] <BtbN> Just any webserver and make ffmpeg output hls I guess?
[00:45:05 CET] <BtbN> Also, how would ffserver help with that in the first place? I'm not aware of any web-browser-streaming qualities it has
[00:52:05 CET] <furq> use the hls muxer, or use nginx-rtmp's hls muxer
[00:52:18 CET] <furq> https://github.com/dailymotion/hls.js/
[00:52:19 CET] <furq> and that
[00:53:39 CET] <sib> @BtbN @furq thanks!
[02:48:11 CET] <blue_misfit> can anyone give me some info about what async would be used for?
[02:48:22 CET] <blue_misfit> *the protocol handler*
[03:28:32 CET] <damdai> x264 core 142 r2479 dd79a61
[03:28:33 CET] <damdai> Encoding settings              : cabac=1 / ref=1 / deblock=1:0:0 / analyse=0x1:0x111 / me=hex / subme=2 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=0 / me_range=16 / chroma_me=1 / trellis=0 / 8x8dct=0 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=0 / threads=6 / lookahead_threads=2 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3
[03:28:33 CET] <damdai> / b_pyramid=2 / b_adapt=1 / b_bias=0 / direct=1 / weightb=1 / open_gop=0 / weightp=1 / keyint=250 / keyint_min=25 / scenecut=40 / intra_refresh=0 / rc_lookahead=10 / rc=abr / mbtree=1 / bitrate=6670 / ratetol=1.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40 / aq=1:1.00
[03:28:33 CET] <damdai> Default                        : Yes
[05:43:51 CET] <k-man> hi
[05:44:05 CET] <k-man> does ffpmeg infer the output format from the extension of the output file?
[05:44:28 CET] <furq> yes
[05:44:43 CET] <furq> you can force it with -f
[05:45:13 CET] <k-man> ah thanks
[05:45:49 CET] <k-man> what would the format be for m4a?
[05:45:55 CET] <furq> probably -f ipod
[05:46:48 CET] <k-man> ah thank you so much!
[05:50:30 CET] <k-man> where do i find the documentation about which file extensions map to which formast?
[05:50:32 CET] <k-man> formats
[09:17:45 CET] <sunil_> I am trying to encode an audio from aac -> vorbis for webm, but I an getting an error. Details are here - http://pastebin.com/p1R9Nkhn
[09:18:53 CET] <atomnuker> sunil_: libvorbis doesn't support bitrates under 45kbps
[10:20:12 CET] <sunil_> atomnuker: I am able to encode a 16k bitrate mono audio using libvorbis
[10:20:24 CET] <sunil_> Is this limitation documented somewhere?
[11:01:30 CET] <fling> which lossless audio codec is less cpu intensive?
[11:01:56 CET] <fling> I want to have a slight compression when capturing
[11:24:00 CET] <durandal_1707> fling: flac
[11:32:00 CET] <fahadash>  I have a video that I have to speed up a few portions of. What is the best way to do it? I am thinking about skipping the -i and using the movie filters with seek option  as many times as there is a change.
[11:39:56 CET] <fling> durandal_1707: thanks.
[11:40:39 CET] <fling> fahadash: change input fps in these portions
[11:59:06 CET] <fahadash> How ?
[11:59:30 CET] <fahadash> I can only change the fps once how can I specify 3 positions?
[12:33:09 CET] <MoonOwl> Guten Tag!
[12:34:21 CET] <MoonOwl> Has anyone ABXed high resolution music vs the redbook standard
[12:54:52 CET] <dannyzb> The E5 1650v3 processor says it has QuickSync support, but it doesn't have a GPU, how can it support Quicksync and is it usable in FFMPEG?
[12:57:05 CET] <nonex86> as far as i remember quicksync actually used a dedicated chip - mfx engine
[12:57:23 CET] <nonex86> but, it also require gpu functionality
[12:57:42 CET] <nonex86> just test it using ffmpeg
[12:58:00 CET] <nonex86> if it works, well, youll get your profit
[12:58:20 CET] <nonex86> if not... just forget it and use software implementation :)
[13:07:35 CET] <dannyzb> nonex86 I don't have the server in my hands yet - "but, it also require gpu functionality" - it says in the docs that for FFMPEG/QS OpenCL GPU support is required, wouldn't that be a no if there is no GPU?
[13:07:47 CET] <jkqxz> Quick Sync is part of the GPU, and therefore is not present on any E5 or E7 processors.
[13:08:29 CET] <dannyzb> jkqxz: http://ark.intel.com/products/82765/Intel-Xeon-Processor-E5-1650-v3-15M-Cache-3_50-GHz Intel says it does
[13:10:02 CET] <jkqxz> Er.  No it doesn't?  That page doesn't mention it at all.
[13:14:36 CET] <DexterF> hi
[13:16:38 CET] <DexterF> having some problems compiling ffmpeg on a raspi in order to have it in playce for omxplayer. the git pull in Makefile.ffmpeg the omxplayer git provides and calls in its own Makefile does not work, returns a network error. so I pulled ffmpeg from git manually, and then ran make -f Makefile.ffmpeg which worked, actually, *but* on sudo make -f Makefile.ffmpeg install it returns:
[13:16:51 CET] <DexterF> strip: 'ffmpeg_compiled/usr/local/lib/*.so': No such file
[13:17:20 CET] <DexterF> so, the ffmpeg specific question is: where/how/when is that director ffmpeg_compiled generated? is that ffmpeg specific at all?
[13:18:28 CET] <nonex86> dannyzb: not sure about the ffmpeg implementaion, but for hardware qs decoding you dont need the opencl at all
[13:19:02 CET] <nonex86> dannyzb: opencl not related to mfx engine api, i mean their software parts are not related
[14:22:44 CET] <dannyzb> nonex86: ok thanks
[14:22:52 CET] <dannyzb> nonex86: probably related to FFM implementation
[16:34:21 CET] <Mateon1> Hi, I'm having trouble transcoding unsigned 8-bit PCM to wav. I'm using ffmpeg 3.0.1; my command is `ffmpeg -c:a pcm_u8 -i piano.u8 piano.wav` and I get `piano.u8: Invalid data found when processing input` - why is that? Shouldn't PCM always decode successfully?
[16:35:27 CET] <c_14> try ffmpeg -f u8 -i piano.u8
[16:36:28 CET] <Mateon1> c_14: That worked. May I know what was wrong with my previous command?
[16:36:43 CET] <c_14> Tried detecting the format and failed
[16:36:54 CET] <c_14> you set the codec, not the format
[16:37:08 CET] <Mateon1> Ah, thank you. :)
[16:51:30 CET] <Nacht> Hello, I noticed that a certain HLS stream has Timed_ID3 tags. Is it possible to capture these for, say 1 min ?
[16:51:40 CET] <Nacht> Stream #0:7: Data: timed_id3 (ID3  / 0x20334449)
[17:10:12 CET] <rjp421> trying to get an ancient logitech quickcam go working, but ffmpeg wont take it.. http://pastebin.com/raw/fhhU607B , closest i found googling has no solution https://www.mail-archive.com/search?l=ffmpeg-issues@lscube.org&q=subject:%22%5Bissue1837%5D+problems+with+FFmpeg+and+Video2Linux2%22&o=newest&f=1
[17:12:04 CET] <rjp421> [video4linux2,v4l2 @ 0x1bd4240] Dequeued v4l2 buffer contains 26058 bytes, but 25344 were expected. Flags: 0x00000001.
[17:12:04 CET] <rjp421> /dev/video1: Invalid data found when processing input
[17:14:01 CET] <rjp421> i tried 'v4l2-ctl -d /dev/video1 --set-fmt-video=width=176,height=144,pixelformat=1' before running ffmpeg, to set the format and res to bayer.. s561 wouldnt work in my attempts. still no luck
[17:15:25 CET] <tommie_> so i'm trying to segment op1a (xdcamhd) files, into chunks of 60 seconds. It does not need to be exactly 60 seconds, aslong as the video "fits" together. unfortunately each segment looses 2 frames.
[17:15:27 CET] <tommie_> http://pastebin.com/4165B1Bu
[17:15:54 CET] <tommie_> 1st segment ends at 60.04, second segment starts at 60.12
[17:16:23 CET] <tommie_> as far as I can figure out, its due to the fact that op1a does not start with i frame, but has 2 B/P frames infront
[17:17:17 CET] <tommie_> stream order vs representation order, where ffmpeg uses the stream order (the i frame) but not the representation order, hence its skipping the 2 B/P frames
[17:18:31 CET] <tommie_> it works fine when using regular mp4, or intra-frame only codecs, all segments fit perfectly together
[17:18:49 CET] <tommie_> seems to be op1a codec only
[17:36:06 CET] <R4md4c> Hi, I'm having an FLV file that I need to convert a portion of it to be played in slow motion into a new file. I've read this link https://trac.ffmpeg.org/wiki/How%20to%20speed%20up%20/%20slow%20down%20a%20video, and the video is played in slow motion, but the duration of the video is not stretched to accomodate the slowness.
[17:36:06 CET] <R4md4c> Here is my command "ffmpeg -i input.flv -y -q 1 -ss 00:00:01 -t 00:00:16 -filter:v "setpts=(1/0,5)*PTS" . I'm expecting the video to be at least 30 seconds, but it is still 16 seconds. Is it possible to fix that?
[17:36:47 CET] <kepstin> R4md4c: you're using '-t' as an output option, which means that the trimming is being applied after the filtering
[17:37:08 CET] <kepstin> if you put the -t before the input filename (the -ss should probably be there too), then it'll apply the trim before the filtering
[17:37:25 CET] <R4md4c> Ah, OK, Interesting, I'll try that.
[17:38:20 CET] <kepstin> options before input files apply to that specific input file, options before an output file (after input files) apply when saving the video to the output file
[17:40:16 CET] <R4md4c> kepstin: It worked now, thanks a lot!
[18:21:38 CET] <rjp421> is this patch https://patchwork.kernel.org/patch/9324933/ related to the "[video4linux2,v4l2 @ 0x1bd4240] Dequeued v4l2 buffer contains 26058 bytes, but 25344 were expected. Flags: 0x00000001." "/dev/video1: Invalid data found when processing input"
[18:22:18 CET] <rjp421> they say "This patch basically makes it throw away corrupted buffers and then also the first 8 buffers after the last corrupted buffer."
[18:24:38 CET] <JEEB> rjp421: sounds like something more relevant to ffmpeg-devel mailing list, where the patches are posted
[18:25:14 CET] <JEEB> except it seems to have been posted there already?
[18:25:16 CET] <JEEB> "Well Ive posted it to the FFmpeg dev list for feedback, so we will see."
[18:26:38 CET] <rjp421> JEEB, im not sure.. my error is http://pastebin.com/raw/fhhU607B and googling took me to https://www.mail-archive.com/search?l=ffmpeg-issues@lscube.org&q=subject:%22%5Bissue1837%5D+problems+with+FFmpeg+and+Video2Linux2%22&o=newest&f=1
[18:26:53 CET] <rjp421> but it still seems to be unsolved?
[18:27:40 CET] <rjp421> i wasnt sure if that was the right mailing list or not
[18:32:51 CET] <rjp421> posibly a duplicate issue, but im using the same webcam, with the same error, as far as i can tell
[18:33:39 CET] <kerio> wait, youtube serves opus audio? ;o
[18:36:55 CET] <JEEB> kerio: for quite a while
[18:37:13 CET] <JEEB> it has AVC+AAC and VP9+opus
[18:44:34 CET] <rjp421> sorry i was confused, that mailing list is not the same error that the patch is for (or is it?)... but http://libav-user.ffmpeg.narkive.com/rLqpcxGy/ffmpeg-webcam-error is the same error with the same type (not model) webcam, and was never solved
[18:45:30 CET] <rjp421> https://www.mail-archive.com/search?l=libav-user@ffmpeg.org&q=subject:%22%5C%5BLibav%5C-user%5C%5D+FFMPEG+%5C%2B+WEBCAM+%5C%2B+ERROR%22&o=newest&f=1
[18:50:09 CET] <rjp421> it seems using v4l-ctl to set the pixelformat of the /dev/videoX to bayer (1), or it gives that "format not found" with "I/O error"
[18:52:59 CET] <rjp421> but then i have those errors in that latest paste :(
[19:18:21 CET] <fahadash> Is there a way to change FPS other than using setpts?
[19:18:38 CET] <kerio> -r
[19:18:54 CET] <kerio> it will drop/duplicate frames as needed
[19:20:18 CET] <Sashmo_> does anyone know if there is real way to force FFPLAY to open on specfic window?
[19:20:52 CET] <furq> fahadash: if you want to change the playback speed, you can do it with some formats by demuxing and remuxing the elementary stream
[19:20:56 CET] <furq> without reencoding
[19:21:13 CET] <fahadash> I don't really care much about not reencoding
[19:21:29 CET] <fahadash> Someone told me something about changing FPS, so I thought there was another filter
[19:21:46 CET] <furq> -r and -vf fps will keep the playback speed and drop or dup frames
[19:21:53 CET] <furq> -vf setpts will change the playback speed
[19:21:56 CET] <fahadash> How can I change the FPS at multiple places in the timeline while keeping all other places same original?
[19:22:03 CET] <furq> with difficulty
[19:22:09 CET] <fahadash> haha
[19:22:12 CET] <kerio> :D
[19:22:28 CET] <fahadash> I was thinking about using the movie filter to source the same file multiple time, and crop out the unwanted content
[19:22:44 CET] <furq> https://ffmpeg.org/ffmpeg-filters.html#Timeline-editing
[19:22:56 CET] <furq> idk if setpts supports that, but if it does then that's what you want
[19:23:01 CET] <furq> and -vsync vfr
[19:23:25 CET] <rjp421> to test the LD_PRELOAD fix for spca561 webcams, do i use one or the other with ffmpeg, or both v4l1compat.so and v4l2convert.so?  like motion suggests http://www.lavrsen.dk/foswiki/bin/view/Motion/SupportQuestion2010x04x28x191448
[19:26:14 CET] <blue_misfit> JEEB, remember the MP4 vs MKV timestamp thing I was looking into yesterday? I determined that you're totally right, the MVK -> MP4 files oscillate between 672 and 656, and the pure MP4 files have smooth 672.
[19:26:40 CET] <JEEB> blue_misfit: ayup
[19:26:43 CET] <blue_misfit> any thoughts on how I can re-write these timestamps when remuxing from mkv to MP4?
[19:27:37 CET] <JEEB> not in ffmpeg at least, since stuff like setpts is limited to re-encoding
[19:27:54 CET] <blue_misfit> right, that's kind of what I concluded
[19:28:04 CET] <blue_misfit> maybe l-smash?
[19:28:43 CET] <JEEB> yeah, but that only handles ISOBMFF-like things
[19:28:53 CET] <blue_misfit> or alternatively... can you think of a good way to merge MP4s (with identical encoding settings of course) into one continuous MP4?
[19:29:06 CET] <blue_misfit> I'm only using MKV because it can nicely concatenate
[19:29:15 CET] <furq> you could use mpegts
[19:29:19 CET] <blue_misfit> s/MKV/mkvmerge
[19:29:41 CET] <JEEB> was 90k enough for 24000/1001 being exact?
[19:30:02 CET] <blue_misfit> 90k?
[19:30:07 CET] <JEEB> mpeg-ts time base
[19:30:14 CET] <blue_misfit> ahh
[19:30:24 CET] <blue_misfit> not sure
[19:30:54 CET] <blue_misfit> so if I make TS fragments instead of MP4 fragments, and then just use ffmpeg concat I might be able to get a clean MP4
[19:30:55 CET] <blue_misfit> hmm
[19:31:10 CET] <JEEB> also fragmented isobmff is worth a try
[19:31:20 CET] <blue_misfit> interesting!
[19:31:39 CET] <blue_misfit> okay, thanks for the ideas, JEEB and furq
[19:43:40 CET] <rjp421> on a pi2, 'LD_PRELOAD=/usr/lib/arm-linux-gnueabihf/libv4l/v4l1compat.so ffmpeg' seemed to "fix" the error... but is that a permanently necessary workaround?
[19:49:46 CET] <blue_misfit> JEEB, damn, using MPEG-TS intermediates and concat to remux to mp4 generates timestamp durations that oscillate between 3754 and 3753 lol
[19:49:48 CET] <blue_misfit> so close
[19:50:36 CET] <bencoh> how do you concat? using cat?
[19:50:50 CET] <blue_misfit> ffmpeg -f concat <list_of_files.txt>
[19:51:02 CET] <blue_misfit> err ffmpeg -f concat -i <list_of_files.txt>
[19:51:17 CET] <JEEB> kind of guessed that a 90000 timebase wouldn't be exact
[19:51:25 CET] <blue_misfit> yeah
[19:51:43 CET] <blue_misfit> damn so... fractional 29.97 and 23.98 in TS are kind of hacky in that sense?
[19:52:13 CET] <JEEB> try using -movflags empty_moov+separate_moof+frag_keyframe
[19:52:28 CET] <JEEB> might be interesting to know how things go
[19:52:38 CET] <blue_misfit> when concating the ts fragments?
[19:52:41 CET] <blue_misfit> or when encoding mp4 fragments
[19:52:44 CET] <JEEB> encoding
[19:52:47 CET] <blue_misfit> ok
[19:55:33 CET] <JEEB> blue_misfit: also I think there even was a bug report on the tracker about lavf not fixing those oscillating timestamps when doing MPEG-TS => MP4
[19:55:46 CET] <blue_misfit> JEEB, holy crap I think it worked
[19:56:01 CET] <blue_misfit> what is this sorcery?
[19:56:11 CET] <blue_misfit> what does this do? lol
[19:56:34 CET] <JEEB> writes an empty moov in the beginning of each thing, writes a separate moof for each track and fragments at keyframes
[19:56:38 CET] <JEEB> welcome to movie fragments
[19:56:45 CET] <JEEB> also nice to know that those can be concat'd
[19:57:04 CET] <blue_misfit> timescale is now 11988 and durations are always 500
[19:57:18 CET] <JEEB> a bit funky but at least constant
[19:57:18 CET] <JEEB> lol
[19:57:26 CET] <blue_misfit> yeah
[19:57:27 CET] <JEEB> also you can see docs @ https://www.ffmpeg.org/ffmpeg-all.html#mov_002c-mp4_002c-ismv
[19:57:58 CET] <blue_misfit> so I'm totally familiar with fMP4 in terms of DASH and Smooth Streaming
[19:59:09 CET] <JEEB> well this is nothing special then
[19:59:18 CET] <JEEB> movie fragments fragmented at IDRs
[19:59:38 CET] <JEEB> the two other options are something the smooth streaming mode does by default
[19:59:40 CET] <blue_misfit> right, totally makes sense
[20:01:10 CET] <blue_misfit> okay, so I just repeated the test using mp4 chunks without the special flags to make fMP4
[20:01:19 CET] <blue_misfit> and I get the variable timestamps again
[20:01:30 CET] <blue_misfit> so what I'm trying to understand is.. why does using fMP4 help the situation?
[20:01:53 CET] <JEEB> no idea, you should check the separate segments regarding how the timestamps in those are set
[20:02:40 CET] <JEEB> in the worst case el GOOG's contribution could be ruining your day (they implemented edit lists more or less badly)
[20:03:00 CET] <JEEB> otherwise it could just be some funky detail in how lavf handles the concat
[20:03:00 CET] <blue_misfit> hmm boxdumper "failed to construct timeline" when I do --timestamp
[20:03:11 CET] <JEEB> now that's funky :P
[20:03:21 CET] <blue_misfit> the chunks do play in mpc-hc tho
[20:03:37 CET] <blue_misfit> windows can't calculate their duration either in Explorer haha
[20:03:53 CET] <blue_misfit> mediainfo gives an invalid floating point operation error on them LOL!!!
[20:03:59 CET] <JEEB> and those aren't even the fragmented ones?
[20:04:12 CET] <blue_misfit> sorry, those are the fragmented ones
[20:04:16 CET] <JEEB> ok
[20:04:30 CET] <JEEB> then a lot of stuff will just fail on them
[20:04:39 CET] <JEEB> boxdumper's thing failing is kind of interesting
[20:04:55 CET] <JEEB> if it happens just by encoding something into fragmented isobmff I'll try it out
[20:05:48 CET] <blue_misfit> seems like it does
[20:05:49 CET] <blue_misfit> too weird
[20:06:08 CET] <blue_misfit> okay, I'm not going to chase this much further. I'll do some more testing just to make sure tho. THANK YOU!
[20:06:11 CET] <JEEB> but boxdumper's box mode works I bet
[20:06:16 CET] <JEEB> np
[20:06:26 CET] <blue_misfit> ya sure does
[20:06:28 CET] <JEEB> didn't know concat'ing fragmented isobmff worked well with ffmpeg
[20:06:40 CET] <blue_misfit> MP4 Explorer can read the structure as well
[20:23:54 CET] <fortean_dad> hi all how do i use ffmpeg to strip audio out of an avi file? i'm using ubuntu 1610 if that helps
[20:24:17 CET] <fortean_dad> to be clear i want to remove the audio and keep the video
[20:27:17 CET] <furq> -i foo.avi -an -c copy bar.avi
[20:29:39 CET] <fortean_dad> furq: ta, will give it a bash
[20:41:32 CET] <kuroro> is there a way to apply pitch countour of 1 audio source (i.e song) to another (i.e. speech) using ffmpeg?
[20:41:47 CET] <kuroro> i want to do something similar to songify
[21:59:31 CET] <tokam> Hello, is there a reason why this command ignores images which do not change at the end of the video?
[21:59:33 CET] <tokam> or sequence
[21:59:51 CET] <tokam> ffmpeg -f image2 -framerate 25 -r 6 -i /var/www/tokam_dev/p172_webcam_video/tmp/1482353854/%d.jpg -s 1920x1080 -vcodec libx264 /var/www/tokam_dev/p172_webcam_video/output_videos/1482353854.avi
[22:41:50 CET] <tokam> how can I create a video with ffmpeg from images and than attach a branding image at the end?
[23:03:08 CET] <tokam> http://video.stackexchange.com/q/20207/17541
[23:03:47 CET] <furq> tokam: add -loop 1 before the second -i, then add -shortest somewhere
[23:04:05 CET] <furq> there are filters which can do it but you shouldn't need them for image2 inputs
[23:04:23 CET] <tokam> -shortest somewhere?
[23:04:25 CET] <furq> actually, don't use -shortest with overlay
[23:04:31 CET] <tokam> how to say how long to loop?
[23:04:33 CET] <furq> add :shortest=1 to the overlay command
[23:04:56 CET] <furq> it'll loop the second input forever, and stop encoding when the shortest input (the first one) ends
[23:05:03 CET] <tokam> I do not want an overlay anymore
[23:05:27 CET] <tokam> I want the first input and than show an image for n seconds at the end.
[23:05:31 CET] <furq> oh
[23:05:33 CET] <furq> https://ffmpeg.org/ffmpeg-filters.html#loop
[23:05:49 CET] <tokam> in my first attempt, I just added the images multiple times to my input sequence
[23:06:05 CET] <tokam> but ffmpeg somehow outsmarted me and knew that the images, which I added did not change
[23:07:39 CET] <furq> http://vpaste.net/msBNx
[23:07:40 CET] <furq> something like that
[23:10:08 CET] <tokam> furq: [AVFilterGraph @ 0x561144c945a0] No such filter: 'loop'
[23:10:29 CET] <furq> what ffmpeg version is that
[23:11:40 CET] <tokam> maybe this helps
[23:11:40 CET] <tokam> http://superuser.com/questions/683643/how-to-add-my-logo-for-the-first-30-seconds-in-a-video-with-ffmpeg
[23:13:06 CET] <tokam> 3.0.5 furq
[23:16:50 CET] <furq> i guess that must be too old for the loop filter
[23:17:06 CET] <c_14> 3.1 afair
[23:17:12 CET] <furq> there you go
[00:00:00 CET] --- Thu Dec 22 2016