[Ffmpeg-devel-irc] ffmpeg.log.20151101

[burek]

[00:48:51 CET] <Fjorgynn> lol
[00:48:58 CET] <Fjorgynn> I think I must buy a enw pc
[09:11:12 CET] <abd5932> Hello, I have to sets of .png frames that I want to unite into a video clip. I would like to display them side by side on black background, each occupying 1/4 of the screen. What is the best way(/tool) to do that? Thank you!
[09:11:21 CET] <abd5932> two sets
[09:45:19 CET] <durandal_1707> abd5932: pad and hstack filters
[09:46:27 CET] <abd5932> durandal_1707: thank you! I'll check the docs...
[09:54:12 CET] <abd5932> durandal_1707: Another question - I would like to output some text beneath the two sets on the padded space in form "type writer" effect... is there a filter to do that?
[10:00:57 CET] <durandal_1707> no, there is filter that just draw text without any effect
[10:01:26 CET] <durandal_1707> *type writer effect
[10:01:41 CET] <waressearcher2> durandal_1707: "durandal_1706" and down were taken ?
[10:11:52 CET] <Fyr> c_14, is it possible to calculate PSNR of multiple files? i.e.:
[10:11:52 CET] <Fyr> ffmpeg -i source.mkv -i output1.mkv -i output2.mkv -i output3.mkv etc
[10:12:27 CET] <Fyr> I know only how to calculate PSNR of two files.
[10:13:20 CET] <c_14> Use a filter complex to compare each file/stream against the source
[10:13:41 CET] <c_14> Might need to specify a (different) stats_file per filter though, otherwise it might get confusing which stats are for which result
[10:13:57 CET] <Fyr> ok, how do I do that?
[10:14:59 CET] <c_14> -filter_complex "[0:v][1:v]psnr=stats_file=stats1;[0:v][2:v]psnr=stats_file=stats2"
[10:15:00 CET] <c_14> etc
[10:15:09 CET] <Fyr> thanks
[10:24:09 CET] <abd5932> durandal_1707: what is the name of the filter to draw simple text?
[10:24:17 CET] <c_14> drawtext
[10:24:24 CET] <abd5932> thank you!
[12:48:08 CET] <toriso> Hello
[12:55:31 CET] <waressearcher2> ffprobe gives me this information: "29.97 tbr, 30k tbn, 59.94 tbc" what are they mean ? those tbr,tbn,tbc ?
[13:10:20 CET] <IntelRNG> A random hit at ixquick shows lots of results for that.
[13:14:50 CET] <waressearcher2> what ixquick ?
[13:47:21 CET] <waressearcher2> is ffmpeg project lives on donations ?
[13:54:12 CET] <durandal_170> yes
[13:54:30 CET] <waressearcher2> you lost your 7 ?
[14:06:40 CET] <durandal_170> so what?
[14:09:55 CET] <waressearcher2> you are like useless now
[14:10:23 CET] <waressearcher2> or should I say you lost 1537 ?
[14:13:14 CET] <durandal_170> dont troll
[14:23:53 CET] <anYc> Is there an easy way with ffmpeg to detect if an I-frame (AVPacket or AVFrame) is an IDR frame?
[14:26:28 CET] <anYc> (in a h264 stream)
[14:28:43 CET] <anYc> or at least an easier way than parsing the data myself
[14:40:37 CET] <JEEB> there's no proper flags for all the stuff, but the keyframe flag should do what you need in general
[14:40:55 CET] <JEEB> it gets more "fun" with IRAPs and all the kinds of pictures with HEVC
[14:46:40 CET] <toriso> I'm getting this message: Using network protocols without global network initialization. Please use avformat_network_init(), this will become mandatory later.
[14:54:04 CET] <anYc> JEEB: thanks but I think I really need to know if it's an IDR frame. I'm currently debuggin my cutting tool (https://github.com/anyc/avcut)
[14:54:43 CET] <anYc> some helper functions would already be nice
[14:55:47 CET] <JEEB> I think only the parser knows exactly what type of picture it is, and that information is not AFAIK exported. that said, picture type I and keyframe flag being set should be a random access point (you would probably have to prepend parameter sets before it in the cut version of course, or otherwise make them available)
[15:04:54 CET] <anYc> I found a bluray video that seems to miss frames before an I frame. Therefore I'd like to test if it works if I cut at IDR frames. I think I handle copying the extradata (with those avcc-annexb filter) already as that fixed other video files.
[15:05:08 CET] <anYc> I'll check if I can parse it, thanks!
[15:06:06 CET] <JEEB> yes, you should only cut on keyframes
[15:06:15 CET] <JEEB> with HEVC, IRAP, with AVC on IDR
[15:06:26 CET] <JEEB> there's also open gop with AVC but I am not going to go into that
[15:06:55 CET] <JEEB> start with using the keyframe flag
[15:07:04 CET] <JEEB> don't remember in which struct it is
[15:10:47 CET] <anYc> I already do, therefore I think I need to cut at "IDR boundaries"
[15:10:56 CET] <anYc> with this video
[15:11:09 CET] <JEEB> uhh
[15:11:13 CET] <JEEB> keyframe != I
[15:11:24 CET] <JEEB> IIRC keyframe flag only gets set in IDR frames
[15:11:30 CET] <JEEB> I might be incorrect of course
[15:11:54 CET] <anYc> oh okay, didn't know that. that would be easy then, I'll try thank you!
[16:13:42 CET] <bencc1> is it possible to use "ffprobe -show_frames" for only part of a video file?
[16:13:44 CET] <bencc1> to save time
[16:37:21 CET] <Admin_234234> hey everyone good day
[16:38:05 CET] <Admin_234234> anyone know why when i am running segmenting with ffmpeg. it deletes segments fine and creates them based on the corect size for a while.. and then randomly.. the segment ( .ts ) file no longer segments... it just increments one file forever until the hard drive runs out of space
[16:38:09 CET] <Admin_234234> a bug ?
[16:38:38 CET] <c_14> What's your commandline and ffmpeg version?
[16:38:48 CET] <c_14> And how long is "a while"
[16:39:04 CET] <Admin_234234> a while coudl be a day or hours
[16:39:15 CET] <c_14> Not always the same amount of time?
[16:39:39 CET] <Admin_234234> ffmpeg version 2.8.1
[16:39:53 CET] <Admin_234234> no i dont know that part exactly actually i went to bed.. came back my hard drive was full
[16:40:01 CET] <Admin_234234> i had like 10 sessions running and 1 or 2 were still ok
[16:40:10 CET] <Admin_234234> others were 150MB size on one .ts file
[16:41:03 CET] <Admin_234234> restarted them about 1 hour ago.. so far all are well
[16:41:59 CET] <Admin_234234> not the first time it has happened.. i thought first time.. ehh something stupid.. now i nkow there is a problem :(
[16:43:22 CET] <c_14> What's the commandline?
[16:45:45 CET] <Admin_234234> http://pastebin.com/3JUcewH2
[16:48:23 CET] <c_14> mhm
[16:48:28 CET] <c_14> What's the source?
[16:48:34 CET] <Admin_234234> UDP source
[16:48:35 CET] <Admin_234234> live stream
[16:48:50 CET] <Admin_234234> mpeg2
[16:49:19 CET] <Admin_234234> most segmented files are 2 - 3 M which is good
[16:49:28 CET] <Admin_234234> but randomly one of them doesn't rotate
[16:49:52 CET] <Admin_234234> just keeps going ++++150MB+ and since i no longer have room in the tmpfs ramdisk it stops
[16:50:22 CET] <c_14> Do you have one of these too-long outputs lying around still?
[16:50:24 CET] <Admin_234234> there has to be a bug there where it messes up on the segmenting and doesn't know to rotate the file
[16:50:32 CET] <Admin_234234> no.. got cleared :(
[16:50:43 CET] <Admin_234234> when i reset it clears all old stuff
[16:50:48 CET] <c_14> Because it will only split on keyframes
[16:50:53 CET] <Admin_234234> but i can get it next time it happens for sure..
[16:50:58 CET] <c_14> Since you're using -c copy, if there's no keyframe it wouldn't split
[16:51:01 CET] <Admin_234234> hum... so you think something is up with the key frames ?
[16:51:13 CET] <c_14> It's a possible problem.
[16:51:15 CET] <Admin_234234> hum... i am using -c copy
[16:51:39 CET] <Admin_234234> actually yes this source is not UDP
[16:51:49 CET] <Admin_234234> sorry i got mixed up with another source... yes these sources are -c copy
[16:53:15 CET] <Admin_234234> i dont want to transcode.. its already h264
[16:53:27 CET] <Admin_234234> i just want to pull the stream and segment it
[16:54:22 CET] <c_14> I see 2 possible problems, one is the keyframe issue. The other is if the timestamps aren't strictly monotonically increasing (but it should warn on that)
[16:54:31 CET] <c_14> It's hard to find out which if it can't be reproduced easily.
[16:55:11 CET] <Admin_234234> yup i know.. well its def. random i think.. doesn't happen for a long time and i leave 50 strams going.. i come back a day later and the drive is maxed out :(
[17:32:29 CET] <wlritchi> I've got a script that wraps ffprobe to get the bitrate of the audio stream in a file, extracting it using grep and sed. However, ffprobe doesn't report anything in kb/s for opus streams. Is there any good way to determine the average bitrate of an opus stream besides extracting it with ffmpeg and getting the size?
[17:44:43 CET] <Fyr> wlritchi, have you tried mediainfo?
[17:47:23 CET] <wlritchi> Fyr: Thanks for the suggestion, but it has the same problem. I'm assuming opus doesn't have any sampling rate field anywhere, and neither tool feels like scanning the full file to figure it out
[17:48:47 CET] <Fyr> OPUS' sampling rate is 48 kHz.
[17:49:03 CET] <wlritchi> Sorry, that was tired brain. I meant bitrate.
[17:49:11 CET] <wlritchi> (sampling rate both tools report)
[17:51:08 CET] <Mavrik> Which of the ffprobe output formattings do you use?
[17:52:53 CET] <Fyr> wlritchi, I just checked, ffmpeg determines OPUS bitrate.
[17:53:11 CET] <Fyr> ffmpeg -i temp.opus -f null -y /dev/null
[17:53:24 CET] <Fyr> Stream #0:0: Audio: pcm_s16le, 48000 Hz, stereo, s16, 1536 kb/s
[17:54:07 CET] <Fyr> sorry, wrong line:
[17:54:07 CET] <Fyr> Duration: 00:02:02.55, start: 0.000000, bitrate: 98 kb/s
[17:54:16 CET] <Fyr> it shows bitrate.
[17:54:18 CET] <Fyr> =/
[17:55:15 CET] <wlritchi> Confirmed, that works. Adding a -codec copy to that line saves on CPU usage as well
[17:56:27 CET] <wlritchi> Actually, will that work if there are multiple streams in the source file?
[17:57:05 CET] <furq> it probably won't work if there's more than one vbr stream in the file
[17:57:12 CET] <furq> depending on which container you use
[18:41:52 CET] <cousin_luigi> Hello.
[18:41:57 CET] <cousin_luigi> What do you make of this http://pastie.org/private/jpdrfxdaeuxqfn2k5rvmq ?
[18:46:11 CET] <DragonsLordKERNE> hi guys!
[18:46:15 CET] <DragonsLordKERNE> c_14, are you there?
[18:46:23 CET] <waressearcher2> DragonsLordKERNE:
[18:46:25 CET] <waressearcher2> waressearcher2:
[18:46:34 CET] <DragonsLordKERNE> I tried to set up a fixed GOP each 250 frames
[18:47:24 CET] <DragonsLordKERNE> using options g, keyint_min, sc_treshold but I failed
[19:21:17 CET] <waressearcher2> does anyone knows how to make audio sound as if the source is more up or more down ?
[19:22:13 CET] <waressearcher2> to make it sound more left or right one have to just lover value of left or right, but how to make it sound higher or lower ?
[19:24:46 CET] <cousin_luigi> bbl
[20:39:04 CET] <AlexYeCu> Hi everybody. A question. I want to capture sounds from sdl application. So, I`m loading snd-aloop kernel module, changing device at openal/alsoft.conf, and after that audacity can capture the sound. But when I do ffmpeg -f alsa -ac 2 -i hw:2,1 -f x11grab  -s 1920x1080 -i :0.0 -strict -2 -acodec aac -vcodec libx264  -threads 0 output.mkv (any audio codec, or even without it), I`m getting cannot set sample format 0x10000 2 (Invalid argu
[20:39:18 CET] <StephenS> currently I generate thumbnail like this: "{$ffmpeg} -i \"{$uploadedfile}\" -deinterlace -an -ss 1 -t 00:00:01 -r 1 -y -vcodec mjpeg -f mjpeg \"{$thumbnail}\" 2>&1" however, its getting the first, and I want somehow to get thumbnail which is stabilized, with no big "distortion", what should I add/use?
[20:39:29 CET] <StephenS> its getting first frame*
[20:45:59 CET] <waressearcher2> AlexYeCu: "(Invalid argu" then no more text
[20:46:56 CET] <ChocolateArmpits> StephenS: Do you mean with no motion ?
[20:47:00 CET] <AlexYeCu> waressearcher2 Invalid argument, then brackets closes. That`s all.
[20:47:47 CET] <StephenS> ChocolateArmpits, kinda, so I get the best possible thumbnail
[20:49:25 CET] <ChocolateArmpits> StephenS: Well best possible is very subjective, not very programmatical under common conditions
[20:50:12 CET] <ChocolateArmpits> Is this a completely automated process ?
[20:50:30 CET] <ChocolateArmpits> You could fetch several thumbnails and then pick  yourself which is the best
[20:51:04 CET] <ChocolateArmpits> Or maybe do motion analysis to find which segments in the video have a stable shot and then pick a frame from there hoping it's somehow meaningful
[20:52:37 CET] <DragonsLordKERNE> ciao ChocolateArmpits
[20:52:46 CET] <ChocolateArmpits> DragonsLordKERNE: hi
[20:52:50 CET] <DragonsLordKERNE> I tried the setting suggested by you, yesterday
[20:52:56 CET] <DragonsLordKERNE> but I fail
[20:52:57 CET] <DragonsLordKERNE> :(
[20:53:18 CET] <ChocolateArmpits> DragonsLordKERNE: did the command fail or was there a problem with the  rendered file ?
[20:53:33 CET] <waressearcher2> you people like long things do you ?
[20:54:03 CET] <DragonsLordKERNE> the output file has not fixed GOP each 250 frames
[20:54:04 CET] <DragonsLordKERNE> :(
[20:55:05 CET] <DragonsLordKERNE> I am wondering how avidemux is able to do this trick
[20:55:15 CET] <ChocolateArmpits> DragonsLordKERNE: ok I'll try running the command myself with ffmpeg
[20:55:30 CET] <DragonsLordKERNE> thank you very much, ChocolateArmpits
[20:55:49 CET] <DragonsLordKERNE> would you like me to provide one of the source I use
[20:55:58 CET] <ChocolateArmpits> nah I have plenty of files
[20:56:01 CET] <ChocolateArmpits> to test on
[20:56:01 CET] <waressearcher2> megan fox
[20:56:12 CET] <DragonsLordKERNE> and also the command line I used?
[20:56:18 CET] <waressearcher2> make it stop
[20:56:25 CET] <StephenS> ChocolateArmpits, it should be automated.
[20:56:43 CET] <DragonsLordKERNE> ChocolateArmpits, however my source is a .MP4 file (H.264)
[20:57:04 CET] <ChocolateArmpits> DragonsLordKERNE: the source shouldn't matter realistically
[20:59:37 CET] <DragonsLordKERNE> ChocolateArmpits, here is what I did: > http://pastebin.com/aaaMP2nw <
[21:01:22 CET] <StephenS> uhm real quick: ffmpeg -i ham.webm -i main_watermark.png -filter_complex "overlay=x=(main_w-10):y=(main_h-10)" -strict -2  out.mp4
[21:01:27 CET] <StephenS> watermark is not showing in video
[21:01:41 CET] <StephenS> it should've been in right bottom corner
[21:01:47 CET] <AlexYeCu> Well, while a question about sample format has no answer, another question. I`m getting twiced frame rate of captured video. But why?
[21:03:55 CET] <StephenS> ah nvm I have missed overlay
[21:20:34 CET] <DragonsLordKERNE> ChocolateArmpits, are you still there?
[21:20:51 CET] <ChocolateArmpits> DragonsLordKERNE: ye, I think I have it working, just wait a sec
[21:21:04 CET] <DragonsLordKERNE> :)
[21:21:24 CET] <ChocolateArmpits> DragonsLordKERNE: yeah it's working
[21:21:31 CET] <ChocolateArmpits> a different command line though
[21:21:33 CET] <ChocolateArmpits> I'll pastebin it
[21:21:40 CET] <DragonsLordKERNE> OK!
[21:21:48 CET] <DragonsLordKERNE> Did you start from my sample
[21:21:59 CET] <DragonsLordKERNE> or you started by yourself?
[21:22:02 CET] <ChocolateArmpits> nah, the keyint_min, g, and sc_threshold triplet
[21:22:06 CET] <ChocolateArmpits> like I suggested yesterday
[21:22:19 CET] <DragonsLordKERNE> I already set them
[21:25:41 CET] <ChocolateArmpits> DragonsLordKERNE: this here it: http://pastebin.com/694dzP6s
[21:25:58 CET] <ChocolateArmpits> I switched to mpeg4, though it's explicitly listed in the codecs page
[21:26:03 CET] <ChocolateArmpits> not *
[21:26:43 CET] <ChocolateArmpits> I wasn't really able to get libxvid to work this way
[21:27:28 CET] <DragonsLordKERNE> Do you know if it also work with XviD
[21:27:31 CET] <DragonsLordKERNE> as codec
[21:27:31 CET] <ChocolateArmpits> So it's basically an alternative
[21:27:38 CET] <DragonsLordKERNE> insted of MP4?
[21:27:53 CET] <ChocolateArmpits> mpeg4 isn't MP4
[21:28:00 CET] <DragonsLordKERNE> sorry
[21:28:08 CET] <DragonsLordKERNE> XviD instead of MPEG-4
[21:28:15 CET] <ChocolateArmpits> xvid is a codec under mpeg4-part 2 spec
[21:29:25 CET] <ChocolateArmpits> "Xvid (formerly "XviD") is a video codec library following the MPEG-4 video coding standard"
[21:29:27 CET] <ChocolateArmpits> from wiki
[21:29:53 CET] <ChocolateArmpits> using mpeg4 produces the same coded stream as would with libxvid
[21:30:11 CET] <ChocolateArmpits> as in, the streams are coded to the same standard
[21:30:18 CET] <Mavrik> Same standard, not the same stream.
[21:30:19 CET] <Mavrik> :)
[21:30:34 CET] <ChocolateArmpits> yeah, I tried to correct myself
[21:30:40 CET] <DragonsLordKERNE> What do you think avidemux is doing setting min keyframe to 1 and max to 250?
[21:30:52 CET] <DragonsLordKERNE> is it simply setting closed gop?
[21:31:07 CET] <Mavrik> Also, wtf, xvid in 2015?
[21:31:09 CET] <ChocolateArmpits> closed gop doesn't necessarily result in equal gop size
[21:32:00 CET] <DragonsLordKERNE> yes but
[21:32:04 CET] <DragonsLordKERNE> I can't understand
[21:32:04 CET] <ChocolateArmpits> running libxvid  with the closed gop flag doesn't result in equal gop size at all
[21:32:08 CET] <ChocolateArmpits> Which part ?
[21:32:50 CET] <DragonsLordKERNE> I simply set the GOP size to 250 within avidemux
[21:32:54 CET] <DragonsLordKERNE> and the output is
[21:33:00 CET] <DragonsLordKERNE> min keyframe = 1
[21:33:05 CET] <DragonsLordKERNE> max keyframe = 250
[21:33:28 CET] <DragonsLordKERNE> but GOPs are each 250 frames
[21:33:30 CET] <DragonsLordKERNE> always
[21:33:36 CET] <DragonsLordKERNE> without any null frames
[21:36:02 CET] <ChocolateArmpits> DragonsLordKERNE: I'm guessing it's then using mpeg4, setting keyintmin to 1 still produces fixed gop size file
[21:36:11 CET] <ChocolateArmpits> through ffmpeg
[21:37:50 CET] <DragonsLordKERNE> -sc_threshold 1000000000 is to disable scene detection?
[21:38:19 CET] <ChocolateArmpits> Nah it's there to please the unhappy encoder "[mpeg4 @ 0000000002dbcea0] closed gop with scene change detection are not supported yet, set threshold to 1000000000"
[21:38:31 CET] <ChocolateArmpits> so yes
[21:39:30 CET] <DragonsLordKERNE> I will also do some tests, thank you very much for your effort
[21:39:36 CET] <ChocolateArmpits> np
[21:39:42 CET] <DragonsLordKERNE> do you know if I can also disable null frames?
[21:39:55 CET] <ChocolateArmpits> as I said yesterday I have no idea about those
[21:40:02 CET] <DragonsLordKERNE> OK, sorry
[21:40:22 CET] <DragonsLordKERNE> 'cause I have something like one frame of desync between audio and video
[21:40:28 CET] <DragonsLordKERNE> and I cannot understand why
[21:40:59 CET] <jorb> so i'm trying to concat several videos
[21:41:19 CET] <jorb> do i need to normalize all of the clips to the same bit rate or something? output isn't working :(
[21:41:53 CET] <ChocolateArmpits> jorb: are you concating via a text file ?
[21:41:58 CET] <ChocolateArmpits> concatenating *
[21:42:12 CET] <DragonsLordKERNE> I did like ChocolateArmpits is telling to you
[21:42:19 CET] <DragonsLordKERNE> and it worked finem, for me
[21:42:31 CET] <DragonsLordKERNE> also with not normalized sources
[21:43:11 CET] <DragonsLordKERNE> ChocolateArmpits, do you know if the options, suggested by you should work also with XviD?
[21:43:15 CET] <sleblanc> Is there a way to query ffprobe for the type of metadata header contained in the file? e.g. Vorbis tags or ID3Vx
[21:43:16 CET] <jorb> ChocolateArmpits: yes.. i'm able to generate output using the text file list of videos (demuxer), but i also tried the video filter method and it wouldn't generate any output
[21:43:25 CET] <jorb> http://stackoverflow.com/a/11175851
[21:44:40 CET] <jorb> i tried all 3 of those methods without much luck
[21:44:40 CET] <ChocolateArmpits> DragonsLordKERNE: if you mean the libxvid encoder, they won't
[21:46:56 CET] <StephenS> hmm shell_exec("{$ffmpeg} -i \"{$uploadedfiletmp}\" -i watermark.png -filter_complex \"overlay=x=(main_w-overlay_w)-10:y=(main_h-overlay_h)-10\" -strict -2 {$uploadedfile}" gives me very bad video quality
[21:47:00 CET] <StephenS> how to preserve quality?
[21:47:04 CET] <ChocolateArmpits> jorb: So if one of them is working, why not just use it? the concat protocol has a limited number of format it supports, the video filter probably have inputs/outputs mistyped
[21:47:39 CET] <jorb> ChocolateArmpits: the demuxer method will generate output, but only the first video plays, the other streams get corrupted somehow
[21:47:51 CET] <jorb> the other methods aren't working at all
[21:48:07 CET] <ChocolateArmpits> jorb: what are the input video files ?
[21:48:44 CET] <jorb> their all H.264 mpeg uhhh
[21:48:47 CET] <jorb> .mp4
[21:48:48 CET] <ChocolateArmpits> StephenS: through more bitrate at it
[21:49:05 CET] <jorb> with aac audio as well
[21:49:06 CET] <StephenS> ChocolateArmpits, such as?
[21:49:09 CET] <ChocolateArmpits> jorb: well yeah but what are the video formats
[21:49:25 CET] <ChocolateArmpits> StephenS: the video gets reencoded, you need to specify encoder settings
[21:49:30 CET] <jorb> libavformat
[21:49:52 CET] <ChocolateArmpits> jorb: no I mean what is the stream format, resolution, profile, level
[21:51:08 CET] <jorb> wStream #0:0(und): Video: h264 (avc1 / 0x31637661), 720x480, 491 kb/s, SAR 8:9 DAR 4:3, 23.98 fps, 23.98 tbr, 24k tbn, 48k tbc
[21:51:20 CET] <jorb>     Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 720x480 [SAR 1:1 DAR 3:2], 1661 kb/s, 15 fps, 15 tbr, 15360 tbn, 30 tbc
[21:51:51 CET] <jorb>     Stream #0:1(eng): Audio: aac (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 128 kb/s
[21:51:58 CET] <jorb>     Stream #0:1(eng): Audio: aac (mp4a / 0x6134706D), 16000 Hz, mono, fltp, 21 kb/s
[21:52:20 CET] <StephenS> well ChocolateArmpits ffmpeg -i ham.webm -i main_overlay.png -codec:v libx264 -crf 18 -preset slow -pix_fmt yuv420p -filter_complex "overlay=x=(main_w-overlay_w)-10:y=(main_h-overlay_h)-10" -strict -2  out.mp4
[21:52:31 CET] <StephenS> that works, but I get out.mp4 twice in size from the original video
[21:53:38 CET] <ChocolateArmpits> jorb: Just look at that, different framerates
[21:54:28 CET] <ChocolateArmpits> StephenS: Then try manually setting the bitrate that would match the original video, -b
[21:57:08 CET] <ChocolateArmpits> jorb: the protocol most likely doesn't work because of different framerates and the aspect ratio; the demuxer gets slightly further but fails because of the same reason too if you're trying to copy the streams. You should try using demuxer AND reencoding also
[21:57:54 CET] <ChocolateArmpits> the perfect formula to making concatenation work is having matching file formats
[21:58:20 CET] <jorb> yeah.. basically i'll need to normalize them before hand
[21:58:36 CET] <jorb> i only got as far as setting the bigger aspect ratio down to the 720x480 display
[21:59:05 CET] <jorb> basically i'm taking a cell phone video and trying to concat a 1920x1080 movie clip
[22:02:00 CET] <DragonsLordKERNE> <ChocolateArmpits> DragonsLordKERNE: if you mean the libxvid encoder, they won't | Yes, I mean libxvid ... I am trying to understand how it works with avidemux + libxvid :|
[22:06:22 CET] <waressearcher2> isn't having "mencoder" and "ffmpeg" is like reinventing the wheel ?
[22:10:28 CET] <StephenS> ChocolateArmpits, video quality chanages
[22:10:35 CET] <StephenS> changes* I need some way to use it constantly
[22:11:07 CET] <ChocolateArmpits> StephenS: I don't understand your problem, do you have video quality problems after encoding ? What are your requirements ?
[22:11:20 CET] <StephenS> ChocolateArmpits, to add watermark and preserve video quality
[22:12:26 CET] <ChocolateArmpits> StephenS: what's the bitrate ?
[22:12:31 CET] <ChocolateArmpits> or the original video
[22:13:07 CET] <StephenS> its volatile
[22:13:50 CET] <jorb> ChocolateArmpits: i got the video sync'd up by running both the videos through the same -vf
[22:13:57 CET] <jorb> i'm confused on what to do for audio though
[22:14:10 CET] <jorb> do i just need to set the same bit rate or are there some other options?
[22:14:34 CET] <StephenS> but  bitrate: 829 kb/s
[22:14:36 CET] <furq> they probably need to have the same samplerate and number of channels
[22:14:42 CET] <StephenS> but thats volatile as I said
[22:14:57 CET] <furq> StephenS: just use a higher crf value
[22:15:12 CET] <ChocolateArmpits> StephenS: ok so if crf 18 produces too big of a file, try 21. General rule is that every 6 steps the bitrate doubles or halves. You're reencoding so you can't have the same bitrate anyways to preserve some semblance of original quality
[22:16:07 CET] <ChocolateArmpits> jorb: do this for the second audio stream " -ar 48000 -ac 2 -b:a 128k "
[22:16:29 CET] <StephenS> crf 21
[22:16:36 CET] <StephenS> gives me the same size
[22:16:46 CET] <ChocolateArmpits> And worse quality ?
[22:17:11 CET] <ChocolateArmpits> Then try 19, 20, it's all about poking to find the right thing for you
[22:18:22 CET] <StephenS> ChocolateArmpits, no quality is good, but size is too big
[22:18:41 CET] <ChocolateArmpits> StephenS: ok then go lower, 24,25
[22:19:24 CET] <StephenS> hmm well for a eye its the same quality, but is there a way to measure it?
[22:19:48 CET] <furq> not reliably
[22:19:52 CET] <DragonsLordKERNE> <ChocolateArmpits> DragonsLordKERNE: if you mean the libxvid encoder, they won't | Yes, I mean libxvid ... I am trying to understand how it works with avidemux + libxvid :|
[22:20:00 CET] <DragonsLordKERNE> ChocolateArmpits, sorry to bother you, once more
[22:20:18 CET] <DragonsLordKERNE> so there are no equivalent setting within libxvid encoder?
[22:20:19 CET] <furq> psnr and ssim exist for that purpose, but they're not that useful with x264
[22:20:27 CET] <DragonsLordKERNE> (with FFmpeg, at least?)
[22:20:55 CET] <ChocolateArmpits> DragonsLordKERNE: what I said was that I wasn't able to get fixed gop with libxvid
[22:21:26 CET] <ChocolateArmpits> DragonsLordKERNE: however mpeg4 produces a stream under same standard that libxvid would produce, so technically it should be an acceptable solution for you
[22:21:55 CET] <ChocolateArmpits> DragonsLordKERNE: considering it gets the fixed gop right
[22:22:34 CET] <DragonsLordKERNE> avidemux seems to work with MPEG4 ASP (xvid)
[22:26:53 CET] <DragonsLordKERNE> ChocolateArmpits, do you know if -gop and -g for libxvid is basically the same?
[22:27:18 CET] <ChocolateArmpits> DragonsLordKERNE: I have never seen the first one used, check codec documentation
[22:27:34 CET] <ChocolateArmpits> it's not listed under libxvid last I looked
[22:34:31 CET] <toriso> it
[22:52:14 CET] <InTheWings> Anyone aware of an ismv/smooth encoder producing invalid offsets ?
[22:53:05 CET] <InTheWings> implicit being 1st byte of tfhd instead of moof
[22:54:42 CET] <huglester> Guys, if some of you worked with nginx-rtmp... I use push, to push stream to /dash application. When I stream simple 'movie', it works ok, but when my device connects and publishes, dash files are not created
[22:54:50 CET] <huglester> if I try to play dash stream via ffplay - it works
[23:50:46 CET] <debianuser> Hello, a short question: how can I set "default" flag for some mkv audio track when doing something like `ffmpeg -i 1.mkv -i 2.ac3 -map 0:0 -map 1:0 -map 0:1 -c copy all.mkv`?
[23:59:35 CET] Action: debianuser found how to do that with mkvtoolnix-gui/mmg but still wonders how to do that with ffmpeg...
[00:00:00 CET] --- Mon Nov  2 2015