[Ffmpeg-devel-irc] ffmpeg.log.20150623

burek burek021 at gmail.com
Wed Jun 24 02:05:01 CEST 2015


[01:12:57 CEST] <aib> I'm splitting and joining a video using segment and concat, without any problems. However, if I process the segments (ffmpeg -i seg01.mp4 -map 0:v -map 0:a oseg01.mp4) they don't join properly; I get audio stutters at segment boundaries. any ideas?
[01:16:54 CEST] <aib> looks like when I do "fmpeg -i seg01.mp4 -map 0:v -map 0:a oseg01.mp4", the output video has an extra frame at the beginning and is missing two at the end
[01:17:02 CEST] <aplund> Using idet I get this output from what is meant to be an interlaced mpeg2 stream.  How on earth is this meant to make sense?  Is it TFF or BFF? http://pastie.org/10253960
[01:17:44 CEST] <c_14> aplund: TFF
[01:18:23 CEST] <aplund> Why so many progressive frames?  It's full of movement... and when I play it, it hardly looks interlaced at all.
[01:19:06 CEST] <c_14> hmm, wait
[01:19:11 CEST] <c_14> didn't even look at the progressive count
[01:19:11 CEST] <aplund> it's a very strange.  It's from dvb in Australia.
[01:21:04 CEST] <c_14> aplund: might be a telecine pattern
[01:22:35 CEST] <aplund> c_14: don't know if it helps but here is the stream info: http://pastie.org/10253965
[01:23:37 CEST] <aplund> my guess is that if tbc is double fps then it's interlaced
[01:23:58 CEST] <aplund> but somehow idet is confused and reports lots of progressive frames.
[01:24:24 CEST] <aplund> Perhaps the broadcasater does something weird to have interlaced streams appear progressive
[01:25:11 CEST] <aplund> (not that I have any idea what tbr, tbn and tbc actually mean)
[01:25:23 CEST] <c_14> You could try something like -vf idet,fieldmatch,yadif=deint=interlaced,decimate
[01:25:35 CEST] <c_14> You might have to framestep through the video to check if it's doing what it's supposed to
[01:36:58 CEST] <aplund> c_14: Lots of output from fieldmatch like :"Frame #216 at 15288.4 is still interlaced"
[01:37:15 CEST] <c_14> That's what the yadif is for
[01:46:46 CEST] <aplund> Is there any way to get ffplay to give you the frame number you are on?
[01:47:15 CEST] <c_14> None I know of.
[07:19:33 CEST] <sanjosekid_> libx264 is encoding our 320x240 frames at 1 Mbps!  our input is realtime and low bandwidth.  does anyone know what settings would produce lower bitrate?
[07:23:45 CEST] <kepstin> probably setting a lower bitrate would do it.
[07:24:43 CEST] <kepstin> if you're doing a "crf" encode and don't want to lower the quality more, try increasing the keyframe interval or using a slower preset?
[07:25:14 CEST] <kepstin> can't provide any real hints unless I know what options you're currently using.
[08:23:53 CEST] <sanjosekid_> thanks @kepstin
[08:25:33 CEST] <sanjosekid_> @kepstin, we were thinking about 1) removing intra-refresh, 2) turning on error-concealment on decoder side, 3) we cannot use slower preset, our device has limited CPU
[08:26:26 CEST] <sanjosekid_> we were going to check CFR based encoding, with bitrate-limit of 400 kbps.  we relaxed CBR.
[08:30:02 CEST] <sanjosekid_> our x264 is 1.5 years old - not sure if that might be an issue, i doubt it
[08:30:43 CEST] <sanjosekid_> our settings are: av_dict_set(&dict.ptr, "g",      "250",          0);    // Default 15     av_dict_set(&dict.ptr, "vprofile", "high",        0);     av_dict_set(&dict.ptr, "profile", "high",        0);     av_dict_set(&dict.ptr, "tune",    "zerolatency", 0); //animation+zerolatency      av_dict_set(&dict.ptr, "preset",  "veryfast",  0);     av_dict_set(&dict.ptr, "threads", "4",          0);      av_dict_set(&dict.ptr, "slice-m
[11:38:23 CEST] <gurpartap1> Is there any facility in ffmpeg that it drop same number of audio packets to synchronize with video........
[14:44:30 CEST] <DRAWNBOX> Hello
[14:45:27 CEST] <DRAWNBOX> When I type in ffmpeg -i "PNG's%04d.png" -
[14:45:27 CEST] <DRAWNBOX> r 60 -vcodec libx264 -crf 15 "Build The Cities Mashup.mp4" it starts to duplicate frames in the render
[14:45:44 CEST] <DRAWNBOX> E:\Blender\Visualizer\Build The Cities Mashup\Output>ffmpeg -i "PNG's%04d.png" -
[14:45:44 CEST] <DRAWNBOX> r 60 -vcodec libx264 -crf 15 "Build The Cities Mashup.mp4"
[14:46:11 CEST] <DRAWNBOX> I really dont know whats wrong with that
[14:46:56 CEST] <DRAWNBOX>  
[14:47:06 CEST] <DRAWNBOX>  http://pastebin.com/PJyGbker
[14:47:12 CEST] <DRAWNBOX> Thats my console
[14:48:49 CEST] <DHE> it thinks the input framerate is 25 and you're asking it resample to 60fps
[14:48:55 CEST] <DHE> you need to make it think the input is 60fps instead
[14:49:25 CEST] <DRAWNBOX> yep
[14:49:52 CEST] <DRAWNBOX> wait, I cant read
[14:49:57 CEST] <DRAWNBOX> How do I do that?
[14:50:29 CEST] <DRAWNBOX> - framerate?
[14:50:39 CEST] <Mavrik> DRAWNBOX, the order is important
[14:50:49 CEST] <DRAWNBOX> oh, i see what I need to
[14:50:50 CEST] <Mavrik> ffmpeg -r 50 -i input -r 25
[14:51:04 CEST] <Mavrik> means "interpret input as 50fps and then convert it to 25 fps"
[14:51:10 CEST] <Mavrik> so if your input frames are at 60fps
[14:51:17 CEST] <Mavrik> move -r 60 before -i :)
[14:54:45 CEST] <relaxed> it's -framerate
[14:55:06 CEST] <relaxed> although -r will probably work too
[14:55:48 CEST] <DRAWNBOX> Now, I get something about past duration to large
[14:56:19 CEST] <relaxed> pastebin your command and output
[14:57:11 CEST] <DRAWNBOX> http://pastebin.com/yqLSzP30
[14:57:44 CEST] <relaxed> now try with -framerate 60
[14:58:11 CEST] <DRAWNBOX> that worked
[14:58:46 CEST] <DRAWNBOX> Im pretty sure it did anyway
[14:58:56 CEST] <DRAWNBOX> Once its done I will see if it looks right
[15:08:23 CEST] <DRAWNBOX> Well, it worrked
[16:19:33 CEST] <BurningCa007> Hello! Just wanted to ask a quick question, can ffmpeg get me raw audio samples from .mp3 files via C++ easily?
[16:20:38 CEST] <BtbN> yes, if that's all you want, it's simple enough
[16:21:11 CEST] <BurningCa007> Nice, thanks BtbN, care to point me in the general direction in the doc?
[16:21:31 CEST] <BtbN> https://github.com/FFmpeg/FFmpeg/tree/master/doc/examples
[16:22:00 CEST] <BtbN> demuxing_decoding.c sounds most fitting for what you want
[16:22:51 CEST] <BurningCa007> what about decoding_encoding.c
[16:23:11 CEST] <BtbN> don't you want raw samples?
[16:23:20 CEST] <BurningCa007> I do, but audio only?
[16:23:31 CEST] <BtbN> So there is nothing to encode
[16:31:16 CEST] <BurningCa007> BtbN: forgive me for asking some more, but is there a specific array that buffers the samples in the demuxing_decoding.c source?
[16:34:17 CEST] <BurningCa007> So av_read_frame() gets me packets and then I call avcodec_decode_audio4 on them?
[16:34:42 CEST] <BtbN> basicaly, yes
[16:35:23 CEST] <BurningCa007> but avcodec_decode_audio4 returns the number of bytes consumed?
[16:36:53 CEST] <BtbN> it returns a statis code
[16:37:11 CEST] <BtbN> and gives you an AVPacket
[16:40:17 CEST] <BurningCa007> uh doc says it stores the decoded data in the area pointed to by an AVFrame
[16:40:29 CEST] <BurningCa007> I presume AVFrame::data is the array I'm looking for?
[16:43:04 CEST] <BurningCa007> BtbN?
[16:47:30 CEST] <BurningCa007> frame->extended_data got it, thanks.
[17:19:05 CEST] <Mavrik> O.o
[19:24:40 CEST] <zipped> hey there guys . I'm trying to capture an entire screen with the following command but it only captures half of the screen , any ideas why ? I'm freaking out here :) :
[19:24:40 CEST] <zipped> ffmpeg -f alsa -ac 2 -i pulse -f x11grab -r 15 -s 1024x768 -i :0.0+448,156  -acodec pcm_s16le -vcodec libx264 -preset ultrafast -threads 0  output.avi
[19:26:27 CEST] <DHE> why are you specifying offsets if you want the whole screen?
[19:26:50 CEST] <zipped> I tired w/o them as well ... didn't work
[19:27:07 CEST] <zipped> your mean the +448,156 , right ?
[19:27:40 CEST] <zipped> basically I just need a high quality screen capture with audio and frate of 15
[19:30:37 CEST] <zipped> ok I think I got it now , but w/o the audio . could you help me please with adding the audio ?
[19:30:45 CEST] <zipped> ffmpeg -f x11grab -s 1920x1080 -r 15 -i :0.0 -qscale 0 -vcodec huffyuv output.avi
[19:40:27 CEST] <zipped> DHE: ^
[19:41:10 CEST] <c_14> zipped: readd the -i pulse?
[19:41:34 CEST] <zipped>  ffmpeg -f x11grab -s 1920x1080 -r 15 -i :0.0 -qscale 0 -vcodec huffyuv -f pulse -ac 2 -i default output.avi
[19:41:42 CEST] <zipped> I tried it like this but didn't work
[19:41:49 CEST] <zipped> I must have got something wrong
[19:42:26 CEST] <c_14> What is your default pulse capture device? Can you listen to it with any other program?
[19:43:23 CEST] <zipped> yeah well , when I ran it like this it did work :ffmpeg -video_size 1024x768 -framerate 25 -f x11grab -i :0.0+100,200 -f pulse -ac 2 -i default output.avi   but then I had the video capturing only half of the screen
[19:44:57 CEST] <zipped> c_14: so I guess that default device is ok , I just don't use the command properly
[19:45:31 CEST] <c_14> eh, oh
[19:45:40 CEST] <c_14> move -qscale 0 -vcodec huffyuv after -i default
[19:47:21 CEST] <relaxed> huffyuv is lossless, so -qscale 0 does nothing unless you meant -q:a 0
[20:10:56 CEST] <zipped1> c_14: perfect !! that did it :)) thanks a lot!!
[20:38:15 CEST] <fabHK> when I used an output name of something.m4a, the file was not recognised by iTunes. Neither was .aac. However, output format .mp4 is recognised by iTunes. Yet, it doesn't seem to allow to remember the last playback position.
[20:38:28 CEST] <fabHK> any suggestions? (sorry for delay, my VPN was acting up)
[21:26:08 CEST] <fabHK> ah, looks like iTunes doesn't support AAC with the HE or HE-v2 profile.... dang, I really wanted to use it for 24kb/s voice compression...
[00:00:00 CEST] --- Wed Jun 24 2015


More information about the Ffmpeg-devel-irc mailing list