[Ffmpeg-devel-irc] ffmpeg.log.20160802

Wed Aug 3 03:05:01 EEST 2016

[00:08:07 CEST] <wallbroken> jkqxz, so, i do it with 32bit version?
[02:18:47 CEST] <ycon_> Hi all, I need to re-encode my audio of my mp4 video to be Mp3 (as its the only format my vid supports)
[02:19:15 CEST] <ycon_> I ran this- but it didn't work ffmpeg -i Video.mpt -acodec libmp3lame audioreformat.mp4
[02:19:23 CEST] <ycon_> *ffmpeg -i Video.mp4 -acodec libmp3lame audioreformat.mp4
[02:20:50 CEST] <furq> what does "didn't work" mean
[02:21:05 CEST] <furq> also you probably want -c:v copy or else it'll reencode the video
[02:27:34 CEST] <ycon_> furq: didnt work, as in it created a file that was cut short of the total time
[02:27:49 CEST] <ycon_> May have been my computer stuffing up ha. So where do I use -c:v copy? LIke this?
[02:27:50 CEST] <ycon_> ffmpeg -i Slide_Downstairs_2.m4v -c:v copy -acodec libmp3lame mp3.mp4
[02:27:56 CEST] <furq> yes
[04:01:03 CEST] <sauh97> Hi! I'm converting videos to play on my phone, and I'd like to scale and crop them in such a way to fix my screen as best as possible.
[04:02:14 CEST] <sauh97> I'm using "scale=w=427:h=240:force_original_aspect_ratio=decrease" but I'd like ffmpeg to crop the video in such a way so there aren't any 'black bars' when I play it back on my phone
[04:16:08 CEST] <sauh97> Found solution: -vf "crop=in_h*16/9:in_h,scale=-1:240"
[04:16:13 CEST] <sauh97> Thanks anyhow :)
[04:17:21 CEST] <xpack> I keep getting errors when compiling ffmpeg following the guide
[04:17:38 CEST] <xpack>    ffmpeg/libavcodec/x86/dirac_dwt_init.c:140: undefined reference to `ff_vertical_compose53iL0_sse2'
[04:17:53 CEST] <xpack> collect2: error: ld returned 1 exit status make: *** [ffmpeg_g] Error 1
[04:20:39 CEST] <furq> xpack: build with --disable-decoder=dirac
[04:20:57 CEST] <furq> also maybe submit a bug report, this is the second time i've seen this reported
[04:43:24 CEST] <xpack> Thanks @furq that fixed the compilation error
[09:20:45 CEST] <seanrdev> I am attempting to merge multiple avi video files. I keep getting timestamp errors. Any idea where to start looking?
[10:53:49 CEST] <spirou> good morning
[10:54:47 CEST] <Fjorgynn>  /me is here
[11:00:26 CEST] <spirou> if I want to save a graphical subtile to a separate file what shall I cal the out filenamne?
[11:01:18 CEST] <spirou> ffprobe don't really list it it though (but perhaps that is "Stream #0:0[0x1bf]: Data: dvd_nav_packet"?), vlc say "DVD Subtitles (spu )"
[11:03:25 CEST] <spirou> the film is in a .iso file (file says "UDF filesystem data (version 1.5)", mediainfo says "Format: ISO 9660")
[14:50:00 CEST] <kdehl> What is IDC (or idc?) in H.264? I see it everywhere, but I can't find a description of it.
[14:57:43 CEST] <bencoh> kdehl: IDR?
[14:57:46 CEST] <jkqxz> As in profile_idc or chroma_format_idc?  I don't know what it stands for, but it marks syntax elements in the bitstream which take more than two discrete values.  (Elements taking only two values are "flag", elements with ranges don't get a suffix.)
[14:57:50 CEST] <bencoh> or IDC?
[14:59:43 CEST] <kdehl> Yeah, I meant idc.
[14:59:54 CEST] <kdehl> jkqxz: Alright, thanks.
[15:00:09 CEST] <kdehl> Weird that you can't find what it stands for though.
[15:00:24 CEST] <kdehl> At least I know what it means now.
[15:23:46 CEST] <zuloyd> hi
[15:24:56 CEST] <zuloyd> is it possible to distribute encoding of a live stream onto multiple servers in the network?
[15:26:18 CEST] <DHE> yes, but you might want to use something better suited than ffmpeg. an outage will likely jam up the whole system
[15:26:29 CEST] <zuloyd> any recommendations?
[16:03:02 CEST] <hero_biz> hi guys.
[16:03:25 CEST] <hero_biz> I have a small question which answer is probably no, but just want to make sure.
[16:04:11 CEST] <hero_biz> if we encode an mkv with ffmpeg which has ordered chapters, is it possible to merges chapters while encoding or not?
[16:17:26 CEST] <c_14> no, ffmpeg does not (as of yet) support ordered chapters
[16:20:05 CEST] <hero_biz> reordering is super irritating.
[16:21:48 CEST] <c_14> if mpv (which has encoding support) does everything you want you could try using that. It has it's own mkv demuxer (which supports ordered chapters)
[16:23:54 CEST] <hero_biz> I need a console encoder. I think think I could use mpv
[16:25:57 CEST] <furq> that seems like a weird thing for mpv to support
[16:26:02 CEST] <furq> are people pining for the good old days of mencoder
[16:26:51 CEST] <hero_biz> hm..
[16:27:28 CEST] <hero_biz> I still don't know why process that I use for reordering chapters will fails sometimes.
[16:28:02 CEST] <hero_biz> I encode all parts with same settings and add I frames to main video where I should place chapters.
[16:28:18 CEST] <hero_biz> then I split main video on those I frames.
[16:28:34 CEST] <hero_biz> and finally merges splits and chapter files together.
[16:28:55 CEST] <hero_biz> I think think process should not mess up things.
[16:29:58 CEST] Action: hero_biz thinks maybe a faulty subtitle mess up everything????
[16:45:40 CEST] <kdehl> Um. Probably a silly question, but about the Emulation Prevention Bytes in, H.264. How does a decoder know that it _is_ an emulation prevention byte, and not just valid encoded video data?
[17:22:41 CEST] <PhoenixBR> Hi there guys!
[17:26:52 CEST] <PhoenixBR> A simple question: which ffmpeg command can I use to transform a .mp4 video (1280x720, 9990kbps, 30fps) to a .mp4 video with the half of this size and bitrate, reducing the size of the file?
[17:50:52 CEST] <intracube> PhoenixBR: do you mean half resolution?
[17:55:23 CEST] <PhoenixBR> intracube: yeah, could be
[17:55:32 CEST] <intracube> ffmpeg -i input.mp4 -b:v 5000k output.mp4
[17:56:06 CEST] <PhoenixBR> intracube: and the half size?
[17:56:10 CEST] <intracube> ffmpeg -i intput.mp4 -b:v 3000k -vf scale=640:360 output.mp4
[17:56:24 CEST] <intracube> you can get away with a lower bitrate since it'd be quarter the area
[17:56:28 CEST] <PhoenixBR> intracube: perfect! let me try
[17:56:48 CEST] <BtbN> 3kbps is quite excessive for 360p
[17:56:56 CEST] <BtbN> 3000
[17:57:35 CEST] <PhoenixBR> BtbN: which resolution do I have try?
[17:57:37 CEST] <intracube> depends what quality you're going for, but yeah, assuming h.264, it could be 2000k and still fairly good quality
[17:59:51 CEST] <intracube> the audio will also be re-encoded. instead you might want to copy from the original file as-is
[18:00:14 CEST] <intracube> ffmpeg -i intput.mp4 -c:a copy -b:v 2000k -vf scale=640:360 output.mp4
[18:15:55 CEST] <PhoenixBR> intracube BtbN thanks guys!
[18:16:01 CEST] <intracube> np
[18:16:02 CEST] <PhoenixBR> it worked!
[18:17:52 CEST] <PhoenixBR> A more advanced question: I generated a green screen video: ffmpeg -f lavfi -i color=green:s=704x576:r=24000/1001 -f lavfi -i anullsrc -ac 2 -t 1 greenscreen.ts
[18:18:55 CEST] <PhoenixBR> then I concatenated with a .ts video like this: ffmpeg -i greenscreen.avi -i 1.avi -filter_complex "[0:v:0] [1:v:0] [0:v:0] concat=n=3:v=1 [v]" -map "[v]" -pix_fmt yuv420p -vcodec rawvideo output.avi
[18:19:17 CEST] <PhoenixBR> ops, sorry, .ts version: ffmpeg -i greenscreen.ts -i 1.ts -filter_complex "[0:v:0] [1:v:0] [0:v:0] concat=n=3:v=1 [v]" -map "[v]" -pix_fmt yuv420p -vcodec rawvideo output.ts
[18:20:00 CEST] <PhoenixBR> that means, 1 second of the greenscreen video before and after the original video 1.ts
[18:22:10 CEST] <PhoenixBR> the output.ts video do not work
[18:22:35 CEST] <PhoenixBR> unless i take off the -vcodec rawvideo parameter
[18:23:13 CEST] <PhoenixBR> *remove this parameter
[18:23:22 CEST] <PhoenixBR> What I am doing wrong?
[18:24:39 CEST] <PhoenixBR> But I really need the original video quality in the output video
[18:24:47 CEST] <intracube> does .ts stream format support raw video?
[18:24:50 CEST] <intracube> I'm not familiar with the filter_complex syntax
[18:27:15 CEST] <Kadigan> Here's a question: I have a video that's exactly 45 seconds long. I run 'ffmpeg -i video.mov -c copy -ss 15.00 -vframes 375 middle15s.mov' and everything is fine for video, but I get a little bit of the audio from beyond the 375 frames.
[18:27:24 CEST] <intracube> I was going to suggest using -c:v copy to copy over the raw stream, but I'm not sure if that works with the concat option
[18:27:29 CEST] <Kadigan> This happens when I specify v/a codecs as well. (the rate is 25fps) Ideas?
[18:28:45 CEST] <Kadigan> (and I know I get audio 'from beyond', because the audio lasts for 365 frames (10f silence, 365f audio), so even if it 'overshot' by a frame or two I should be in the 10f of silence, yet I get about 15f of audio from the next segment, ie. 375 video frames + 390 audio frames)
[18:35:42 CEST] <PhoenixBR> intracube: another way to connect this .ts videos?
[18:36:10 CEST] <intracube> PhoenixBR: I never found an easy way with ffmpeg
[18:37:01 CEST] <PhoenixBR> intracube: do you recommend another toll?
[18:37:07 CEST] <PhoenixBR> tool*
[18:38:01 CEST] <intracube> I can't think of one atm
[18:38:39 CEST] <PhoenixBR> intracube: it's ok, thanks!
[18:42:28 CEST] <intracube> PhoenixBR: just tried the first command here and it seems to work: https://trac.ffmpeg.org/wiki/Concatenate#demuxer
[18:43:15 CEST] <intracube> I didn't add the full folder structure to the file list since it complained about an 'unsafe filename'
[18:43:38 CEST] <intracube> file 'clip_01.ts'
[18:43:43 CEST] <intracube> file 'clip_02.ts'
[18:43:46 CEST] <intracube> etc
[18:44:11 CEST] <intracube> txt file in the same folder as the video clips
[18:45:01 CEST] <average> is there an ffmpeg planet ? (news aggregator)
[18:45:08 CEST] <average> where do people get their ffmpeg news/posts ?
[18:45:41 CEST] <average> like for example if I'm into ffmpeg, where do I go to get my fix ?
[18:48:09 CEST] <bencoh> average: I just follow both irc channels and do git fetch when I feel like it (or have a specific need)
[18:48:31 CEST] <average> bencoh: that's fine and reasonable
[18:48:37 CEST] <bencoh> you can check the mailing-lists as well
[18:48:42 CEST] <average> but supposing I'm more at the user-level
[18:49:00 CEST] <average> the internals level is a bit of a different space
[18:49:19 CEST] <PhoenixBR> intracube: but have you tried with .ts files?
[18:50:01 CEST] <bencoh> average: rss and/or mailing-list then?
[18:50:42 CEST] <intracube> PhoenixBR: yep
[18:50:55 CEST] <intracube> the file plays ok in VLC, but not mplayer for some reason
[18:50:55 CEST] <PhoenixBR> intracube: let me try it
[18:54:16 CEST] <average> bencoh: rss would be great, like you know how many large OSS projects have a planet
[18:54:50 CEST] <average> bencoh: I mean a news aggregator (it just so happens that the software running that aggregator is called planet or venus http://www.planetplanet.org/ ; http://www.intertwingly.net/code/venus/ )
[18:55:17 CEST] <average> bencoh: like if there was something like that, that would be a one-stop-shop for ffmpeg users to get their fix
[18:56:05 CEST] <bencoh> I doubt there is
[18:56:59 CEST] <alyawn> I'm attempting to stream an mpegts source to an rtsp output and I'm consistently getting a crash with the newer ffmpeg (3.1.1 and current HEAD). I can seem to figure out if this is a known issue or not. Are any of you seeing this issue?
[18:57:41 CEST] <alyawn> http://pastebin.com/fz7JyMD3
[18:59:27 CEST] <bencoh> alyawn: I guess you could try and bissect, since you already know how to build from git :)
[19:00:15 CEST] <alyawn> sure thing bencoh
[20:25:32 CEST] <NetworkingPro> hey everyone
[20:25:53 CEST] <NetworkingPro> Anyone know where I could find a doc with examples of transcoding and RTSP stream to mjpeg?
[20:26:04 CEST] <NetworkingPro> Ive googled and cant seem to find it.
[20:37:51 CEST] <DHE> if you can find an example of rtsp in, and an example of mjpeg out, you can just stitch them together most of the time
[20:46:18 CEST] <Guest56501> I want a copy of the images in a video - i think video images are bmp? anyway I need the best (copy) of EVERY image. is this the best way to do it?  ffmpeg -i *.mp4 -q:v 1  output_%04d.bmp  ffmpeg -i *.mp4 -q:v 1  output_%04d.jpg  ffmpeg -i *.mp4 -q:v 1  output_%04d.png
[20:48:39 CEST] <DHE> there is no quality option for png, png is lossless
[20:57:58 CEST] <Guest56501> ok but the rest is ok? it will give best quality?
[21:31:38 CEST] <kdehl> Um. So. A question. The scale2ref filter, why does it take two outputs? I'm just interested in the rescaled video stream, not the stream that I used as a size reference...
[21:32:34 CEST] <kdehl> I must have misunderstood something.
[21:34:46 CEST] <kdehl> This works:
[21:34:47 CEST] <kdehl> ffmpeg -i left.mp4 -i right.mp4 -y -an -filter_complex "[0]pad=w=in_w*2[padded]; [1][0]scale2ref[0][right_scaled]; [padded][right_scaled]overlay=W/2" output.avi
[21:35:40 CEST] <kdehl> But I'm not sure how. That first output connection (or whatever you call it) of scale2ref, why is that 0? That refers to left.mp4, does it not?
[22:09:43 CEST] <brontosaurusrex> how do i say: -vf "crop to height if height<width else reverse" ?
[22:35:46 CEST] <brontosaurusrex> cough
[23:41:06 CEST] <mikjaer> I am recording a video stream with a magewell hdmi->usb adapter using ffmpeg, but when i take out the hdmi cabel and reinsert it (or another on) while the recording is running ffmpeg freezes the last frame and keeps recording the sound only ... how do i proceed debugging this?
[00:00:00 CEST] --- Wed Aug  3 2016