[Ffmpeg-devel-irc] ffmpeg.log.20141219

[burek]

[00:14] <mac_nibblet> Trying to compile ffmpeg
[00:14] <mac_nibblet> but it won't detect zlib on ubuntu 14.04
[00:14] <mac_nibblet> install all the requried zlib packages
[00:37] <beastd> mac_nibblet: do you have zlib development files installed (debian based distros usually have dev in a separate package)
[04:15] <shevy> is there a simple and generic command to tell ffmpeg to extract the audio format from e. g. a .mp4 or .avi file?
[04:32] <pzich> shevy: you can use -map 0:a to grab just the audio, but you need to specify the correct file extension if you don't want to transcode
[07:50] <solarant> Hi
[07:51] <solarant> What amongst the three will make ffmpeg faster?
[07:51] <solarant> a faster processor, more ram, a powerful graphic card
[07:52] <solarant> I am currently generating previews of videos using ffmpeg on a ubuntu machine with a i5 processor and 8gb ram
[07:52] <solarant> and built in graphic card
[07:58] <ubitux> better cpu
[08:04] <solarant> thank you ubitux
[08:04] <solarant> will get the cpu upgraded
[08:04] <solarant> thanks
[08:05] <solarant> ubitux:  will I need more ram if I am decoding files that are 50 Gb to 300 GB in size
[08:06] <ubitux> you will need more ram only if you are processing using complex filter graph in which there is a need for caching frames
[08:09] <solarant> ubitux:  I dont think I am doing that
[08:09] <solarant> I am using resource space and it only creates a preview
[08:09] <ubitux> what's your need exactly?
[08:09] <solarant> converts 30gb to 300 mb
[08:10] <ubitux> from what codec to what codec?
[08:10] <solarant> I am using a DAM that uses ffmpeg to convert .mov files to flv
[08:10] <ubitux> what codecs?
[08:17] <wh-hw> hi, all , i add a function to vf_delogo.c , and declare it in avfilter.h , but when i use it on my code , i link it
[08:17] <wh-hw> can not link it ,     undefined reference to
[08:30] <solarant> ubitux:  the DAM does not and I am not sure what is the exact command it uses
[08:30] <solarant> ubitux:  is there any other wayto find out
[08:30] <ubitux> i don't know what your dam is
[08:37] <DrSlony> solarant "ffprob video.mov" or video.flv
[08:38] <DrSlony> (why are you using flv????)
[08:48] <solarant> ubitux:  it is Resource Space'
[08:48] <solarant> DrSlony: just chekcing
[08:53] <solarant> ubitux:  output from ffprobe of the flv file - http://paste.ubuntu.com/9567568/
[08:54] <solarant> output from the ffprobe of the mov file - http://paste.ubuntu.com/9567570/
[08:54] <solarant> DrSlony:  Resource Spaces uses flv for preview generatio
[09:00] <solarant> ubitux:  did you mean the encoder - then it is Lavf54.63.104
[09:01] <ubitux> i meant the codec, and it's flv1
[09:01] <ubitux> (flv being the container)
[12:09] <targitaj> hello
[12:09] <targitaj> can anybody help me
[12:09] <targitaj> ?
[12:20] <__jack__> targitaj: about ?
[13:08] <targitaj> I install Unreal Media Server, run command ffmpeg -i 1.flv -f flv "rtmp://127.0.0.1:5130/live/ggg"
[13:09] <targitaj> error occurs
[13:09] <targitaj> HandShake: client signature does not match!
[13:09] <targitaj> RTMP_ReadPacket, failed to read RTMP packet header
[13:09] <targitaj> rtmp://127.0.0.1:5130/live/ggg: Unknown error occurred
[13:09] <targitaj> what the problem
[13:11] <targitaj> hm changed settings
[13:11] <targitaj> HandShake: client signature does not match!
[13:11] <targitaj> RTMP_ReadPacket, failed to read RTMP packet header
[13:11] <targitaj> rtmp://127.0.0.1:5130/live/ggg: Unknown error occurred
[13:11] <targitaj> I will google
[13:49] <targitaj> Can't find solution ffmpeg +
[13:49] <targitaj> Unreal Media Server +
[13:49] <targitaj> ffmpeg -i 1.flv -f flv "rtmp://localhost:5130/live/zzz"
[13:50] <targitaj> error
[13:50] <targitaj> HandShake: client signature does not match!
[13:50] <targitaj> RTMP_ReadPacket, failed to read RTMP packet header
[13:50] <targitaj> rtmp://127.0.0.1:5130/live/ggg: Unknown error occurred
[13:56] <sabkaraja> Hi, I have posted an issue on the trac which is marked as enhancement.I am willing to pay if someone can take a look and fix it. https://trac.ffmpeg.org/ticket/4182
[14:26] <sabkaraja> Hi, I have posted an issue on the trac which is marked as enhancement.I am willing to pay if someone can take a look and fix it. https://trac.ffmpeg.org/ticket/4182.. Please contact me at sabkaraja at hotmail.com
[14:39] <BtbN> Are you sure this isn't just a broken file? Is something else able to decode that?
[15:29] <shevy> can ffmpeg be used to modify audio or video, other than format conversions? for instance, if I wish to change the pitch or some part of the stream or something like that, or apply some kind of audio and/or video filter
[15:39] <Mavrik> shevy, yes, it has audio and video filters :)
[15:39] <Mavrik> see filter documentation :)
[15:41] <shevy> ah ok
[15:41] <shevy> I don't think I ever used them before yet
[15:58] <iive> though it might be tricky if you want to use then for one specific part in the video
[15:58] <iive> then/them
[17:45] <claz> is there by any chance an option to make ffmpeg dump rtsp requests/responses?
[18:16] <shevy> iive hmm let me think
[18:33] <squeegily> I want to save the N-th frame of a video as a Jpeg image. How would I do this?
[18:36] <c_14> use the select filter
[18:40] <squeegily> c_14: I already read https://www.ffmpeg.org/ffmpeg-filters.html#select_002c-aselect but I have no idea how to go about selecting a specific frame
[18:42] <Mavrik> select=eq(n, 5)
[18:42] <c_14> select=eq(n\,frame_number--- yes, that
[18:45] <squeegily> It just stays at frame=    0 fps=0.0 q=0.0 size=N/A time=00:00:00.00 bitrate=N/A
[18:46] <squeegily> I used -vf select="eq(n\,5)"
[18:46] <c_14> It might take a while.
[18:46] <c_14> But it'll get there.
[18:46] <c_14> It just has to count every frame.
[18:46] <c_14> Until it gets to in this case 5
[18:47] <c_14> Though
[18:47] <c_14> if it really is 5, it should be pretty fast.
[18:47] <Mavrik> yeah, adding -vframes 1 will help :)
[18:48] <c_14> That shouldn't matter if you use the jpeg muxer, should it?
[18:48] <squeegily> ffmpeg -f video4linux2 -i /dev/video0 -vf select='eq(n\,5)' -vframes 1 test.jpeg
[18:48] <squeegily> This is staying at frame = 0
[18:48] <Mavrik> c_14, hmm, not really sure, will it cut off immediately or wait for the second frame to die? dunno :)
[18:48] <c_14> Mavrik: it cuts of for me
[18:48] <c_14> squeegily: i'm not sure v4l2 has frame numbers per se...
[18:49] <squeegily> The first frame or 2 of the webcam are messed up as it initializes and adjusts for lighting/etc.. what should I do?
[18:49] <c_14> squeegily: do you want all frames > than N?
[18:49] <squeegily> c_14: I want the first frame >N; the Nth frame
[18:50] <squeegily> Just a single frame so the Jpeg encoder doesn't have a stroke
[18:50] <squeegily> ..would VLC be better for this?
[18:50] <Mavrik> hmm
[18:51] <Mavrik> no idea
[18:51] <c_14> I think it should work, no idea why it doesn't.
[18:51] <c_14> Can you try with file input?
[18:51] <c_14> Just to make sure that your command works technically.
[18:56] <squeegily> c_14: Wow, the command worked. The quality was unbelievably awful, the the command created an image of the video. ffmpeg -i ../Anime/S+W_LowRes/10.mkv -vf select='eq(n\,90)' -vframes 1 test.jpeg
[18:56] <c_14> So it's something with the v4l2 source somewhere...
[18:56] <squeegily> So.. it seems the problem just lies with accessing the webcam
[18:57] <Mavrik> squeegily, might wanna use -qscale to set jpeg quality
[18:57] <Mavrik> I think the default setting is terrible :)
[18:57] <squeegily> Mavrik: I even tried a PNG image and I think it's just the input video quality
[18:57] <c_14> Dirty hack: just pipe ffmpeg -f v4l2 -i /dev/video0 -f mkv pipe:1 | ffmpeg -i - -vf select[foobar]
[18:57] <Mavrik> ah
[18:57] <Mavrik> yes
[18:57] <squeegily> c_14: What purpose would that serve?
[18:58] <squeegily> I tested it out and ffmpeg -i /dev/video0 -vframes 70 video.mkv doesn't work either.. It just drops a ton of frames
[18:58] <c_14> Then there might be somthing with your device.
[18:59] <squeegily> It records 1 frame successfully and the webcam lights up the whole time
[18:59] <Mavrik> squeegily, note that for some cameras when using V4L2 you need to set framerate / resolution / etc.
[18:59] <squeegily> VLC can view the webcam with no problems.. They both use the same interface (V4L2), so the problem likely lies with ffmpeg
[19:00] <squeegily> Mavrik: Stream #0:0: Video: h264 (libx264) (H264 / 0x34363248), yuv422p, 1280x800, q=-1--1, 7 fps, 1k tbn, 7 tbc
[19:00] <c_14> squeegily: what FFmpeg version are you running?
[19:00] <Mavrik> squeegily, https://www.ffmpeg.org/ffmpeg-devices.html#video4linux2_002c-v4l2
[19:00] <Mavrik> squeegily, see that you set explicitly what v4l2 driver has to tell to camera
[19:00] <Mavrik> sometimes that's the issue with ffmpeg
[19:01] <squeegily> The latest version from the site. My distro's repos are outdated, so I'm not using it from the repos
[19:02] <DrSlony> squeegily instead of "The latest version from the site." be specific
[19:03] <squeegily> I downloaded it like yesterday.. It's 2.5-
[19:05] <squeegily> Trying version N-42022-gc89751a- now
[19:05] <DrSlony> there is more than one "latest version from the site."
[19:07] <squeegily> DrSlony: I tried the latest version (2.5) and the latest Git build with the same result. The builds were from http://johnvansickle.com/ffmpeg/ which is linked to by the official FFMPEG site
[19:07] <squeegily> So version 2.5 and version N-42022-gc89751a both didn't work
[19:09] <squeegily> Out of time now.. maybe I'll get it working next time
[20:08] <shevy> iive let's say I have a .mp4 files. I use ffmpeg to extract video only, without audio. no problem. now... is there a way to split up the individual video frames into standalone "pictures", modify a few of these, then merge them back in without any loss in video quality?
[20:08] <shevy> I figure that I could always apply some filters on a few of these frames, without affecting the other stuff that is in the video, then merge it back into a full video again, and then multiplexing back the audio into the .mp4 (or .avi, whatever)
[20:15] <vlt> shevy: Yes. It's possible. You can split the video to single images, edit some and later use ffmpeg to create a video from the the image sequence. But ...
[20:15] <vlt> There's only no quality loss if you use a lossless codec later, like ffv1 or huffyuv (or lossless h.264).
[20:16] <shevy> hmm damn
[20:16] <shevy> or wait
[20:16] <vlt> shevy: Recoding to any lossy codec virtually everytime means loss. Hence the name ;-)
[20:16] <shevy> can I convert, let's say, xvid, first into a lossless codec, modify the frames... ah damn... I guess, when i go back to xvid, I will lose quality again :(
[20:16] <shevy> yeah
[20:17] <vlt> shevy: Exactly.
[20:17] <shevy> hmm
[20:17] <vlt> Exporting to a png image sequence is already a lossless concersion.
[20:18] <vlt> *conversion
[20:19] <shevy> yeah but if I have to get it back into a video, I lose quality right?
[20:19] <pzich> at least some
[20:19] <pzich> may be minimal
[20:19] <pzich> depending on quality
[20:19] <shevy> I wonder how programs like avisynth + virtualdub go about this
[20:19] <shevy> pzich yeah well... like a guy with glasses, as the years pass by, he needs stronger and stronger glasses to defeat the blur :D
[20:20] <vlt> shevy: You don't have to lose quality when going back to video. Just using a lossy codec *could* mean losing quality.
[20:20] <shevy> I am way too scared now
[20:21] <shevy> the blur will get worse and worse :(
[20:22] <iive> in theory if the source video is made of key-frames only, you could replace one frame.
[20:22] <iive> but if you have prediction frames, you'd have to reencode everything after the changed frame
[20:22] <iive> up to the next keyframe.
[20:23] <shevy> aha
[20:23] <iive> afaik, ffmpeg doesn't support such trick modes
[20:26] <iive> you could try cut the video in 3 parts, the middle one is the segment where the frame is. reencode just it, then merge. however it must be encoded using the same setting/parameters as the rest of the video. aka at least same extradata.
[20:28] <shevy> yeah well
[20:29] <shevy> that's interesting with the frames, I remember reading about it years ago like on doom9.org
[20:29] <shevy> but I'll try to keep things as simple as possible, my brain isn't on par to deal well with complications
[20:29] <shevy> or complex stuff for that matter
[20:32] <iive> :)
[21:04] <tnelsond> How would one "value propogate" the white values of a video? Can this be done with this dilate filter?
[21:10] <tnelsond> Essentially I am trying to increase the thickness of a video of white text on a black background.
[22:13] <DrSlony> tnelsond check out the frei0r filters
[22:20] <tnelsond> DrSlony: Thanks. However, I managed to get satisfactory results with multiple blend modes and boxblurs.
[22:20] <DrSlony> all with cli ffmpeg?
[22:20] <DrSlony> hardcore
[22:22] <tnelsond> Yeah, ffmpeg rocks, I just don't know why I have to hardcode these dvdsubs in a round-about way. Ffmpeg must not properly support bitmap dvd subtitle hardcoding yet?
[23:53] <shevy> got another question
[23:53] <shevy> I am using ffmpeg to identify the video in an old .mpg file
[23:53] <shevy> and it gives me this line:
[23:53] <shevy>   'Stream #0:0[0x1e0]: Video: mpeg1video, yuv420p(tv), 368x272 [SAR 178:163 DAR 4094:2771], 1150 kb/s, 25 fps, 25 tbr, 90k tbn, 25 tbc'
[23:53] <shevy> does this mean that the name of the codex is "mpeg1video" ?
[23:53] <shevy> *codec
[23:56] <DrSlony> yes
[23:59] <shevy> ok thanks. I wasn't sure, the name mpeg1video did sound a bit peculiar
[00:00] --- Sat Dec 20 2014