[Ffmpeg-devel-irc] ffmpeg.log.20131126

[burek]

[00:05] <psc1963> hello, i am a newbie that is trying to use ffmpeg to convert amr-wb found in an rtp stream to a wav file so I can listen to it.  does anyone have a command example i could try?
[00:13] <llogan> psc1963: ffmpeg -i rtmp://[username:password@]server[:port][/app][/instance][/playpath] output.wav
[00:13] <llogan> oh, rtp: rtp://hostname[:port][?option=val...]
[00:14] <llogan> http://ffmpeg.org/ffmpeg-protocols.html#rtp
[00:19] <Ankman> oh, people here :-)
[00:22] <Ankman> anyway, may be someone can help. i have a big matroska file which i cannot play with my laptop. i usually render it down. i wonder if it's possible to keep the aspect ratio. until now i used the -s and then widthXheight. but i have to calcualte it. isn't it possible to tell ffmpeg i want to render it down to 1/4 and it takes care of it?
[00:23] <JEEB> yes, look at the examples for the scale video filter
[00:23] <JEEB> in the documentation @ ffmpeg.org
[00:23] <llogan> http://ffmpeg.org/ffmpeg-filters.html#scale
[00:23] <Ankman> thanks
[00:23] <llogan> specifically the "-1" example
[00:25] <Ankman> ok, will try that with ffplay first
[00:27] <robadob> I've become stuck with a weird linker issue when trying to convert an ffmpeg AVFrame into a QT QVideoFrame. 2 of the constructors for QVideoFrame can't be linked ('undefined reference to..') if i have 1 of many ffmpeg headers(avcodec/swscale/avformat etc) included (in extern "C") tags. I've been fiddling with this for about a day now before tracing
[00:27] <robadob>  it back to these ffmpeg headers, but i'm still completely stumped as to why they should affect an unrelated library/how to fix it.
[00:38] <Ankman> thanks again
[01:17] <Logicgate> How do I get the author from an mp4 with ffmpeg?
[01:17] <Logicgate> it's not within show_format or show_streams
[01:23] <klaxa> define "author"
[01:28] <Logicgate> im screenshottingh
[01:28] <Logicgate> hold up
[01:29] <Logicgate> http://i.imgur.com/05DawBz.png
[01:34] <klaxa> looks like metadata
[01:34] <klaxa> ffmpeg -i video.mp4 -f metadata metadata.txt; grep Author metadata.txt
[01:34] <klaxa> or something like that
[01:35] <Logicgate> klaxa, I did that earlier, there is no metadata.
[01:36] <klaxa> hmm...
[01:38] <llogan> you want to set the author metadata?
[01:38] <Logicgate> I want to read it first then set it llogan
[01:39] <Logicgate> i know how to set it with -metadata var="value" when encoding
[01:39] <Logicgate> I need to read it
[01:40] <llogan> please provide a sample
[01:41] <Logicgate> Sample of what??
[01:41] <Logicgate> I'll send you a video
[01:41] <Logicgate> Is it possible to add codecs to ffmpeg after it's compiled??
[01:41] <llogan> no
[02:48] <anewuser> hi, curious, just hearing ffmpeg does GME but can't find more info
[03:08] <tm512> What bottlenecks might there be that would cause a recording to drop frames seemingly randomly? I'm recording with x11grab but randomly, the video locks up and then continues at a later frame
[03:08] <tm512> my cpu shouldn't be a problem for this
[03:09] <tm512> Happens with x264 ultrafast -qp 0, and huffyuv
[03:14] <llogan> anewuser: what's GME?
[03:17] <anewuser> game music emu
[03:17] <anewuser> a library to play videogame music file formats
[03:17] <anewuser> a general reference would be like listening supernintendo music on computer
[03:18] <anewuser> http://blargg.8bitalley.com/libs/audio.html#Game_Music_Emu
[03:20] <llogan> anewuser: i'm not sure, but check this out (from a ffmpeg dev) http://gamemusic.multimedia.cx/
[03:21] <tm512> llogan: including hundreds or thousands of lines?
[03:21] <tm512> nothing seems out of the ordinary
[03:21] <llogan> you can trim any repeating lines
[03:22] <llogan> anewuser: only problem is that i don't use the chrome spybrowser
[03:23] <tm512> I'll have to redo it in a bit to get the exact output, also it's run from a script, but I can get the general command that is being run
[03:23] <tm512> http://hastebin.com/raw/lumuluyalo
[03:25] <llogan> anewuser: oh, i see. yes, you can --enable-libgme in ./configure. i've never used it.
[03:26] <anewuser> llogan: haha, did not know about that site
[03:27] <tm512> I had the idea that my HDD was at fault, but I dunno how, I can copy from /dev/zero at 100+ MB/s
[03:32] <anewuser> ohhhhh
[03:33] <anewuser> does that mean it's also enabled in the linux world??
[03:33] <anewuser> one tool to rule them all
[03:34] <llogan> http://ffmpeg.org/ffmpeg-formats.html#libgme
[03:35] <anewuser> ohhh
[03:36] <anewuser> never knew it was part of the plan to add vgm files
[03:36] <anewuser> ruling the world XD
[03:36] <anewuser> vgmstream up next haha
[05:00] <ubuntuaddicted> hello guys, i am getting a ton of errors when trying to use ffmpeg (avconv) to decode and re-encode a video stream from a device node. here's the beginning output from the terminal http://pastebin.com/Lzu9R1GW and the avconv command actually freezes and the hd-pvr stops recording by itself. why would this be? i can use cat /dev/video1 > video.ts just fine and it captures video until I hit ctrl-c.
[13:47] Action: jokar hello All, How are you? How can I convert Mpg to VOB? 
[13:47] <jokar> Can you show me a ffmpeg command to convert mpg to vob?
[13:49] <relaxed> ffmpeg -i input.mpg -map 0 -c copy output.vob
[14:29] <brontosaurusrex> how would i burn srt subs to self-generated black that is correct in size and length?
[15:18] <Rajat> hi
[15:18] <Rajat> i need some help
[15:18] <Rajat> to install  ffmpeg-2.1 on rhel6
[15:19] <Rajat> hi
[15:19] <Rajat> any one there
[15:21] <Rajat> hi
[15:59] <dccc> i'm am trying to create a mobile app that resizes an mp4 with the api. can someone give me an overview of the flow in ffmpeg that accomplishes this? or perhaps point me to documentation?
[16:00] <dccc> from what i've read so far, it seems like i need to decode the file as a stream and recode it with the next size in the context.
[16:00] <dccc> new size*
[16:11] <Mavrik> dccc, there are examples in doc folder of ffmpeg source
[17:23] <brontosaurusrex> is it possible to get length if my only input is srt soft subs?
[17:25] <brontosaurusrex> like ffmpeg -i my.srt -f lavfi -i color=s=1920x1080:color=black -vf subtitles=filename=my.srt -r 25 -c:v qtrle -an out.mov < is my current cli
[17:25] <brontosaurusrex> but the encode goes on forever
[17:25] <klaxa> try adding -shortest
[17:26] <brontosaurusrex> klaxa, tried, not working
[17:26] <klaxa> hmm
[17:57] <ubitux> brontosaurusrex: -vf subtitles is *not* soft sub
[17:57] <ubitux> you're doing both soft and hard in your cmd
[17:58] <brontosaurusrex> ubitux, i used soft only to get the lenght, but its not working anyway
[17:58] <brontosaurusrex> ill have to parse subs and then use -t i guess
[17:59] <ubitux> why do you need that?
[17:59] <ubitux> use ffprobe to get information about a media
[18:05] <dharriso> hi, using the ffmpeg api to transcode audio only streams in an MP4 container or ADTS
[18:05] <brontosaurusrex> about srt as well?
[18:06] <dharriso> the audio is AAC, i need to extract the AAC profile from the input stream.  For MP4 i can extra the AudioSpecifcConfig from the extradata to extract this
[18:06] <dharriso> but for ADTS streams the extradat is not available
[18:07] <dharriso> the ADTS header has a profile but if its HE-AAC it only give AAC-LC
[18:08] <dharriso> is there a way to extract this information from the AAC frame data?
[18:10] <dharriso> So im looking to see if the Profile is HE-AAC or HE-AACv2, you know with the SBR details
[18:11] <dharriso> one of the spec says about implicite detection checking if the sampleFrequency specified in the ADTS and the frame data differ then it can be assumed to SBR
[18:11] <dharriso> is there somewhere in ffmpeg this is done that I could use as a reference
[18:11] <dharriso> thanks
[18:12] <brontosaurusrex> ubitux, about srt as well?
[18:58] <ubitux> brontosaurusrex: yes?
[19:10] <viric> hm I fail to use drawtext
[19:11] <viric> I don't see any text over the output video
[19:11] <brontosaurusrex> ubitux, ffmpeg doesn't seem to know the duration of the soft subtitle format like srt
[19:12] <ubitux> why do you need that?
[19:12] <ubitux> it makes no sense to have the "duration" of a subtitles
[19:13] <viric> aha
[19:13] <viric> ah got it. the default x=100,y=100 isn't very usefil
[19:13] <viric> :)
[19:13] <viric> x=0,y=0 I mean
[19:15] <viric> can drawtext use antialiasing?
[19:15] <ubitux> it uses freetype
[19:15] <ubitux> so yes, it has anti aliasing
[19:15] <viric> ah
[19:17] <viric> right
[19:17] <brontosaurusrex> ubitux, burning subs to black video
[19:17] <ubitux> and how the duration would help?
[19:17] <brontosaurusrex> it would stop encoding when the subtitles end
[19:20] <ubitux> ffprobe -show_entries packet=pts_time,duration_time -of compact
[19:20] <ubitux> try to parse this
[19:20] <ubitux> like, ffprobe -show_entries packet=pts_time,duration_time -of compact | tail -n 1
[19:20] <ubitux> and do a pts_time+duration_time
[19:20] <ubitux> should be enough
[19:20] <brontosaurusrex> don't have ffprobe compiled for osx, would that work with ffmpeg -i ?
[19:20] <ubitux> no
[19:21] <ubitux> (you can try different writers of course)
[19:22] <brontosaurusrex> so far i'am awk-ing the last timecode from srt and using that with ffmpeg ... -t awked
[19:22] <brontosaurusrex> but seems stupid
[19:22] <ubitux> just use ffprobe
[19:23] <ubitux> there is no reason for not having it deployed with a traditionnal install
[19:23] <brontosaurusrex> ill remember for the next time i will attempt to compile ffmpeg, thansk
[19:23] <brontosaurusrex> thanks*
[19:23] <ubitux> it's by default
[19:25] <viric> I want to do a title in black background, then fade to the input video. I fail to get the fade working
[19:26] <viric> http://sprunge.us/EVTB
[19:26] <brontosaurusrex> another q: with -lavfi color=s=1920x1080:color=black
[19:26] <brontosaurusrex> i'am not getting to speed
[19:26] <brontosaurusrex> stuck at about 200 fps
[19:26] <brontosaurusrex> is there another faster way?
[19:29] <viric> the text fades out, not the black bg
[19:39] <viric> damn it, the 'color' source is without alpha
[19:40] <viric> with drawbox it works
[19:54] <viric> how can I set the filter input pixel format, different than the filter output?
[19:55] <viric> I want 'alpha' channel in the filter
[20:01] <psc1963> llogan, thank u for your help!  that got me started in the right direction, and I have it working with the sdp option now
[20:01] <user0_> echo
[20:02] <viric> I fail to achieve an alpha fade. Any idea? http://sprunge.us/eWZT
[20:02] <user0_> Hi i have a lot of jpg files to convert to video, can someone help and tell me what should i put on arg -i ? (input file name) ?
[20:02] <DelphiWorld> hey guys
[20:02] <DelphiWorld> what's the secret of AC3/DTS audio?
[20:02] <DelphiWorld> who own them?
[20:05] <psc1963> does anyone know how to add in support to ffmpeg for g722.1 aka siren7?
[20:07] <brontosaurusrex> DelphiWorld, dolby and dts
[20:08] <brontosaurusrex> ok, got around 350 fps with some cropping, that should do for a while
[20:09] <brontosaurusrex> DelphiWorld, http://en.wikipedia.org/wiki/DTS_(sound_system) and http://en.wikipedia.org/wiki/Dolby_AC-3
[20:10] <brontosaurusrex> user0_, i have a script that does renamed symlinks and push that to ffmpeg  , are you on nix like system?
[20:10] <brontosaurusrex> user0_, but i think ffmpeg lately upgraded that section
[20:11] <DelphiWorld> thank brontosaurusrex
[20:11] <DelphiWorld> and how do ffmpeg support them without license?
[20:11] <DelphiWorld> just wondring
[20:11] <brontosaurusrex> DelphiWorld, i let them
[20:12] <DelphiWorld> brontosaurusrex: you? who's you:P
[20:12] <brontosaurusrex> nah, i think this is some sort of solution with source code vs. binaries
[20:12] <brontosaurusrex> but don't quote me
[20:12] <brontosaurusrex> a "solution"
[20:12] <DelphiWorld> ah
[20:13] <DelphiWorld> got ya
[20:14] <brontosaurusrex> hmm, what framerate would I need for smooth subs? would 1 be enough? lmao
[20:14] <brontosaurusrex> or maybe 2?
[20:14] <brontosaurusrex> fps that is
[20:14] <llogan> psc1963: sounds good. but i forgot what i did.
[20:15] <DelphiWorld> brontosaurusrex: maybe you know, what's dvdnav?
[20:16] <brontosaurusrex> DelphiWorld, some sort of dvd library for dvd navigation rendering
[20:16] <llogan> user0_: http://ffmpeg.org/faq.html#How-do-I-encode-single-pictures-into-movies_003f
[20:17] <user0_> brontosaurusrex: shure i'm on nix
[20:17] <user0_> llogan: ty
[20:17] <llogan> also http://trac.ffmpeg.org/wiki/Create%20a%20video%20slideshow%20from%20images
[20:17] <user0_> i go the last link
[20:17] <user0_> but i dont understand the -i input file name format
[20:17] <llogan> using "-vf fps" is recommended over -r for now due to some issues with -r
[20:17] <user0_> %03d
[20:17] <llogan> (for the output)
[20:18] <llogan> show an example of your input file
[20:18] <llogan> oh, output
[20:18] <user0_> 00606E60E78E()_0_20131126170132_34511.jpg
[20:18] <llogan> %03d.jpg would start with 001.jpg and go from there
[20:19] <user0_> 00606E60E78E()_ -> never change
[20:19] <psc1963> I was looking for a way to convert amr-wb in rtp to wav, thats what you helped me with yesterday.
[20:21] <viric> hey I used a pastebin
[20:21] <psc1963> now I am trying to do the same thing for 722_1 or siren7, but see no ffmpeg support for that audio codec
[20:22] <brontosaurusrex> user0_, this script http://paste.debian.net/67882 is inside a droplet (on osx) and will accept folder as input
[20:22] <brontosaurusrex> user0_, probably interesting from line 100 and on
[20:22] <DelphiWorld> psc1963: what you're trying to do for G.722.1?
[20:22] <viric> oh got it! format=argb before fade in, makes it work. fade doesn't handle yuva420p!
[20:22] <llogan> i see support for g.722 but not g.722.1
[20:23] <llogan> viric: you used pastebin but ignored the output
[20:23] <psc1963> take a 722.1 stream that is rcvd in rtp and convert it to a wav file
[20:23] <viric> llogan: ahhhh ok
[20:23] <llogan> user0_: i was confused. try using a glob pattern as shown in the trac link i gave you
[20:23] <viric> sorry
[20:24] <DelphiWorld> psc1963: http://www.freeswitch.org/
[20:24] <viric> in any case, argb was the key for fade to work with alpha
[20:25] <psc1963> there is a libg722_1 there at freeswitch i think.  do u suggest not using ffmpeg for this need?
[20:25] <llogan> viric: did you use the split because the drawbox and drawtext would overwise not work with fade?
[20:26] <llogan> psc1963: do you have a g.722.1 sample file?
[20:26] <psc1963> there is also a libsiren thread that may be worth trying.  has anyone used it?
[20:26] <psc1963> yes, i have one.
[20:27] <viric> llogan: I used split, because otherwise I've to manually set the color src of the same size as the input.
[20:27] <viric> split + drawbox instead of color
[20:27] <llogan> i see. seems to be a good workaround/lazy bonus
[20:28] <viric> it's tricky... the argb thing killed me for two hours almost :)
[20:28] <user0_> 00606E60E78E()_%d_%d_%d.jpg
[20:28] <llogan> no
[20:28] <llogan> ffmpeg -r 1 -pattern_type glob -i '*.jpg'
[20:28] <user0_> No such file or directory
[20:28] <llogan> except ignore that -r 1
[20:29] <user0_> kkkk
[20:29] <llogan> viric: yeah, i've failed at fade a few times and then keep forgetting
[20:30] <user0_> ksh: ffmpeg: Argument list too long
[20:31] <user0_> but it seems working ty :)
[20:31] <viric> I will draw a front picture as png, with gimp. And then, movie + drawtext + fade. Good that drawtext understands utf8!
[20:33] <llogan> psc1963: please provide a sample file if possible
[20:39] <psc1963> its a wire capture in pcap file format.  would u need it in rtpdump dump format to look at it?
[20:40] <llogan> psc1963: i'm too ignorant of the format to know
[20:53] <DelphiWorld> psc1963, use freeswitch. play the stream in a conference, start recording from conference to a wav file, done
[20:54] <psc1963> tx!
[20:55] <llogan> or provide a sample and submit a feature request in ffmpeg and someone may implement a decoder natively or via an external library
[20:56] <psc1963> will do that as well.  tx again.
[21:36] <Timothy_Gu> What's the difference between mov subtitle and text subtitle?
[21:37] <Timothy_Gu> (See mov2textsub and text2movsub bitstream filters)
[21:40] <Timothy_Gu> Seriously anyone?
[22:01] <relaxed> Timothy_Gu: maybe they're stored differently in the mov container.
[22:04] <relaxed> Although with most of these bitstream filters I don't see why ffmpeg can't do the right thing automagically.
[22:05] <relaxed> Take for instance: -bsf:v mjpeg2jpeg
[22:06] <Rabbitt> Sample footage 4k raw from a Canon C500 https://www.dropbox.com/sh/gsqjxbbabz8th5j/ZrtIZGbfFG
[22:07] <relaxed> Why would anyone ever want to copy mjpeg to jpgs that can't be decoded properly?
[22:07] <relaxed> It's retarded.
[22:15] <Timothy_Gu> relaxed: speaking of mjpeg2jpeg, seems like it doesn't even work
[22:16] <Timothy_Gu> by reading the code
[22:16] <relaxed> It worked last time I needed it.
[22:17] <Timothy_Gu> All the magic done in line 101-103 are in the out buffer, but the "out" buffer isn't even outputted
[22:19] <relaxed> it would be trivial to test
[22:20] <Timothy_Gu> You wanna do it? Not me cause I got nothing but an unrooted nexus 7 on my hand.
[22:46] <dharriso> anyone who might help with AAC audio question
[22:59] <viric> llogan: if I replace color=c=black by movie=a.png, fade fails again
[23:00] <viric> despite setting argb
[23:13] <viric> hm some pts thing I guess
[23:22] <viric> grmbl
[23:22] <viric> [Parsed_showinfo_2 @ 0x152ee60] n:0 pts:7194 pts_time:0.0399667 pos:15364 fmt:yuva420p sar:45/44 s:640x360 i:P iskey:0 type:B checksum:C3E08DC1 plane_checksum:[BC6C2649 CFEA068B 9011B05E 871DB080]
[23:22] <viric> what relates pts and pts_time?
[23:22] <viric> tb=1/25
[23:22] <viric> [Parsed_showinfo_2 @ 0x152ee60] n:1 pts:14394 pts_time:0.0799667 pos:17284 fmt:yuva420p sar:45/44 s:640x360 i:P iskey:0 type:B checksum:83AE8F83 plane_checksum:[C33F47AA 5BEC0CAA D64A8AA0 871DB080]
[23:26] <llogan> dharriso: what is the question?
[23:27] <llogan> viric: i'm not sure. hard to know what's going on without command and complete console output
[23:27] <viric> it's showinfo on a direct MTS file
[23:27] <viric> 25 fps, 50 tbr, 90k tbn, 50 tbc
[23:27] <viric> I guess that's the key... the 90k thing :)
[23:28] <viric> but I don't know what it is
[23:28] <dharriso> llogan: Im transcoing AAC content using the API.  I want to matain the AAC profile.  With AAC in MP4 cotnainer I can parse out the AudioSpecificConfig in the extradata to find this out e.g. HE-AAC
[23:28] <dharriso> however when the AAC is wrapped in ADTS no extra data
[23:28] <dharriso> it only gives the profile as AAC-LC from the ADTS header
[23:29] <dharriso> is it possible to extract this information from the AAC payload
[23:29] <dharriso> mediainfo on the file says HE-AAC
[23:29] <llogan> i've never used the API
[23:29] <dharriso> o right
[23:30] <viric> llogan: http://sprunge.us/gGXe
[23:30] <Rabbitt> https://bugzilla.libav.org/show_bug.cgi?id=598
[23:31] <llogan> Rabbitt: we are FFmpeg, not libav
[23:31] <Rabbitt> hmm, okay, someone did say in here yesterday that it wouldn't matter
[23:32] <llogan> bullshit
[23:32] <Rabbitt> I thought as much
[23:32] <Rabbitt> apologies
[23:32] <Rabbitt> I've read the controversy around the fork, so sincerest apologies
[23:32] <llogan> no problem, but if you're using ffmpeg, and want a feature in ffmpeg, then you should use the ffmpeg bug tracker
[23:33] <llogan> yes, it can be a confusing situation for users due to the hijacking of several ffmpeg related vocabulary
[23:33] <dharriso> llogan; do i need to check back later and ask the same question, for API, or should i use another channel?
[23:34] <llogan> dharriso: you can try the libav-user mailing list or try later. not as many api answers here unfortunately
[23:34] <llogan> http://ffmpeg.org/contact.html
[23:34] <llogan> for ML info
[23:35] <dharriso> llogan: thanks, will do both as you suggest
[23:37] <viric> llogan: any idea about the PTS not being time/TB?
[23:37] <viric> time * TB
[23:37] <viric> is TB = 1/framerate? or not necessarily
[23:37] <viric> ?
[23:44] <cbsrobot> Rabbitt: did you try http://www.libraw.org/ ?
[23:44] <Rabbitt> never heard of it
[23:44] <Rabbitt> so no
[23:44] <cbsrobot> they seem to support the c500 since release 16 alpha
[23:45] <Rabbitt> that doesn't really help much, that's only useful if you want to edit frames like they are from a camera
[23:45] <Rabbitt> a DSLR
[23:46] <cbsrobot> what do you want to do with it ?
[23:46] <Rabbitt> not interesting to me, I want ffmpeg to stream it as video... you have no idea how much processing goes into just a single frame of raw?
[23:46] <Rabbitt> it's a little insane
[23:46] <Rabbitt> it's like talking about doing a timelapse of 20k pictures you manually edit each
[23:47] <Rabbitt> I want to stream it into another format so that it can be worked on with say kdenlive or whatever video editor I prefer to use
[23:47] <cbsrobot> Well I used dcraw (the base of libraw) to work on cr2 files a few years ago
[23:48] <Rabbitt> photography workflows are insanely more single image oriented than video workflows
[23:48] <cbsrobot> normally you convert the raw pictures in something editable
[23:49] <Rabbitt> I use darktable these days, and they use the library, so I should be able, in theory, with the alpha version of this lib access the .rmf files, but that does not in any way help me turn a large video into something useful for a video workflow.  The processing time per raw is on average 5-10 seconds...  Can you imagine 100k frames in a video?
[23:49] <viric> does drawtext support carriage returns?
[23:50] <viric> newlines
[23:50] <cbsrobot> I even had a dailies workflow for the A-Cam dII which records single cinemadng files
[23:50] <cbsrobot> but sure it's a bit insane
[23:50] <Rabbitt> dcraw is great for photography, but it isn't a viable tool for raw video... right now there's a huge deficit in this side of things.
[23:51] <Rabbitt> ideally you want to do colour grading, white balance and what-not at any time in a video edit
[23:52] <Rabbitt> only commercial software does this at the moment
[23:52] <Rabbitt> and even then
[23:52] <cbsrobot> dcraw converts the 14 bit raw images into 16 bit tiff files - that was what I was able to process
[23:53] <cbsrobot> but I think nowadays davinci resolve can process them too
[23:53] <cbsrobot> it's even free for 2k projects (if I'm not mistaken)
[23:54] <cbsrobot> what software can process the c500 files ?
[23:54] <Rabbitt> yeah, DaVinci does it
[23:54] <Rabbitt> it can do it without any in-between processing
[23:58] <Rabbitt> hmm, if it ran under Wine I'd consider using it
[00:00] --- Wed Nov 27 2013