[Ffmpeg-devel-irc] ffmpeg.log.20120517

[burek]

[00:00] <burek> that's my last try :)
[00:00] <killown> burek, I know what was the problem
[00:00] <burek> try -i :1.0
[00:01] <killown> yes
[00:01] <killown> [hostname]:display_number.screen_number[+x_offset,y_offset]
[00:01] <killown> its right 1.0
[00:01] <killown> worked, burek thanks a lot
[00:01] <burek> does it work
[00:01] <burek> ok
[00:01] <burek> :)
[00:02] <killown> I think I need make a tutorial for that, I think there is many people having trouble with screen recorder on linux
[00:02] <burek> that would be cool :)
[00:04] <killown> I am very impressed with webm format, I can make my screencasts and convert to webm and it will provide such small file with 1080 quality
[00:04] <killown> I am talking about 30MB file with 2 minutes content
[00:04] <killown> I know its relative to the way I encode it
[00:04] <JEEBsv> and the vp8 encoder isn't even good :P
[00:04] <burek> you could put a few cmd lines, along with screenshots, just for a showcase :)
[00:04] <killown> but..webm FTW
[00:05] <JEEBsv> it might become usable after D_S gets finished with xvp8, which is sponsored by Mozilla -- but until that vp8 is just a PITA to use
[00:06] <JEEBsv> also, looking at the example lines you were thrown, libx264 and matroska were used
[00:06] <JEEBsv> not webm
[00:06] <JEEBsv> (webm = vp8 + ogg vorbis in matroska)
[00:06] <JEEBsv> [limited subset of matroska]
[00:07] <killown> JEEB, I think webm is the future, even if it is not done yet
[00:07] <share> File for preset 'medium' not found
[00:07] <JEEBsv> oh the vp8 spec is ready
[00:07] <share> how to solve this?
[00:08] <burek> share,
[00:08] <burek> can you please use pastebin.com, to show your command line and its output?
[00:08] <JEEBsv> share: use -preset ? and I guess you're using libx264?
[00:08] <share> yes i am
[00:08] <JEEBsv> and burek is useful for a change
[00:08] <share> -vpre
[00:08] <JEEBsv> use -preset for internal presets
[00:08] <JEEBsv> and -vpre for external ffpreset files
[00:09] <JEEBsv> 'medium' is the default libx264's internal preset
[00:09] <JEEBsv> thus, -preset
[00:09] <share> im trying to stream desktop
[00:09] <share> rtmp://127.0.0.1/8080: Operation not permitted  :P
[00:09] <JEEBsv> sounds like fun
[00:10] <ShinyObjects> Hey everyone. I'm making an app that takes a raw h264 stream  and converts it to flv or f4v
[00:10] <killown> root permission?
[00:10] <ShinyObjects> What's a good starting  place in the ffmpeg library's documentation for me?
[00:10] <killown> dont know
[00:10] <JEEBsv> killown: it's over 1024
[00:10] <JEEBsv> shouldn't need root, and even if it would need it -- it shouldn't be run as such
[00:10] <burek> ShinyObjects, ffmpeg's source code docs/examples/ dir?
[00:11] <ShinyObjects> Awesome burek - that looks like a great start.
[00:11] <killown> right
[00:11] <ShinyObjects> Thanks!
[00:11] <burek> :beer: :)
[00:11] <burek> when nothing else works, well, read the docs (manual) :D
[00:12] <burek> rtmp://127.0.0.1/8080 is not a valid rtmp address
[00:12] <burek> try rtmp://127.0.0.1:8080
[00:15] <ShinyObjects> Haha, burek: I've set pidgin up to highlight messages containing beer. You said just the right thing
[00:15] <burek> ^^ :)
[00:16] <share> JEEBsv: it's not fun if i cant do it
[00:17] <share> so ffmpeg can act as server right..
[00:18] <share> cannot*
[00:19] <JEEBsv> did you try the way burek noted the url?
[00:19] <JEEBsv> otherwise, I don't know how well ffmpeg can dance the rtmp dance
[00:20] <share> ah ill try
[00:20] <share> No application or playpath in URL!
[00:22] <share> and also tried with sudo
[00:22] <burek> your rtmp url is invalid
[00:22] <burek> what exactly are you trying to do?
[00:22] <burek> can you please use pastebin.com, to show your command line and its output?
[00:23] <share> im trying to create screencast
[00:25] <share> burek: http://pastebin.com/raw.php?i=1Gj933cR
[00:26] <burek> ok, now, do you have any rtmp server installed?
[00:26] <burek> like Wowza or Red5?
[00:26] <share> no xD
[00:26] <burek> well then, who are you sending the output to? :)
[00:26] <share> so I guess the easy way is to use justin.tv or something
[00:26] <burek> the best is to use wikipedia and google and read more about rtmp or screencasting
[00:27] <burek> before starting actually to work on it :)
[00:27] <share> is this a science
[00:29] <burek> in the beginning usually :) but during time, while experience develops, it stops being that :)
[00:41] <Aleks3Y> hey guys , can someone help with -g option ?
[00:42] <Aleks3Y> i need to convert all the videos to mp4
[00:44] <floater_f432c> what is -g ?
[00:45] <Aleks3Y> gop_size
[00:46] <floater_f432c> ah, never used that
[00:46] <Aleks3Y> i have to manage script that someone wrote already before , can i post it here and maybe you can suggest something ?
[00:46] <burek> Aleks3Y, what exactly do you need with -g
[00:46] <Aleks3Y> just ffmpeg part :)
[00:46] <burek> can you please use pastebin.com, to show your command line and its output?
[00:46] <Aleks3Y> burek: it was already there and set to 60
[00:47] <burek> ok, and what do you want to do with it
[00:48] <floater_f432c> http://v4l2spec.bytesex.org/spec/
[00:49] <Aleks3Y> here you go guys: http://pastebin.com/H7shZmjY
[00:49] <Aleks3Y> i just found that -g 60 is very high, no? everywhere i was reading people suggesting 12-15
[00:50] <Aleks3Y> ffmpeg 0.7.12
[00:51] <burek> Aleks3Y, did you read and understand what gop is?
[00:53] <Aleks3Y> burek: almost as far as my video knowledge let me go
[00:53] <Aleks3Y> as i understood higher number, lower size of encoded file
[00:54] <Aleks3Y> but higher latency
[00:55] <burek> hm
[00:55] <Aleks3Y> so if this is for video streaming 60 should be ok, IBBP ?
[00:55] <burek> search for gop
[00:55] <Aleks3Y> read it
[00:56] <Aleks3Y> that's why i'm confusing
[00:56] <burek> gop tells you how frequently to create intra frames
[00:56] <burek> -g 60 means one intra frame each 60 frames
[00:56] <burek> if your video is 60 frames per second that means one intra frame each second
[00:56] <burek> if it's 15 fps then 1 i-frame each 4 seconds
[00:56] <burek> etc
[00:57] <burek> so, it really depends on your fps
[00:59] <Aleks3Y> can be anything i guess
[00:59] <Aleks3Y> people just upload files
[00:59] <Aleks3Y> question about default presets as well, should i use them or it's better to specify everything yourself?
[01:00] <burek> wait
[01:00] <burek> focus on what I'm telling you
[01:00] <Aleks3Y> ok, lets finish that )))
[01:01] <burek> the point is that -g 60 doesn't represent anything meaningful if you don't provide fps rate
[01:01] <burek> so
[01:01] <burek> can you please use pastebin.com, to show your command line and its output?
[01:01] <burek> for one of your video files
[01:01] <burek> which has -g 60
[01:01] <Aleks3Y> i sec
[01:04] <Aleks3Y> want me to finish it, or jsut avg fpm?
[01:04] <Aleks3Y> fps
[01:04] <TACPILOT> currently trying to create a sequence of openexr images from an mkv video. Since ffmpeg does not encode exr I must pipe output to another app.
[01:05] <TACPILOT> trying to use convert, but that app will not treat the stream as frames
[01:05] <TACPILOT> so output is just a single image
[01:06] <TACPILOT> ffmpeg -i clip.mkv -f image2pipe | convert - image%d.exr
[01:06] <TACPILOT> err ffmpeg -i clip.mkv -f image2pipe - | convert - image%d.exr
[01:06] <burek> Aleks3Y, using -vpre to specify libx264 preset is deprecated
[01:06] <burek> you should use -preset
[01:07] <burek> also, if you don't know the fps of your input file
[01:07] <burek> then -g 60 really doesn't make any sense
[01:07] <burek> unless you specify -r <fps> for the output
[01:07] <burek> because if your input is 60 fps then you'll have 1 i-frame each second
[01:07] <burek> as explained before
[01:08] <gurrag> What sort of flags can you give to ffmpeg on the command line to prevent it from skipping during audio transcoding [non real time] ?
[01:08] <burek> people use -g 250 and similar values, but that doesn't mean you should use it too
[01:08] <burek> so, once again, Aleks3Y, what exactly is your goal?
[01:08] <Aleks3Y> burek: http://pastebin.com/XFB3xUXi
[01:08] <Aleks3Y> burek: found fastest way to convert video to mp4 and stream them after
[01:09] <burek> TACPILOT, your 'convert' script is wrong maybe?
[01:09] <Aleks3Y> burek: currently i'm resizing only videos that are bigger then 720xZZZ, so that saved some time on conversion
[01:09] <burek> gurrag, what skipping?
[01:10] <burek> Aleks3Y, just to find fastest way? :D
[01:10] <gurrag> Occasionally when trying to transcode non-compressed .wav files to compressed .ogg, I've noticed that there are occasionally segments where the audio cuts out or skips in the output
[01:10] <TACPILOT> burek: it turns out to be the nature of convert and how it treats streams .. dug for days till I found them stating it cant be done
[01:11] <TACPILOT> convert reads the entire stream then closes it before beginning to process
[01:11] <burek> ffmpeg -i input.avi -vcodec libx264 -preset ultrafast -acodec copy -s 720x480 -threads 0 /root/test.mp4
[01:11] <burek> Aleks3Y,*
[01:11] <TACPILOT> then treats the whole stream as a single image
[01:11] <TACPILOT> the only way is to send convert one image at a time
[01:12] <burek> gurrag, try other output format (mp3 or aac) and see if there are skips/cuts too
[01:12] <TACPILOT> any suggestions on how to do that ?
[01:12] <burek> TACPILOT, find another script? :)
[01:12] <Aleks3Y> burek: any disadvantages?, your smile made me thinks :D
[01:12] <TACPILOT> point me
[01:13] <burek> -f image2pipe sends several images, one by one, one after another
[01:13] <burek> that's all
[01:13] <burek> you just need a script that will know how to handle that
[01:13] <burek> Aleks3Y, quality probably will be bad, but try and see :)
[01:13] <TACPILOT> I have not yet been able to locate another ... suggestions ?
[01:14] <burek> let me ask google :)
[01:14] <TACPILOT> maybe u have better luck than I .. I fresh pair of eyes can make all the diff  :)
[01:15] <burek> TACPILOT, did you read this http://www.imagemagick.org/discourse-server/viewtopic.php?f=2&t=18320
[01:16] <TACPILOT> yes
[01:16] <burek> well there you go :)
[01:18] <TACPILOT> not sure how that applies here .. their woking with ppm piping it all the way back to ffmpeg in which ffmpeg cant encode exr ..
[01:19] <Mista_D> Anyway to preserve Closed Captioing(CEA-608 aka Line21), going from MPEGTS(MPEG2/MP2) to MPEGTS(H.264/AAC)?
[01:19] <Aleks3Y> burek: i want to keep the quality close to original, but resize it if it's bigger then 720 to 720, i'm checking presets lossless_fast and lossless_medium, main difference is flag coder=0/1
[01:20] <Mista_D> Aleks3Y: basically coder=1 means you're encoding with CABAC and its slower, but better compression.
[01:21] <burek> Aleks3Y, lossless_fast and -vpre in general is deprecated
[01:21] <burek> you should use -preset
[01:21] <burek> and check: x264 --help
[01:21] <burek> for all the available -preset -tune and -profile options
[01:21] <Freakshow> aha! it is deprecated! ;-)
[01:21] <Aleks3Y> burek: ya, got it, i'm talking about file names )
[01:21] <Aleks3Y> for presets)
[01:21] <Freakshow> just kidding
[01:22] <burek> Aleks3Y, they are not used anymore for libx264
[01:22] <Aleks3Y> oh, ok, let me man it )
[01:22] <burek> Freakshow, can you use pastebin.com to show the content of your convert script
[01:22] <burek> sorry TACPILOT
[01:22] <burek> :)
[01:23] <Freakshow> I'm sorry... what script are we talking about?
[01:23] <Freakshow> with respect to the ad-cue tone?
[01:23] <burek> Mista_D, did you try -dcodec copy and -scodec copy
[01:24] <Mista_D> I'll try decodec.
[01:25] <Aleks3Y> burek: i don't have x264, only libx264
[01:25] <Aleks3Y> probably it's too old setup
[01:25] <Aleks3Y> but running latest deb's from debian multimedia - repo
[01:25] <Freakshow> burek: which script are you referring to?
[01:25] <TACPILOT>  burek: no worries .. TY for looken  :)
[01:25] <Aleks3Y> time to create my own deps from ffmpeg repo
[01:25] <burek> Aleks3Y, I meant libx264
[01:25] <burek> :)
[01:25] <Aleks3Y> debs*
[01:26] <burek> also
[01:26] <Freakshow> oh... duh, I need to learn how to read
[01:26] <burek> always try to use latest git version of ffmpeg
[01:26] <burek> if possible
[01:26] <burek> Aleks3Y, http://ffmpeg.gusari.org/viewtopic.php?f=25&t=14
[01:26] <Mista_D> burek: Unrecognized option 'dcodec'... try SVN?
[01:26] <burek> Mista_D, what version of ffmpeg do you use?
[01:27] <burek> Freakshow, it was a mistake, sorry :)
[01:27] <Freakshow> no no, I'm sorry it took me that long to realize what was going on
[01:27] <Mista_D> burek: from February...
[01:27] <Aleks3Y> burek: i don't need anything else for input files?
[01:27] <Aleks3Y> last time i had to compile a lot of staff for ffmpeg, like lame, etc
[01:28] <Mista_D> burek: trying GIT from April 20th...
[01:29] <burek> Mista_D, http://git.videolan.org/?p=ffmpeg.git;a=commit;h=e3b540b42464395eadbc42b87331589e2e9e4301
[01:29] <burek> it should be available since May 2011
[01:30] <Mista_D> burek: "ffmpeg version 0.10.2.git-dd5e1fc" accepted `-dcodec`.
[01:30] <burek> Aleks3Y, well, it depends what do you need
[01:30] <burek> but if you want, you can also quick-try static build from http://ffmpeg.gusari.org/static/
[01:31] <Aleks3Y> any incoming video to x264
[01:31] <burek> it contains the latest static git version of ffmpeg so you can just download and run it to see if it works
[01:31] <Aleks3Y> h
[01:32] <burek> Mista_D, honestly, with multimedia projects as active as ffmpeg is, it is usually wise to use the latest git :)
[01:33] <Aleks3Y> burek: ok, so for now the best will be just to replace -vpre and keep -g ? or jsut remove -g at all for now? like: -preset medium -g 60
[01:34] <Aleks3Y> also, if i'm reconverting audio to 2ch is it takes a lot of extra time and copy is will be much faster, or not big difference?
[01:35] <share> is it possible to add a delay in streaming? for example 30 seconds
[01:35] <Mista_D> burek: agree, active in FFmpeg terms =insane
[01:35] <share> -itsoffset
[01:35] <Freakshow> however, burek: do you happen to know anything about restreaming an flv with ffmpeg and having it 'not' strip out ad cue tones?
[01:37] <burek> Aleks3Y, if you just want to re-encode the video fast (keeping the audio), then I would just use -preset ultrafast and -acodec copy
[01:37] <burek> Mista_D, :D
[01:38] <burek> Freakshow, what are ad cue tones? :D
[01:39] <burek> share, -itsoffset is not what you need.. that options delays audio/video stream so that you can move it forward/backward to match the other audio/video stream
[01:39] <burek> for example if your audio is late, you can "push" the video stream forward to sync a/v
[01:39] <burek> using -itsoffset
[01:39] <Aleks3Y> ok, i think without tests hard to decide , one more Q... how to check if file is already mp4 or it's another container? like mkv? they both have h264 when i check info via ffmpeg
[01:40] <Freakshow> DTMF cue tones inserted into the audio track for players to que up and insert ads while a consumer is watching a stream
[01:40] <burek> Aleks3Y, just type ffmpeg -i input.avi
[01:40] <burek> and see the output :) it tells you everything :)
[01:40] <Aleks3Y> burek: ya, and both of them have h264
[01:40] <burek> Freakshow, sorry :( never used those :(
[01:41] <Aleks3Y> right now i'm taking that output and doing a check like that: if $info =~ /h264/ && $info !~ /matroska/
[01:41] <Aleks3Y> then it's mp4
[01:41] <Aleks3Y> but i'm not sure it's goood)
[01:41] <Aleks3Y> burek: the idea to not convert mp4 files
[01:42] <Freakshow> burek: thanks... it's just odd that a straight -vcodec copy and -acodec copy is pulling them out
[01:42] <burek> Aleks3Y, you can use ffprobe to check that
[01:42] <burek> or a decend parser and ffmpeg
[01:43] <burek> consider http://pastebin.com/5nZxqWu0
[01:43] <burek> Input #0, flv, from 'Temp\Temp.flv':
[01:43] <burek> (container = flv)
[01:43] <burek> Stream #0:0: Video: h264 (High), yuv420p, 1280x720, 500 kb/s, 25 tbr, 1k tbn, 50 tbc
[01:43] <burek> (vcodec = h264/libx264)
[01:43] <burek> etc
[01:43] <burek> Freakshow, can you type ffmpeg -i input.avi
[01:44] <burek> and use pastebin, just to peek at it
[01:47] <Freakshow>     metadatacreator : inlet media FLVTool2 v1.0.6 - http://www.inlet-media.de/flvtool2
[01:47] <Freakshow>     hasCuePoints    : true
[01:48] <burek> can you please use pastebin.com, to show your command line and its output?
[01:49] <Freakshow> oh sure sure, sorry was looking at something else as well
[01:50] <Freakshow> just a sec
[01:50] <share> burek: do you know how to capture only the center of screen_
[01:50] <share> ?
[01:50] <Mista_D> Can't copy CC (line21/CEA-608) from mpeg2 to x264(mpegts). Latest GIT with '-dcodec copy -scodec copy'...
[01:51] <burek> share, why is that difficult?
[01:51] <burek> Mista_D,
[01:51] <burek> can you please use pastebin.com, to show your command line and its output?
[01:51] <share> burek: i already streamed successfuly
[01:51] <Mista_D> burek: its a data feed in video ES, and it might be difficult.
[01:52] <share> burek: but i only want to stream the center of screen
[01:52] <burek> share,
[01:52] <burek> can you please use pastebin.com, to show your command line and its output?
[01:52] <Freakshow> burek: http://pastebin.com/adFbRTWm
[01:52] <burek> Mista_D, I don't understand? :)
[01:53] <share> burek: im using 1920x1080 resolution. to decrease cpu usage I changed output resolution to 640x480
[01:54] <share> and qualitfy to fast. im also using -i :0.0+nomouse
[01:54] <burek> err.. Freakshow, what's the point of -b:v, -bt -r, when you are using -vcodec copy? :)
[01:54] <Freakshow> true true
[01:54] <Freakshow> yes, I deserved that
[01:54] <Freakshow> :P
[01:55] <burek> and btw, when you finish the copy process, check the file sizes of both input/output files :)
[01:55] <Freakshow> was that to me or someone else?
[01:55] <burek> share,
[01:55] <burek> can you please use pastebin.com, to show your command line and its output?
[01:56] <share> burek: http://pastebin.com/g8MUKT8b
[01:56] <burek> Freakshow, to you (file sizes)
[01:57] <share> device: :0.0+nomouse -> display: :0.0 x: 0 y: 0 width: 640 height: 480
[01:57] <share> and also   Incompatible pixel format 'bgra' for codec 'libx264', auto-selecting format 'yuv420p'
[01:58] <burek> share, -i :0.0+10,20+nomouse
[01:58] <share> ill try thanks
[01:58] <burek> that will move recording rectangle to (10,20)
[01:58] <burek> (upper-left corner)
[01:59] <Freakshow> the file sizes are the same
[02:00] <Freakshow> it appears that some metadata is preserved while some is stripped off
[02:00] <Mista_D> burek: http://pastebin.com/vzsEF3Rv here's a command.
[02:01] <Mista_D> It seems it can't detect CC1, trying ffprobe..
[02:02] <share> burek: yes but i would like to center it
[02:04] <burek> Freakshow :)
[02:04] <burek> try using binary comparison on files :)
[02:05] <burek> Mista_D, can you try using the latest static git version of ffmpeg?
[02:05] <burek> http://ffmpeg.gusari.org/static/
[02:05] <burek> just to see if it can detect it
[02:05] <share> found it http://ubuntuforums.org/showpost.php?s=74ccc1001a3abda56a8eb9b1238836e7&p=10683822&postcount=4
[02:06] <burek> share subtract width of your desktop with 640 (same for height and 480) and divide by 2
[02:06] <burek> you'll get your X,Y for -i
[02:08] <Freakshow> burek: they'll definitely be different given the metadata is missing
[02:11] <burek> Freakshow, just test it :)
[02:11] <Freakshow> ok ok
[02:12] <burek> just type md5sum input.avi
[02:12] <burek> and md5sum output.avi
[02:12] <burek> and check the hashes
[02:13] <Freakshow> well, I was checking on the target end
[02:13] <share> xwininfo | grep -e Width -e Height -e Absolute
[02:13] <Freakshow> but I can do it locally first
[02:14] <Freakshow> different.
[02:16] <Freakshow> and I stripped out the output and just dumped it locally
[02:16] <burek> yes
[02:16] <Freakshow> ffmpeg -i out1.flv -acodec copy -vcodec copy -f flv new_out1.flv
[02:16] <burek> locally
[02:16] <burek> so, they differ?
[02:16] <Freakshow> and yes, they differ
[02:16] <burek> that's a bug
[02:16] <burek> but
[02:16] <burek> since you use version 0.9.x
[02:16] <burek> that's old
[02:16] <burek> so you should try with latest git
[02:17] <burek> or try the static, before you decide to compile everything
[02:17] <Freakshow> well, this is on bsd... so that's a bit different since the port for bsd is what I'm using
[02:17] <Freakshow> I'll get the git and get going
[02:17] <Freakshow> :P
[02:17] <burek> just try :)
[02:18] <Freakshow> no worries... I'll check it out
[02:21] <Freakshow> give me a tick
[02:24] <Freakshow> unfortunately that's not compiled for BSD...
[02:24] <Freakshow> ./ffmpeg
[02:24] <Freakshow> ELF binary type "0" not known.
[02:24] <Freakshow> -bash: ./ffmpeg: cannot execute binary file
[02:25] <Freakshow> it's ok, I'll build it
[02:25] <Freakshow> but outside of md5, cmp also indicated they are different
[02:25] <Freakshow> cmp new_out1.flv out1.flv
[02:25] <Freakshow> new_out1.flv out1.flv differ: char 16, line 1
[02:27] <burek> hmh, well they shouldn't differ
[02:28] <burek> I mean
[02:28] <Freakshow> agreed
[02:28] <burek> if ffmpeg didn't recognize something, it would be logical just to copy it
[02:28] <burek> unless specified to "clean" the output of "unsupported" stuff or something
[02:28] <Freakshow> makes sense
[02:29] <Freakshow> and that's exactly the behavior I was expecting
[02:29] <Aziroshin> Hello. Using gentoo. Pastebin: http://pastebin.com/tMjGpGUc . I am recording from a microphone. The problem is, that the voice is getting "sparkly" (like choppy but less severe). The problem decreases considerably if I don't specify any audio codec and default to .mkv
[02:30] <Aziroshin> I tested the microphone with the teamspeak 3 testing functionality; it doesn't display such behaviour. Unfortunately, testing the microphone through loopback somehow doesn't work.
[02:31] <Aziroshin> I am absolutely new to encoding, I should probably add.
[02:33] <burek> Aziroshin, did you try anything beside libvorbis
[02:33] <burek> like mp3 or aac
[02:41] <share> burek: if i switch to a game ffmpeg stops recording
[02:46] <burek> maybe your game draws directly on video card
[02:46] <burek> and not through the X11
[02:46] <share> im using Wine btw
[02:46] <share> opengl
[02:48] <burek> well did you read the game's manual
[02:48] <burek> maybe it has some shortcuts for recording the scenes
[02:49] <share> yes it can record but the game is needed to play the file
[02:49] <burek> did you try ffmpeg -i gamesavefile
[02:51] <share> it's not a video file
[02:52] <burek> did you try
[02:55] <share> tes
[02:55] <share> yes it doesnt open
[02:55] <share> and i would have to record and then play with ffmpeg to stream
[02:56] <burek> did you try providing a small sample here, to let people analyze it
[02:56] <burek> maybe it's a plain video file
[02:56] <burek> packed into some dummy container
[02:56] <hi117> Aziroshin: try saving as wav using arecord, this will check if its your mic or your encoding
[02:57] <share> burek: maybe but i would have to record and thats not what i want
[02:57] <share> burek: so theres no way to stream "from opengl"
[02:57] <share> ingame
[02:57] <hi117> and worst case you can pipe arecord's output to ffmpeg and encode from there
[02:58] <burek> share, I dunno.. never done it..
[02:59] <share> there is glc
[03:00] <share> https://wiki.archlinux.org/index.php/GLC
[03:00] <share> it also records
[03:01] <Aziroshin> arecord on the right device gives me: "arecord: set_params:1059: Sample format non available
[03:01] <Aziroshin> "
[03:15] <Aziroshin> I found out about alsaloop. It indeed is the hardware.
[07:45] <killown> can't get audio with this ffmpeg -f alsa -ac 2 -i pulse -f x11grab -r 30 -s 1920x1080 -i :1.0 -acodec pcm_s16le  -vcodec libx264 -preset ultrafast -crf 0 -threads 0 nop.mkv   (trying to record screen)
[07:45] <killown> does I need specify my mic?
[07:49] <zap0> what is the simplest container format for video that is easy to parse?
[08:26] <shevy> hmmm .avi ?
[09:30] <cbsrobot> shevy: subtil joke !
[10:56] <elkng> what codec is better to use: H.264 or VP8 ?
[12:03] <burek> elkng, it depends on what beer you like the most :)
[12:03] <burek> killown,
[12:03] <burek> can you please use pastebin.com, to show your command line and its output?
[12:07] <elkng> what is "killown" ?
[12:08] <sacarasc> killown is a member of the channel.
[12:08] <elkng> what gets less file size with same quality H.264 or VP8 ?
[12:10] <Mavrik> currently H.264 has better encoders, so H.264 leads slightly
[12:13] <the_p> hi. i would like to remove only a few frames of the end of a video. Can i do this with ffmpeg?
[12:15] <burek> the_p, yes you can :)
[12:16] <elkng> is it possible to remove some frames or replace them with images like replace frame 500th with image in file image.jpg ?
[12:17] <burek> the_p, all you need is this: ffmpeg -i input.avi -c copy -t <specify_duration_in_seconds> output.avi
[12:18] <burek> duration can be specified as hh:mm:ss.ttt
[12:18] <burek> where ttt are milliseconds
[12:18] <burek> so -t 0.5 will be half a second
[12:18] <burek> or -t 01:30:00 if you want an hour and a half long video
[12:19] <burek> elkng, yes it is, use video filters
[12:51] <Neil1> Morning folks - I'm having an issue compiling ffmpeg on the latest Ubultu server version I downloaded last night.  I'm following the instrustions on https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide but the ./configure section of the FFmpeg compile returns an error - ERROR: libfaac not found.  Sadly I have no idea how to either modify a configure file or install libfaac so thought I'd ask here first for some help
[12:53] <burek> well, that's not too hard Neil1
[12:53] <burek> did you ever compile anything on your own?
[12:54] <Neil1> this is my first time - please be gentle with me
[12:54] <burek> :)
[12:55] <burek> http://www.audiocoding.com/downloads.html
[12:55] <burek> go there and download FAAC Source	Version 1.28 ZIP Package
[12:55] <burek> unpack it into a new folder and just do: ./configure
[12:55] <JEEB> faac is not really GPL (uses non-GPL source code as base), and thus it was removed from many distributions
[12:55] <burek> then do make and make install
[12:56] <Neil1> can i do that with git, as per all the other things I've downloaded so far?
[12:56] <JEEB> nah, it has no git repo
[12:56] <burek> sadly faac doesn't use git
[12:56] <Neil1> ok
[12:56] <JEEB> there hasn't been any work on faac in years :/
[12:56] <JEEB> at least five or so methinks
[12:56] <the_p> hi. i finally managed to get my h264 video running in flowplayer. but there is a problem that the index seems to be at the end of the file which means the whole video needs to be downloaded before it can be played. is there a way to make ffmpeg write the index at the beginning of the file?
[12:57] <burek> Neil1, if you didn't ever compile anything, check here: http://ffmpeg.gusari.org/viewforum.php?f=25
[12:57] <JEEB> the_p, compile the qt-faststart app in the tools folder :)
[12:57] <burek> it might be of some help
[12:57] <JEEB> it moves the index to the beginning
[12:58] <burek> the_p, when you encode your video, do it like this: ffmpeg -i input.avi -vcodec libx264 -acodec ... <other params> -flags +global_header -f flv out.flv
[12:59] <the_p> burek: i'm coding to mp4 not flv does this make a difference? sorry for the maybe supid questions i'm doing this the first time
[12:59] <JEEB> the_p, see what I said
[12:59] <JEEB> what I said was for mp4
[12:59] <JEEB> flv shouldn't need this anyways IIRC
[12:59] <the_p> ok thanks
[12:59] <elkng> burek: "did you ever compile anything on your own?", he says he using "Ubultu"
[12:59] <JEEB> there's also a python script called qtfaststart
[12:59] <JEEB> which does the same thing
[12:59] <JEEB> and then there's the remuxer tool in L-SMASH
[12:59] <burek> :D
[13:00] <JEEB> there's plenty of ways to deal with the index atom in mp4 :)
[13:00] <burek> the_p, it's the same for mpg
[13:00] <burek> mp4*
[13:00] <burek> the_p, when you encode your video, do it like this: ffmpeg -i input.avi -vcodec libx264 -acodec ... <other params> -flags +global_header -f mp4 out.mp4
[13:00] Action: JEEB goes check if what burek says is actually true :P
[13:01] <JEEB> because I don't think that does what he thinks
[13:01] <the_p> i now used the qt-faststart application i'll see soon if it worked
[13:02] <burek> that will also help
[13:03] <elkng> the_p: use python version
[13:03] <JEEB> both should work
[13:03] <elkng> python wont crash
[13:05] <JEEB> yeah, I had read that supposedly the python version was made better
[13:05] <JEEB> https://github.com/danielgtaylor/qtfaststart
[13:06] <JEEB> but if it didn't crash it should've worked just fine
[13:06] <JEEB> then there's remuxer/muxer apps from L-SMASH that are pretty good too for general mp4/mov/3gp playing
[13:07] <the_p> thanks a lot it worked
[13:09] <Neil1> sorry guys - it's been a while since i used a unix system - $ tar -xvz ./faac-1.28.zip isn't working for me - what am I doing wrong?
[13:09] <JEEB> Neil1, just use the unzip app
[13:09] <JEEB> not sure if tar supports dot-zip
[13:10] <JEEB> and I think you don't have to set the type manually any more, 'tar xvf filename.tar.*' should work
[13:10] <JEEB> also someone tell burek that flags +global_header doesn't do what he thinks it does http://git.1f0.de/gitweb?p=ffmpeg.git&a=search&h=HEAD&st=grep&s=CODEC_FLAG_GLOBAL_HEADER
[13:10] <JEEB> because he has chosen to ignore me
[13:11] <burek> Neil1, it's: unzip bla.zip
[13:11] <burek> the archive is not .tar.gz
[13:11] <burek> so, mkdir tmp1
[13:11] <burek> cd tmp1
[13:11] <burek> unzip bla.zip
[13:11] <elkng> unzip faac-1.28.zip
[13:11] <Neil1> got it - thanks guys - as I say, it's been a while...
[13:17] <JEEB> also someone tell burek that the +global_header flag does not remux an mp4/mov/3gp after the fact and put the index atom into the beginning because he has chosen to ignore me? :P (at least looking at what git grep "CODEC_FLAG_GLOBAL_HEADER" gives) -- him giving incorrect help is herp derp.
[13:18] <Neil1> ok - there is no 'configure file in /faac-1.28 - but there is a configure.in - anyway, does libaacplus, listed here accomplish the same thing I'm trying to do here?  http://ffmpeg.gusari.org/viewtopic.php?f=25&t=37&sid=a5cc82661ff24791c0ce0c810921cf56 - it seems 'easier'?
[13:19] <JEEB> uhh, you sure there wasn't a pre-configure-created tarball of faac?
[13:19] <JEEB> I think there was
[13:19] <JEEB> yes
[13:19] <JEEB> "Version 1.28 bootstrapped TAR.BZ2 Package"
[13:19] <JEEB> that is what you want
[13:20] <burek> Neil1, is there any autogen.sh or something?
[13:20] <JEEB> him not seeing what I write is pretty herp derp :P
[13:20] <JEEB> also, libaacplus is pretty much the same, it just encodes he-aac, not "normal" (LC) aac
[13:20] <burek> well aacplus might seem easier to install but it's a non-free library
[13:21] <JEEB> they're are BOTH non-free you idiot
[13:21] <Neil1> well sudo apt-get install faac seems to have accomplished something -
[13:21] <burek> so, it depends on what you need your ffmpeg for
[13:21] <burek> Neil1, you need libfaac-dev or faac-dev
[13:21] <JEEB> yes, you'll need the development libraries and headers
[13:22] <Neil1> well I'm looking to transcode eithe windows media9 files (old I know, but it's what i have to work with) or apple mov foles to h264
[13:22] <JEEB> for most things, just get faac
[13:22] <JEEB> LC-AAC is what you want
[13:22] <burek> Neil1, there is one more simple thing you might try
[13:22] <burek> http://ffmpeg.gusari.org/static/
[13:22] <burek> try it, and if it works, there you go
[13:23] <burek> if not, you can continue with compiling
[13:24] <JEEB> ezoe, can you tell burek as he has chosen to ignore me that A) the +global_header flag does not remux an mp4/mov/3gp after the fact and put the index atom into the beginning because he has chosen to ignore me? :P (at least looking at what git grep "CODEC_FLAG_GLOBAL_HEADER" gives)
[13:24] <JEEB> and B) both faac and libaacplus are nonfree
[13:26] <Neil1> well apt-get install faac seemed to have run but the ffmpeg ./configure is still giving the same 'libfaac not found' error - I'll give the compile a go now
[13:28] <JEEB> yes, because you have no libraries or headers
[13:28] <JEEB> which come from the -dev package
[13:28] <JEEB> apt-cache search faac
[13:29] <JEEB> (or find, don't remember)
[13:40] <Neil1> well I'm making progress - it seems none of the libraries or headers are installed so I'm going through and installing them now
[13:45] <Neil1> ok - next question - the last 'enable' is 'enable-x11grab' and I get an error from the ./compile of 'X11 not found'  Now there's loads of X11 dev files so assuming /i need one (I thought X11 was graphics?) so which one do I need.  I'm running all of this in a shell - I have no GUI
[13:46] <burek> x11grab-dev?
[13:51] <burek> hm, it seems you will need: libx11-dev, libxext-dev, libxfixes-dev
[13:51] <burek> so, try: apt-get install libx11-dev libxext-dev libxfixes-dev
[13:52] <burek> (I got this from reading the file ffmpeg/src/configure and searching for x11grab)
[13:53] <Neil1> ah - thanks for the tip on where to look!  installing now...
[13:53] <burek> ok :)
[13:56] <Neil1> running 'make' :-)
[13:57] <Neil1> getting lunch - this looks like it'll be a while - thanks again folks
[13:58] <burek> if you have multiple cpu cores, you can use -j
[13:58] <burek> like make -j 5
[13:58] <burek> or something
[13:58] <burek> (cores+1)
[14:13] <Neil1> Ok- all seemed to compile and install :-)
[14:14] <Neil1> the line 'hash x264 ffmpeg ffplay ffprobe' gives an error 'hash not found' and I can't find a package called hash - do i need this and where would I get it?
[14:15] <sacarasc> It's part of Bash, I think...
[14:16] <burek> just hash -r
[14:16] <burek> if it doesn't exist, even better
[14:17] <burek> it just clears the cache of previously executed programs
[14:17] <burek> so it runs the fresh ffmpeg and not some remnants of old one in cache
[14:17] <Neil1> well seeing as it wasn't installed before I'll soldier on
[14:25] <elkng> someone said: "fact and put the index atom into the beginning the +global_header flag does not remux an mp4/mov/3gp after the fact and put the index atom into the beginning"
[14:25] <elkng> and "both faac and libaacplus are nonfree"
[14:25] <elkng> is that true ?
[14:26] <elkng> "the +global_header flag does not remux an mp4/mov/3gp after the"
[14:30] Action: burek looks curiously at fflogger
[14:30] Action: burek feeds fflogger :)
[14:30] <JEEB> just git grep for "CODEC_FLAG_GLOBAL_HEADER" and see where it's used :P
[14:30] <burek> :)
[14:31] <JEEB> (and you will also get the meaning as well)
[14:31] <burek> I guess bots also need some love ^^
[14:31] Action: shevy makes love to burek
[14:37] <zap0> is there a pix_fmt for 10bit packed?   like r210 format
[14:43] <JEEB> yuv420p10 was the planar one, right?
[14:44] <zap0> probably,  i want rgb not yuv
[14:45] <elkng> is yuv better than rgb ?
[14:45] <JEEB> I wonder if that's how ffmpeg feeds libx264 RGB
[14:45] <JEEB> elkng, completely different colorspaces
[14:45] <zap0> elkng, its not better. its different.
[14:46] <zap0> thus making it better for different situations.
[14:46] <JEEB> most digital video that's available to people is YCbCr (which is then collorquially called 'YUV')
[14:46] <elkng> photo edition is better in YUV ?
[14:46] <JEEB> wat
[14:46] <JEEB> zap0, anyways -- check how ffmpeg feeds libx264 RGB
[14:46] <zap0> elkng, no.
[14:47] <zap0> JEEB, interesting idea.   i have no idea how to 'check'
[14:48] <elkng> professional photo editors support YUV that makes them better than gimp that doesnt support it
[14:48] <JEEB> zap0, also wasn't RGB a packed format to begin with? Unless I misremember
[14:49] <zap0> gimp is lame (at least last time i checked)  it still does 8bit per channel.  thats aweful.
[14:49] <zap0> JEEB,  ffmpeg -pix_fmts       can you see anything rgb10 there?
[14:49] <JEEB> oh right
[14:49] <JEEB> 10bit
[14:49] <JEEB> yeah... not so sure 'bout that
[14:50] <JEEB> ok, I see a 16bit RGB format
[14:50] <JEEB> at least I think so
[14:51] <zap0> yeah, i can do  rgb48le  but thats 16bits
[14:51] <zap0> i need 10.
[14:52] <zap0> there are plenty of 10bit codecs; but i can't see how you specifiy the pixel format
[14:53] <JEEB> I see 10bit YCbCr just fine (as well as 9bit)
[14:53] <JEEB> not so much about rgb
[14:56] <zap0> this chart http://www.mplayerhq.hu/DOCS/codecs-status.html   search for 'r210'  you see about 3 entries..  the last one indicates RGB32
[14:56] <zap0> but that doesn't appear in the   ffmpeg -pix_fmts   list
[14:56] <JEEB> you probably shouldn't use mplayer's docs for ffmpeg's pix_fmts :)
[14:57] <JEEB> git grepping through ffmpeg's repository sounds like a better idea
[14:58] <JEEB> too bad the official gitweb for ffmpeg doesn't let you do git grep
[14:58] <JEEB> so you either have to clone it and use it locally
[14:58] <zap0> ffmpeg -codecs   has  r210 "uncompressed RGB 10bit"
[14:58] <JEEB> oh
[14:58] <zap0> so what is its pix_fmt?
[14:59] <JEEB> lessee
[14:59] <JEEB> avctx->pix_fmt             = PIX_FMT_RGB48;
[14:59] <JEEB> from libavcodec/r210dec.c
[14:59] <zap0> yeah, i know mplayer doc is wrong source;  its the only source i could find. and its really just hinting at the issue/problem.
[14:59] <JEEB> so inside ffmpeg it's dealt as 16bit RGB
[14:59] <JEEB> and then down/upconverted when encoding/decoding
[15:03] <zap0> the task:     i have some typical 8bit-per-channel videos; some rgb, some yuv, etc..   i want to output an  uncompressed rgb10 file.  that would be 32bit per pixel  2bit-zero,10-bit-red, 10bit-green, 10bit-blue.
[15:03] <zap0> presuming this  r210  codec is  uncompressed 10bt RGB.    what would my command line be?
[15:04] <JEEB> I'm not exactly sure how the 10bit picture is saved in r210 technically, but it is indeed a 10bit RGB format
[15:04] <JEEB> set vcodec or c:v to r210
[15:05] <JEEB> and it will be encoded to that format
[15:05] <zap0> ffmpeg -i file.mp4 -vcodec r210 out.?
[15:05] <JEEB> out dot avi or something
[15:06] <JEEB> or mkv or something
[15:06] <JEEB> depends on what you exactly want
[15:06] <JEEB> + you have to deal with audio as well in most cases methinks so set acodec/c:a accordingly
[15:06] <zap0> ok,  part 2)  what is the simplest container to parse.
[15:06] <zap0> think project does not need audio.
[15:06] <zap0> this proj/
[15:07] <JEEB> then -an
[15:07] <JEEB> no audio
[15:08] <zap0> is there a "no-container" option?
[15:08] <zap0> its trying to derive the container type from the file extention
[15:09] <zap0> -f rawvideo    ok, now something has come out..  does that sound about right
[15:10] <zap0> lemme just calc the size if correct
[15:10] <killown> can't get audio with this ffmpeg -f alsa -ac 2 -i pulse -f x11grab -r 30 -s 1920x1080 -i :1.0 -acodec pcm_s16le  -vcodec libx264 -preset ultrafast -crf 0 -threads 0 nop.mkv   (trying to record screen)
[15:11] <killown> does I need specify my mic?
[15:11] <zap0> w00t!  it adds up to 4bytes per pixel.   i think its working!!!
[16:19] <raven> how to add new audio streams to a file? what is the new option istead of -newaudio?
[16:33] Last message repeated 1 time(s).
[16:34] <JEEB> check the documentation on mapping
[16:34] <JEEB> http://ffmpeg.org/ffmpeg.html#Stream-selection
[16:38] <raven> JEEB where exactly is the right chapter
[16:43] <raven> how to add new audio streams to a file? what is the new option istead of -newaudio?
[16:49] <ubitux> you don't need anything particular, maybe -map
[16:51] <raven> but how?
[16:51] <ubitux> what are you trying to achieve?
[16:51] <raven> i always get only one audio stream
[16:51] <raven> i need one video and 6 audio streams
[16:51] <raven> -i video -i a1 -i a2 ........
[16:52] <ubitux> try ffmpeg -i video -i a1 -i a2 -i a3 -map 0:0 -map 1:0 -map 2:0 -map 3:0 out.mkv
[16:53] <raven> i did but i still get only a1 and nothing more
[16:53] <ubitux> (-map <input-file>:<input-stream>)
[16:53] <ubitux> mmh
[16:54] <ubitux> raven: can you paste the complete command line and output on pastebin?
[16:55] <Mista_D> How does one preserve closed captioning(CEA-608 aka Line21)? Encoding MPEGTS(MPEG2/MP2) to MPEGTS(H.264/AAC).
[16:55] <raven> ffmpeg -i 7.dv -i 1.mp3 -i 2.mp3 -i 3.mp3 -i 4.mp3 -i 5.mp3 -i 6.mp3 -i 7.mp3 -vcodec mpeg2video -b:v 4000k -aspect 16:9 -acodec copy -acodec copy output.mkv -map 0:0 -map 1:0 -map 2:0 -map 3:0
[16:55] <raven> acodec and map is not complete but its only for testing
[17:02] <raven> ubitux ?
[17:02] <ubitux> raven: move the map before the output.mkv
[17:03] <ubitux> (you have 2x -acodec copy, you can drop one)
[17:03] <raven> sry so simple things take so long headache :)
[17:07] <ubitux> try using -map -1 btw
[17:07] <ubitux> (or -map 0 i always forget)
[17:08] <ubitux> instead of the multiple -map
[17:08] <raven> ok it works now
[17:08] <raven> tnx very much
[17:08] <ubitux> and btw, it's simple to avoid these kind of error as long as you keep in mind the syntax; ffmpeg <input-options> -i input <output-options> output
[17:09] <ubitux> raven: does "ffmpeg -i 7.dv 1.mp3 -i 2.mp3 -i 3.mp3 -i 4.mp3 -i 5.mp3 -i 6.mp3 -i 7.mp3 -vcodec mpeg2video -b:v 4000k -aspect 16:9 -acodec copy -map 0 output.mkv" works?
[17:10] <raven> no then its only 1video and 1 audio again
[17:11] <ubitux> mmh
[17:11] <ubitux> ah rien, it's all streams from file 0, my bad.
[17:11] <ubitux> right*
[19:04] <killown> can't get audio with this ffmpeg -f alsa -ac 2 -i pulse -f x11grab -r 30 -s 1920x1080 -i :1.0 -acodec pcm_s16le  -vcodec libx264 -preset ultrafast -crf 0 -threads 0 nop.mkv   (trying to record screen), anyone help me?
[19:04] <killown> I am trying to record a gameplay
[19:11] <avocado> killown: look at the man page for alsa or pulse. i don't think you're using it correctly.
[19:12] <avocado> you should have something like -f alsa -i hw:0 or -f pulse -i default (according to man page)
[19:15] <killown> avocado, thanks a lot, worked wiith -f pulse -i default
[19:34] <madwill> so turns out ffmpeg needs to be compiled with --enable-libpulse to allow for pulse audio capture when using x11grab ?
[19:34] <killown>  File for preset 'lossless_ultrafast' not found
[19:34] <killown> :/
[19:34] <madwill> <haha it does not exist anymore
[19:35] <madwill> killown don't use -vpre lossless_ultrafast  you now use -preset ultrafast with -crf 0 for lossless
[19:36] <killown> madwill, I cant understand the lag when I use this ffmpeg -f pulse -i default -ac 2 -f x11grab -r 30 -s 1920x1080 -i :0.0 -acodec pcm_s16le  -vcodec libx264 -preset ultrafast -crf 0 -threads 0 twoworlds.mkv
[19:37] <madwill> well its as fast as it gets to me. maybe you can use pure lossless mpg or something crazy that gives multiple gb per minutes
[19:37] <madwill> or  lower the game graphics
[19:37] <madwill> have you looked at your system monitor seeing if you're at 100%
[19:38] <madwill> And for the sound, i'm doing the compile with libpulse to see if i can get some GOD DAMN AUDIO for myself
[19:38] <madwill> pulse seems to be pretty ok to me and shows activity but i only ever get a single second of sound
[19:38] <madwill> oh its ready BRB
[19:52] <killown> madwill, :/
[19:53] <killown> ffmpeg -f pulse -i default -ac 2 -f x11grab -r 30 -s hd1080 -i :0.0 -acodec pcm_s16le -b:a 128k  -vcodec libvpx -quality good -cpu-used 0 -b:v 1200k -maxrate 1200k -bufsize 2400k -preset ultrafast -crf 0 -threads 4 twoworlds.mkv
[19:53] <killown> a lot lag just  recording desktop
[19:55] <killown> can't believe my phenon x4 955 overclocked at 3.8ghz + gpu gtx 580 3gb cant hold this
[19:56] <killown> I am only testing the desktop has no game opened
[19:56] <killown> and still lags
[19:58] <killown> hu the problem is the codec
[19:58] <killown> libvpx
[19:58] <killown> rawvideo is what I need
[19:59] <madwill> yeah phemon isen't actually a very good cpu
[19:59] <madwill> depending on the game you run in the background
[19:59] <madwill> Raw video is what you are looking for
[19:59] <madwill> btw compiling with --enable-libpulse is pretty much what you need to get sound
[20:01] <killown> Unknown encoder 'h264_vdpau'
[20:03] <killown> using a vdpu codec should fix the high cpu usage
[20:04] <killown> Unknown encoder 'mpegvideo_vdpau'
[20:04] <killown> doesnt work at all
[20:09] <Mavrik> why are you trying to encode with library meant for decoding?
[20:41] <pisto> hi. What could be an efficient codec for a video composed of simple circles and bold lines moving around?
[20:43] <pasteeater> pisto: probably libx264, but of course it depends on how your planning on paying the output, etc.
[20:43] <pasteeater> *playing
[20:45] <pisto> hm, vlc? what kind of info you're asking?
[20:48] <pasteeater> you didn't give many specifics or restrictions, so that's why i said "probably".
[20:51] <pasteeater> example: ffmpeg -i input -c:v libx264 -preset medium -crf 24 output.mkv
[20:52] <pasteeater> general usage is to use the slowest preset you have patience for and the highest crf value that still gives an acceptable quality
[20:55] <pasteeater> if there is audio involved you can copy it or re-encode it
[20:57] <Pieplay> I have recently bought a dlna mediaserver, and have been testing various commercial and opensource ways to get the dvd to a file
[20:58] <Pieplay>  the winff batcher for ffmpeg gave the best results
[20:58] <Pieplay> but i'm losing alot of quality ripping the dvd to a file before the convertion
[20:59] <Pieplay> is there a lossless way to do this
[21:04] <pasteeater> Pieplay: how are you ripping the dvd?
[21:06] <killown> I am using this (ffmpeg -f pulse -i default -ac 2 -f x11grab -r 25 -s hd1080 -i :0.0 -acodec pcm_s16le -b:a 128k  -vcodec libx264  -quality good -cpu-used 0 -b:v 1200k -maxrate 1200k -bufsize 2400k -preset ultrafast -crf 0 -threads 4 twoworlds.mkv )  and every 8 frames console shows a small stop, maybe its a write disk problem?
[21:06] <Pieplay> Tried VLC MPEG2/MKV, H264/MKV,Copy/MKV and formatfacory on windows with about the same settings, but they all seem to convert the file and quality seems to be lost
[21:09] <pasteeater> Pieplay: handbrake might be easiest to use and has decent x264 settings, but i am not a windows user and have no other suggestions for you
[21:10] <Pieplay> Well see I found how to make a good x264 file with ffmpeg
[21:11] <Pieplay> on windows ffmpeg because this system has a better cpu
[21:12] <Pieplay> Just need to be able to make the dvd into a file without loosing all quality,
[21:12] <pasteeater> i don't know, but -quality and -cpu-used are libvpx options, not libx264
[21:12] <pasteeater> killown: ^
[21:12] <killown> ok
[21:12] <Pieplay> Linux is no problem for me
[21:12] <killown> pasteeater, I am mount 1GB ram to the disk
[21:13] <killown> mounting*
[21:14] <pasteeater> consider removing -threads 4. default behavior is -threads 0 (does not mean "no threads")
[21:14] <killown> tmpfs           2,0G   48M    2,0G   3% /home/killown/ramdisk
[21:15] <pasteeater> Pieplay: vobcopy or tccat are two linux options
[21:15] <killown> I am using ramdisk but still lags so I think I am missing something on ffmpeg
[21:17] <killown> pasteeater, I am using now ffmpeg -f pulse -i default -ac 2 -f x11grab -r 25 -s hd1080 -i :0.0 -acodec pcm_s16le -b:a 128k  -vcodec libx264  -quality good -cpu-used 0 -b:v 1200k -maxrate 1200k -bufsize 2400k -preset ultrafast -crf 0 -threads 0 twoworlds.mkv and still lags
[21:17] <Pieplay> Okay giving vobcopy a try
[21:17] <killown> the cpu usage average is 70%
[21:17] <killown> 70% of one core
[21:18] <killown> every 3 seconds the video does a little stop
[21:19] <killown> I am using memory ram /ramdisk so its not a write bandwidth issue
[21:19] <pasteeater> how do you know the issue isn't with your player?
[21:19] <killown> pasteeater, vlc and mplayer?
[21:19] <killown> and if I upload the video to youtube the same?
[21:20] <pisto> can I feel free to encode my h264 in mkv video with any framerate? or are there known glitches with vlc/windows media player?
[21:20] <killown> my video card is one the most power of the world, its gtx 580 light extreme, and also I compiled ffmpeg to use vdpau
[21:20] <pasteeater> what i mean is that lossless -r 25 -s hd1080 might be too much for your computer to handle encoding and/or decoding
[21:20] <killown> hmmm
[21:21] <Srijan> hi, i am new to ffmpeg and using stream audio and video with it in real time. I have successfully streamed audio  played using ffplay. I am also able to stream video on rtp, but while playing i get error
[21:21] <Srijan> Unable to receive RTP payload type 96 without an SDP file describing it
[21:21] <Srijan> can someone help me out?
[21:22] <killown> pasteeater, -r 15 still lags
[21:23] <killown> pasteeater, rawvideo should fix it?
[21:28] <pasteeater> "-quality good -cpu-used 0 -b:v 1200k -maxrate 1200k -bufsize 2400k" are being ignored. VBV is incompatible with constant QP and b:v and crf are mutually exclusive. quality and cpu-used are libvpx encoder options, not libx264. removing them wont make a difference since they are ignored, but you shouldn't use them in this case.
[21:29] <Srijan> any one can help?...its urgent
[21:44] <killown> pasteeater the lag is due to the audio record
[21:44] <killown> I have no lag with ffmpeg -f x11grab -s hd1080 -r 25 -i :0.0 -preset ultrafast twoworlds.mkv
[21:54] <killown> pasteeater, sound card problem or I am still  missing something
[21:55] <killown> pasteeater, I think I need separate the audio from the video during the recording, but I don't know how to do that
[21:57] <beastd> killown: what happens if you only record the audio? is it playing back correctly?
[21:58] <killown> beastd, I am recording audio with this ffmpeg -f alsa -ac 2 -i pulse   -acodec pcm_s16le -vcodec libx264 -b:a 40024k  -preset ultrafast -threads 0  -y  test.mp3
[21:58] <killown> has no problem
[21:58] <killown> high quality record
[21:59] <beastd> hmm
[21:59] <beastd> no glitches and no stutter?
[21:59] <killown> beastd, just perfect
[21:59] <killown> same to the video
[22:00] <killown> but when I try both on one container it lags
[22:00] <killown> cpu usage average is 70% of one core
[22:00] <killown> I am using a phenon x4 955
[22:00] <killown> ffmpeg compiled with vdpau
[22:00] <beastd> killown: not sure why it doesn't work together than. could be some interdependency. or maybe something with the timing causes the problem.
[22:01] <killown> beastd, the audio has no lag but the video has a frame drop every 3 seconds
[22:04] <beastd> killown: did you try to turn on ffmpeg verbous output? look here for a description http://ffmpeg.org/bugreports.html how to turn it on.
[22:05] <beastd> killown: you could try to see if there are any hints in the output to why frames are missing/dropped or similar
[22:12] <killown> [matroska @ 0xa31c440] Writing block at offset 2488384, size 10890, pts 4080, dts 4080, duration 40, flags 0
[22:12] <killown> *** drop!
[22:13] <killown> beastd, the lag is on this drop
[22:13] <killown> beandog, sorry, not in thins
[22:14] <killown> this*
[22:14] <beastd> killown: ?
[22:16] <killown> I am going to show it in video
[22:17] <beastd> killown: i have one more idea for a test you could carry out.
[22:18] <beastd> killown: you said if you leave out the audio input your video records fine, right?
[22:19] <killown> beandog, yes
[22:19] <beastd> *beastd
[22:20] <killown> sorry
[22:20] <beastd> killown: try to start the video record and an independent audio record (another ffmpeg process and another output file) at the same time
[22:20] <killown> ok
[22:21] <beastd> check if the video only file contains frame drops
[22:24] <killown> beastd, not even one framedrop
[22:24] <killown> both are being executed individually
[22:24] <killown> know has no lag in video
[22:24] <killown> and I am recording the audio to
[22:24] <beastd> killown: ok
[22:25] <beastd> killown:  then it is the combination that causes trouble. could e.g. be caused by timing related issues. but that is just a wild guess.
[22:26] <killown> beastd, very weird, so I will need use two container and after join them
[22:27] <beastd> killown: would be a workaround i think. resulting positive/negative delay may be a problem
[22:28] <beastd> you could try to start both processes quickly after each other with a short shell script
[22:29] <beastd> then mux the video and audio into a single file using ffmpeg again after the recording
[22:29] <beastd> see if the result suits your needs.
[22:29] <killown> I think there is a way to tell ffmpeg to use separated containers audio  < > video during screen record
[22:30] <beastd> killown: this is possible too. try it out. i just guess it will drop video frames too. but i maybe wrong...
[22:31] <killown> beastd, I think the issue is the ffmpeg need deal with video and audio sync at realtime
[22:32] <killown> don't know the syntax to split it
[22:33] <killown> ffmpeg -f pulse -i default -i audio.mp3 -f x11grab -s hd1080 -r 25 -i :0.0  -b:a 128k -preset ultrafast twoworlds.mkv  doesnt work
[22:33] <killown> -acodec copy ouputfile.mp3
[22:33] <killown> Unknown decoder 'copy'
[22:38] <killown> ffmpeg -f x11grab -s hd1080 -r 25 -i :0.0 -preset ultrafast -y video.mkv &&  ffmpeg -f alsa -ac 2 -i pulse   -acodec pcm_s16le -vcodec libx264 -b:a 48000k  -preset ultrafast -threads 0  -y  twoworlds2.mkv
[22:38] <killown> killall ffmpeg to terminate both at the same time
[22:40] <beastd> killown: if you t to try recording to different files with one command you could tr
[22:41] <beastd> sorry, if you still are willing to try using one command writing to output files you could try this
[22:45] <killown> ?
[22:45] <beastd> killown:   ffmpeg -f alsa -ac 2 -i pulse -f x11grab -s hd1080 -r 25 -i :0.0 -vcodec libx264 -preset ultrafast -an -y video.mkv -acodec pcm_s16le -b:a 48000k -vn -y twoworlds2.mkv
[22:46] <beastd> killown: you need to double check codec options and stuff i just juggled around the pieces of your command line
[22:59] <beastd> killown: could you record to different files with a single ffmpeg command? did the video file have frame drops again?
[22:59] <killown> beastd, no lag now, thank you
[22:59] <killown> I bet this is a stdout issue
[23:00] <beastd> killown: my command worked?
[23:00] <killown> with your command line has no flood
[23:00] <killown> beastd, yes, both audio and video recorded with no lag
[23:01] <beastd> killown: that's good news. hope you can do now whatever you wanted to do
[23:01] <beastd> killown: but could you please tell me about the stdout flooding?
[23:02] <killown> beastd, when the logging and verbose is enabled, it flood as much as the console can handle so happens this little stops that affect the frame write to the disk
[23:03] <killown> even with no verbose and no console flood I think there is a flood happening on stdoud
[23:04] <beastd> killown: what is printed with no verbose?
[23:05] <killown> I will show you what happen
[23:06] <killown> beastd, http://www.youtube.com/watch?v=h4J0HjOd_0w
[23:06] <killown> every 3 seconds the flood get a little stop
[23:06] <killown> and it affect the video frames
[23:07] <beastd> oh
[23:08] <killown> beastd, but its because the verbose and logging enabled
[23:08] <beastd> and without it it works well?
[23:08] <killown> no
[23:09] <killown> has no flood and video stills lags
[23:09] <killown> because that I think there is a flood on  stdout  in somewhere
[23:10] <beastd> i guess it is the same lag if you use -loglevel quiet
[23:10] <beastd> but you could quickly try
[23:12] <killown> beastd, with loglevel quiet still lags
[23:13] <beastd> killown: maybe you could report this as a bug on our tracker?
[23:16] <beastd> killown: be sure to include complete information. describing the frame drops and including the log of both commandlines audio/video grabbing to one file and audio/video grabbing into 2 files should be enough.
[23:16] <killown> ok
[23:16] <killown> beastd, thanks for helping me
[23:16] <beastd> killown: np. have fun
[23:17] <beastd> bye...
[00:00] --- Fri May 18 2012