[Ffmpeg-devel-irc] ffmpeg.log.20120205

[burek]

[00:00] <brontosaurusrex> jackNDbox, it depends on what you want to do later, i just : --enable-gpl --enable-libx264 --enable-libvpx
[00:01] <jackNDbox> i'm building a porn site and a video sharing site
[00:01] <jackNDbox> using phpmotion
[00:02] <jackNDbox> for one of em
[00:03] <brontosaurusrex> i mean ffmpeg, do you need mp3 encoder for porn? or theora? probably not
[00:03] <jackNDbox> yea youre probably right. ffmpeg automatically does audio .. right? just in a lower quality?
[00:04] <jackNDbox> I was just being lazy following that guide above
[00:06] <brontosaurusrex> well, video sites this days usually encode to AVC/AAC in some form or another, webm (You would probably want aotuv oggenc here anyway) is also somehow interesting
[00:07] <brontosaurusrex> i don't know anybody that would use xvid/mp3 this days
[00:07] <brontosaurusrex> only my opinion...
[00:09] <brontosaurusrex> p.s. actually even libvpx is kinda useless, its almost impossible to find correct command line, better to just pipe from ffmpeg to vpxenc anyway
[00:10] <jackNDbox> makes sense.. the stuff before that
[00:10] <jackNDbox> thanks for the info i'll be back here probably in a little while
[00:10] <jackNDbox> i'm gonna go get a pizza =)
[00:10] <iive> jackNDbox: look for the phrase "mp3lame" in that log file
[00:22] <brontosaurusrex> how to encode to 444 prores, command line?
[02:47] <Ginks> any idea why ffmpeg would seg fault with "ffmpeg -i 253_temp.flv -vf "movie=/home/mypetman/public_html/images/watermark.png [watermark]; [in][watermark] overlay=10:10 [out]" -b:v 472k 253_w.flv"
[02:47] <Ginks> ?
[02:47] <Ginks> only happens with flv videos
[02:51] <edrx> hello - how do I reduce the number of frames of a video? it's for recording a conference with few slides and a speak that does not move too much... the camera produced a .mov with 24 fps, and I'd like to reduce that to 8 fps by keeping only 1 in each 3 frames...
[02:52] <edrx> s/speak/speaker/
[02:54] <edrx> (btw, I am on debian, and using the command line)
[03:14] <rypervenche> Hi all. My screencasting script broke a while back, apparently due to an update in ffmpeg. It uses x264, however at first I record lossless data. I have since fixed my script using "-vcodec libx264 -preset ultrafast", but is this lossless data? Should I be using somethign else?
[03:16] <rypervenche> Before I was using -vpre lossless_ultrafast. Is there still a lossless option or should I only use the ultrafast option?
[03:38] <iive> rypervenche: just a moment
[03:42] <rypervenche> iive: Thank you.
[03:45] <iive> it will take a little bit more time to compile 0.10 release. in the meantime
[03:45] <iive> you can get lossless by encoding with quantizer 0 in x264
[03:46] <Ginks> qmax, qmin, qscale?
[03:47] <Ginks> I'm going to take a video and add an overlay to it
[03:47] <iive> either by -qscale 0  (that's ffmpeg option ) or -x264opts qp=0
[03:47] <rypervenche> iive: So they removed the lossless preset then? This is the only way to get a lossless video?
[03:48] <Ginks> just want to get the best quality prior to encoding it to 264
[03:48] <Ginks> size doesn't matter, because I'll be encoding it with 264
[03:48] <iive> rypervenche: seems so, the .ffpreset files are gone
[03:49] <iive> there is x264-ipod*.ffpreset
[03:49] <iive> are
[03:50] <iive> hum, there is -qp 0 (ffmpeg option maped to x264 directly).
[03:52] <iive> hum, it says rate control method.... huh
[03:56] <iive> n8 ppl
[04:54] <pills> I have 4 videos I am trying to combine into a single video. This is easy enough to do using the video filter options. My problem is the input videos don't seem to decode properly (first frame is dropped). This seems to be due to how they were encoded wrt timecodes. I can fix this problem for a single video by adding the -vsync 0 option however when combined with the video filters only the first video seems to inherit the vsync option. Is
[04:54] <pills>  there anyway around this (reencoding is fine)? Thanks.
[05:14] <burek> try to reencode your videos
[05:14] <burek> to get valid streams
[05:14] <burek> and then use filter
[05:14] <burek> or even better, if you have room on your hdd, use uncompressed version :)
[05:15] <burek> and only compress back when using the filter
[05:15] <burek> that way you recompress only 1 time
[05:34] <pills> what params do I feed the encoder to ensure they will be decoded correctly?
[08:48] <Sazpaimon> mind if i ask a general streaming question around here?
[08:49] <Sazpaimon> (yes I know I'm asking to ask, sorry)
[12:38] <necktie> hi. can anyone help me with updating ffmpeg and x264? i compiled both from svn/git. is there an easy command that will get the up to date? or do i have to completely re-compile them?
[12:41] <tdr> yep, rebuild them
[13:02] <LexSfX> necktie: write a short script for yourself that simply performs the git and build commands you used in the first place.
[13:13] <necktie> LexSfX: alright, i will try. thanks
[14:35] <necktie> hi. i'm trying to compile ffmpeg but i get this error "WARNING: Please upgrade to VA-API >= 0.32 if you would like full VA-API support.". what does this mean and how can i fix it?
[14:39] <necktie> anyone?
[15:11] <iive> necktie: you need newer vaapi libraries.or newer distro
[15:11] <Mavrik> necktie, your va-api library is too old
[15:12] <Mavrik> and you won't be able to use nVidia hardware acceleration
[16:05] <dragos240> Hello. I'm having a few issues streaming through rtsp with ffserver. My config file is here http://pastebin.com/vmNghAhJ and the command I use to stream to the ffserver is here along with the error message I recieve http://pastebin.com/wQLuSw0r
[16:06] <dragos240> I've been troubleshooting for a while and I cannot figure out for the life of me why this hates me.
[16:07] <dragos240> Oh. And my ffmpeg version is the latest git revision that I checked out 30m ago. However stable versions act the same way
[16:21] <mystica555_> dragos240: i could be wrong, but i dont think mu-law or a-law support 44.1khz ?
[16:21] <dragos240> Hm
[16:21] <mystica555_> i thought they were 8 bit 8khz (with 11bits effective sample loudness resolution due to non-linear quantization scale)
[16:22] <dragos240> We are talking about the sample rate, correct?
[16:23] <mystica555_> mostly; also it doesnt seem that stream2.sdp is even being parsed by ffmpeg to create the codecs as you list (i dont know whether or not it even is supposed to pull from there, but if you notice in your debug output there are no -acodec or -vcodec bits, and its seemingly choosing mu-law randomly instead of libfaac)
[16:24] <mystica555_> nor is it using libx264 as the video codec
[16:24] <mystica555_> try adding -acodec libfaac -b:a your-bitrate -vcodec libx264
[16:24] <mystica555_> to your commandline
[16:25] <mystica555_> and probably -b:v 128K as well
[16:25] <mystica555_> -b:a 64K
[16:25] <mystica555_> or something
[16:25] <mystica555_> granted, hes ircing as root, that doesnt entirely show critical thinking skills
[16:25] <mystica555_> (which i find are required to parse logs)
[16:27] <dragos240> I noticed that.
[16:27] <dragos240> I don't know why it randomly does that.
[16:28] <dragos241> I kept getting disconnected
[16:29] <mystica555_> fwiw you may not want to IRC as root; if there is a backdoor or a known vulnerability in your client, anyone could just 0wn your system.
[16:29] <dragos241> Right.
[16:29] <dragos241> I'm not using root now, but I was.
[16:29] <mystica555_> i suggest adding commandline options to ffmpeg to force video/audio codecs; the SDP file is simply what the rtsp client pulls for its own config i believe
[16:30] <mystica555_> and bitrates for that matter
[16:30] <mystica555_> something like: -acodec libfaac -vcodec libx264 -b:a 64K -b:v 128K
[16:30] <dragos241> Hm.
[16:30] <dragos241> I'll see if that does anything
[16:30] <mystica555_> it should do a lot of things
[16:31] <mystica555_> because if you compare your SDP file to the actual debug output of ffmpeg, you arent matching on any counts
[16:32] <mystica555_> except for framerate and capture size, which you specify on the commandline
[16:32] <dragos241> Hm. Now this is an odd result
[16:32] <dragos241> Unknown decoder 'libx264'
[16:33] <dragos241> I could probably switch to theora and vorbis and see if that does anything.
[16:34] <Mavrik> libx264 is not a decoder
[16:34] <Mavrik> the decoder is probably called "h264"
[16:35] <dragos241> I'll try that
[16:36] <dragos241> Okay. I changed libx264 to h264 and libfaac to aac. I get a new error.
[16:36] <dragos241> [alsa @ 0x14dcda0] sample format 0x15002 is not supported
[16:36] <dragos241> hw:0,0: Input/output error
[16:37] <dragos241> It's complaining about sample format. Maybe the sample rate is to blame?
[16:39] <dragos241> Changing both config and command line instances of 22050 to 8000 does not fix the problem. But those may be invalid.
[16:43] <mystica555_> try opening the device as plughw: vs hw:
[16:43] <mystica555_> that'll force a software samplerate conversion if required
[16:43] <mystica555_> (some chips are limited to 48khz)
[16:44] <mystica555_> Mavrik: i thought he was using this to encode, not decode
[16:44] <dragos241> [alsa @ 0x1ed0da0] sample format 0x15002 is not supported
[16:44] <dragos241> plughw:0,0: Input/output error
[16:45] <mystica555_> hrm..
[16:45] <mystica555_> i hate alsa and sound recording...
[16:46] <dragos241> Is there a way I can tell ffserver and ffmpeg to ignore sound altogether?
[16:46] <mystica555_> -acodec null iirc
[16:46] <dragos241> Nope. Not a decoder.
[16:47] <dragos241> That would have been my first thought though
[16:47] <dragos241> The aim here is to stream video, audio is second.
[16:49] <mystica555_> where are you putting the option in your encode commandline? it is position-sensitive
[16:50] <mystica555_> if you do it before -i its a decoder option, if its after the -i file, its the encoder it seems
[16:50] <dragos241> Hm. That may explain it then
[16:51] <relaxed> dragos241: -an
[16:52] <mystica555_> dragos241: so you'd probably want libx264 and libfaac again
[16:53] <dragos241> Yes I did that mystica
[16:53] <dragos241> As it complained to me
[16:53] <mystica555_> k
[16:53] <dragos241> Now I put -af, -acodec, and -vcodec in front of -i this time with their proper parameters
[16:54] <dragos241> As well as -b:a and -b:v
[16:54] <dragos241> Here is what I fed ffmpeg
[16:54] <dragos241> ffmpeg -f x11grab -r 25 -s 384x240 -i :0.0 -f alsa -i hw:0,0 -acodec libfaac -vcodec libx264 -b:a 64K -b:v 128K -ar 8000 -f rtp rtp://192.168.1.102:8090/stream.sdp
[16:55] <mystica555_> do you need 8khz? 64kbit will support 44.1 mono, 22.05 stereo
[16:56] <dragos241> Ah. No I was trying to get this to work before hand. Let me try 22050 on both the config and ffmpeg args
[16:56] <mystica555_> k
[16:57] <dragos241> Alright those were changed. I'm back to previous errors:
[16:57] <dragos241> Could not write header for output file #0 (incorrect codec parameters ?)
[16:57] <mystica555_> can you pastebin the full output again?
[16:57] <dragos241> Yes. Give me a second to do that.
[16:58] <dragos241> http://pastebin.com/LtUKSSyn
[17:00] <dragos241> Fortunately now, the correct codecs are being used
[17:01] <necktie> hi. can anyone help me understand how to crop? i want to crop the black bars from a video. i need to crop 2 pixels on the left, 4 right, 20 top, and 28 bottom. the video is 1920x1080.
[17:04] <mystica555_> dragos241: i can't seem to find the proper documentation to understand how this all works...but im seeing examples of commandlines pointing to a .ffm file instead of a .sdp file; or simply doing an rtp address as rtp:localhost:port
[17:04] <mystica555_> after -f rtp
[17:06] <dragos241> I will see if changing that has an effect. However, I need an sdp file for a program that I use. But if this works, I'll see if it accepts the ffm.
[17:07] <pingec> necktie -cropleft 2 -cropright 4 -croptop 20 -cropbotoom 20 ?
[17:07] <pingec> I have a raw h264 file that is recognized as data. It has no header, I am able to play it only with "ffplay -f h264 aaa.264" is there a way I could create a header for this file so that it gets recognized in every player? Or can I convert it into a legal file without any actual transcoding?
[17:07] <dragos241> pingec: I don't think that works on newer versions of ffmpeg
[17:07] <dragos241> Evidently they use -vf crop now. As for how it works, I'm stumped.
[17:08] <pingec> ah
[17:08] <necktie> pingec: ya. it used to be that simple. now it's hard as hell. at least for me
[17:08] <pingec> get an older version :P
[17:08] <pingec> the options are still listed in the help tho
[17:09] <dragos241> Hm. That's curious. Maybe someone's neglecting to update the man files?
[17:09] <necktie> i've read the documentation. i still have no clue
[17:09] <Mavrik> dragos240, http://ffmpeg.org/libavfilter.html#crop
[17:10] <mystica555_> pingec: try doing ffmpeg -f h264 -i yourfile -vcodec copy something.mkv/mp4 and it'll mux it for you
[17:11] <mystica555_> then being inside said container format will fit your requirement of universal recognisability
[17:11] <pingec> thanks lets try
[17:11] <necktie> Mavrik: ya. that's what i read. still can't figure it out. can you help? i don't want the answer, i just want to figure out how it works
[17:11] <Mavrik> for cropping, it's simple
[17:12] <Mavrik> you apply the crop filter with "-vf crop=<parameters>"
[17:12] <Mavrik> where you pass parameters delimited with semicolons - as in documentation "out_w:out_h:x:y"
[17:12] <Mavrik> meaning, if you want to crop a 400x400 rectangle, starting at 100,200 you pass
[17:12] <Mavrik> -vf crop=400:400:100:200
[17:13] <Mavrik> there are examples right under the description in the documentation file
[17:13] <necktie> Mavrik: so how would i get the parameters for 2 pixels on the left, 4  right, 20 top, and 28 bottom. the video is 1920x1080.
[17:14] <Mavrik> -vf crop=in_w-6:in_h-48:2:20
[17:14] <Mavrik> (your box has to be (2+4) smaller in width, (28+20) smaller in height and it has to start on 2,20
[17:14] <mystica555_> dragos241: oh btw i just read your comment about the sdp file; the sdp file should be made by the server and present in its web dir; it seems to be wrong or unused for how to describe to ffmpeg where to stream
[17:14] <pingec> mystica555_ thanks, great trick this copy parameter
[17:14] <mystica555_> pingec: i just got done remuxing some divx+mp3 into .mp4 containers for my HP Touchpad because it couldnt demux .avi :\
[17:15] <mystica555_> but...now, COSMOS IN BED!
[17:15] <pingec> hehe :)
[17:15] <necktie> Mavrik: i think i get it now. thanks so much! :)
[17:16] <dragos241> 07:18 < dragos241> mystica555_: Yeah the info on it is kinda sparce
[17:16] <dragos241> I don't know if that went through
[17:16] <dragos241> I lost connection for a tad
[17:16] <mystica555_> it didnt
[17:16] <mystica555_> first time im seeing the line ;)
[17:16] <dragos241> Just wanted to ensure it went through
[17:17] <mystica555_> k
[17:17] <dragos241> Now, experimenting with feeds and streams. The stream depends on a local feed.
[17:17] <dragos241> It doesn't seem to be picky what it's called nor what extension it uses
[17:18] <dragos241> The feed that is
[17:18] <mystica555_> im not sure i follow what context of 'feed' and 'stream' you are using
[17:19] <dragos241> Oh, sorry
[17:19] <dragos241> In the config you have a feed and a stream
[17:19] <dragos241> The feed is what ffmpeg passes into to, and the stream is read from a media player once info from ffmpeg is passed.
[17:19] <dragos241> For example a feed would be feed.ffm
[17:20] <dragos241> A stream may be stream.mp4
[17:20] <mystica555_> ok
[17:20] <dragos241> In my case, my feed is stream.sdp and my stream is stream2.sdp
[17:21] <mystica555_> ok
[17:21] <mystica555_> i can only wonder whether or not ffmpeg is doing some auto-guessing based on extension
[17:21] <mystica555_> thus perhaps naming it .ffm would let ffmpeg know what its doing better 8shrug*
[17:22] <Mavrik> ffmpeg chooses demuxer depending on the extension, so yeah ;)
[17:22] <dragos241> Interesting
[17:22] <dragos241> Naming the feed stream.ffm did nothing
[17:23] <mystica555_> hm
[17:24] <dragos241> However. I still have the output to when I did that
[17:24] <dragos241> I'm posting it on pastebin
[17:24] <mystica555_> k
[17:24] <dragos241> http://pastebin.com/UUpHFwiz
[17:25] <dragos241> Whilst I can't see anything different here, maybe there's something minute.
[17:28] <dragos241> Also if you need the stream file. I have that too.
[17:29] <dragos241> The current one uses the sdp.
[17:29] <dragos241> It was posted earlier, but since has been changed.
[17:29] <dragos241> http://pastebin.com/Kh3eUUja
[17:30] <necktie> Mavrik: one last question. the video is 1920x1080. and i plan on encoding it as 720p. do i need to scale or set dar/sar to do this? or do i just do -s 1280x696? 696 would be the new height instead of 720 due to the cropping..
[17:32] <mystica555_> dragos241: what happens if you omit -f rtp from your output ?
[17:33] <dragos241> Ah. I think I've tried that. It interprets the output as a file.
[17:33] <dragos241> And says it cannot find it.
[17:33] <dragos241> I'll try again though
[17:33] <dragos241> [NULL @ 0x23358b0] Unable to find a suitable output format for 'rtp://192.168.1.102:8090/stream.sdp'
[17:33] <dragos241> rtp://192.168.1.102:8090/stream.sdp: Invalid argument
[17:34] <dragos241> Is the result.
[17:34] <mystica555_> try http for the stream feed file
[17:35] <mystica555_> afaik http is how the static files get served; rtp is simply for the live stream data
[17:35] <dragos241> Hm. I just wiped the audio codec argument and replaced it with -an
[17:35] <dragos241> It works
[17:35] <mystica555_> ah interesting
[17:35] Action: mystica555_ again decries alsa recording
[17:36] <dragos241> Not sure if it plays though
[17:36] <Mavrik> necktie, I suggest you do scale, -s is deprecated
[17:36] <Mavrik> necktie, also, you'll have to put scale filter AFTER crop filter (obviously)
[17:37] <Mavrik> so something like
[17:37] <Mavrik> -vf crop=in_w-6:in_h-48:2:20,scale=-1:696
[17:37] <Mavrik> this will crop video and scale it while keeping the aspect ratio
[17:43] <necktie> Mavrik: when i try that, i get this "width not divisible by 2 (1281x696)"
[17:44] <Mavrik> yuck
[17:45] <Mavrik> I would have thought the scale filter can fix that
[17:45] <Mavrik> necktie, you can just manually set scale=1280x696
[17:45] <Mavrik> er
[17:45] <Mavrik> scale=1280:696
[17:53] <necktie> Mavrik: that worked. thanks again
[17:58] <Tjoppen> what's the way to use x11grab nowadays? -f x11grab etc. says "Unknown input format: 'x11grab'
[17:58] <Tjoppen> even though configura has --enable-x11grab
[17:59] <Mavrik> Tjoppen, check ffmpeg -formats
[18:00] <Tjoppen> ffmpeg -formats|grep x11 -> nothing
[18:00] <Tjoppen> let me update to latest master
[18:01] <Mavrik> Tjoppen, also, funny thing I noticed: if you're missing x11-dev libs, ffmpeg compiles without x11 support even if you go --enable :)
[18:02] <Tjoppen> !!
[18:02] <Tjoppen> libx11-dev is already the newest version.
[18:04] <Tjoppen> yep, master doesn't work either
[18:05] <Tjoppen> libxcb-image0-dev perhaps?
[18:18] <Tjoppen> libx11-xcb-dev mayhaps
[18:18] <Tjoppen> nope.avi
[18:23] <Tjoppen> there we go. needed libxext-dev and libxfixes-dev
[18:24] <Mavrik> wow :)
[18:27] <Tjoppen> today's quest: grab desktop, webcam, microphone and output audio while maintaining sync
[18:30] <Mavrik> hah, good luck :)
[18:32] <Tjoppen> might want yuv444p for the capture.. h.264 doesn't support rgb I suppose?
[18:32] <Tjoppen> I wonder.. ffv1 perhaps
[18:33] <Tjoppen> yep. excellent
[18:34] <Mavrik> Tjoppen, x264 supports RGB, but not via ffmpeg afaik
[18:35] <Tjoppen> ic. well, I think ffv1 will suffice
[18:35] <Tjoppen> or did it not do inter compression?
[18:38] <Tjoppen> apparently not. libx264 + losless_ultrafast + yuv444p it is then
[19:28] <tokam> how to convert an mkv video using ffmpeg to mp3
[19:31] <tokam> unknown encoder .mp3
[19:36] <tokam> libmp3lame is missing
[20:35] <Peace-> just compiled
[20:35] <Peace->  colorspace not supported in LJPEG
[00:00] --- Mon Feb  6 2012