[Ffmpeg-devel-irc] ffmpeg.log.20120506

[burek]

[00:03] <sgfgdf> burek, `fmpeg -i a.mp3 -acodec libaacplus -ar 44100 -ac 2 -ab 32k b.aac' http://cl.ly/2C2Y362a2h3E3A2L1R0X
[00:04] <burek> do you have any sample mp3s?
[00:04] <sgfgdf> burek, the mp3 file is 7.9MB while the aac is 819KB
[00:04] <burek> yes I know, 10x smaller
[00:04] <sgfgdf> burek, you mean that file i'm trying to convert?
[00:04] <burek> I'm just curious what kind of audio is it
[00:05] <burek> yes
[00:05] <sgfgdf> burek, do i cut some seconds or keep it as is?
[00:06] <burek> keep it as original
[00:07] <burek> if you can, put it on some file hosting service and give me a link
[00:07] <burek> and I'll try to do the encoding to see the diff
[00:10] <sgfgdf> burek, can i pm you the link or post it here?
[00:12] <burek> yes sure
[00:12] <burek> pm me
[00:13] <gusto> th
[00:13] <gusto> i am just wondering, because i see a TV channel who is saying that it's live
[00:13] <gusto> but what is live? ... digital TV has some latency, so it can not be live
[00:14] <burek> :)
[00:14] <gusto> so either they lie or they do not know what time it is. it's past midnight here, who would be up to be live on the TV now?
[00:14] <sacarasc> What is live? Baby don't stream me, don't stream me, no more.
[00:14] <burek> :D
[00:15] <gusto> who?
[00:16] <gusto> some baby is being streamed? that can be a felony, you know it?
[00:18] <gusto> however
[00:23] <gusto> i had that impression that amy macdonald lost some weight, or is it the improved mpeg2 quantizer?
[00:27] <aphid> live || canned
[00:28] <burek> gusto, maybe -vf yadif :)
[00:29] <gusto> ???
[00:29] <gusto> what is that?
[00:30] <burek> ultimate weight solution :)
[00:32] <gusto> well, i enabled it, but i do not see any difference
[00:47] <gusto> ha?
[00:47] <gusto> some parameters i have to hit in?
[02:39] <PapaSmurf007> nice ffmpeg comment: //this function is dedicated to the braindamaged gcc
[02:39] <cbreak> they could switch to VC if they hate GCC so much...
[02:41] <PapaSmurf007> it'
[02:42] <iive> cbreak: if only vc was C99
[02:42] <PapaSmurf007> yes!
[02:42] <cbreak> well
[02:42] <cbreak> they did have C in 89 too
[02:42] <PapaSmurf007> the comment was for the function "ff_epzs_motion_search"
[02:42] <cbreak> don't be so picky
[03:10] <echelon> hi
[03:10] <echelon> how do i pass rtmp parameters in ffplay?
[03:11] <echelon> like this.. http://piscui.webear.net/ustream.php?url=http%3A%2F%2Fwww.ustream.tv%2Fchannel%2Fnevadaconvention
[03:15] <echelon> nvm, got it
[03:15] <echelon> just had to put quotes around it
[04:30] <echelon> does ffplay have any volume controls?
[09:07] <kcj> Quick way to strip audio from a .mov?
[09:08] <kcj> Tried "ffmpeg -i file.mov -vcodec copy out.mov" with no luck.
[09:08] <kcj> I get a file but it won't play.
[11:38] <burek> try -acodec copy
[11:39] <burek> oh, to strip :) then -an
[12:36] <wolfgangw> hi, i'm trying to use lutrgb for a matrix transform on rgb, like ffplay -vf lutrgb="r=r*3+g*2+b*1:g=[...]:b=[...]" (omitted the g and b calcs here). apparently this does not work ("Undefined constant or missing '(' in 'r*3+g*2+b*1'"). is there a way to reference other components in each component calculation? (ffplay version N-34350-gf51e501, which is the ubuntu 11.10 default, i reckon)
[12:37] <burek> that error msg says that your shell is complaining about "special characters"
[12:37] <burek> which need to be escaped
[12:38] <burek> most probably with \
[12:40] <wolfgangw> burek: oh? so referencing 'g' and 'b' when evaluating 'r' should work?
[12:40] <burek> http://ffmpeg.org/ffmpeg.html#lut_002c-lutrgb_002c-lutyuv
[12:45] <wolfgangw> burek: yes, thanks, that page is where i realized that this might be possible with lutrgb
[12:45] <wolfgangw> here's the line i'm actually calling:
[12:45] <wolfgangw> ffplay -vf lutrgb="r=r*3.2404542+g*-1.5371385+b*-0.4985314:g=r*-0.9692660+g*1.8760108+b*0.0415560:b=r*0.0556434+g*-0.2040259+b*1.0572252" j2c.mxf
[12:46] <burek> did you try without r,g,b as params
[12:46] <burek> r=5,g=4,b=3 for example
[12:47] <burek> btw, are you creating fractals?
[12:47] <wolfgangw> yes, something like lutrgb="r=val*2:g=val*1:b=val*0" works
[12:47] <burek> I think r,g,b are not valid on the right side of =
[12:48] <wolfgangw> not fractals, xyz->rgb transform
[12:48] <wolfgangw> bummer
[12:48] <burek> only w, h, val, minval, maxval, clipval, negval, clip(val) and gammaval(gamma)
[12:48] <burek> you could suggest that feature in trac
[12:49] <burek> it would be very interesting one I believe
[12:49] <wolfgangw> indeed. and yes, the page actually says that expressions can contain the list you mention (and not r, g, b)
[12:50] <burek> yes, I realized that :)
[12:50] <burek> but I think it wouldn't be that much of a work
[12:50] <burek> to add r,g,b too
[12:50] <burek> :)
[12:51] <burek> maybe you can even take a look at the source code
[12:51] <burek> and just apply a quick patch ? :)
[12:52] <burek> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavfilter/vf_lut.c;h=99df9e58b686c8e98227cbccfc859605156fa0ec;hb=HEAD
[12:52] <wolfgangw> great, thanks
[12:52] <burek> line 86,87,88
[12:52] <burek> those are defaults
[12:53] <burek> so you might use OFFSET(comp_expr_str[R])
[12:53] <burek> as your 'r' :)
[12:53] <burek> :beer: :)
[12:53] <wolfgangw> you in munich? i'll get you some beer you won't forget ;)
[12:54] <burek> :D
[12:54] <burek> no, but if come by, I'll let you know :)
[13:00] <wolfgangw> oh, this is 8bit, yes? line 267
[13:00] <burek> I guess every component (r,g,b) is 8 bit
[13:01] <burek> including alpha
[13:02] <wolfgangw> wasn't there some work on >8bit components, in relation to prores work iirc? sorry, don't know the codebase well
[13:03] <burek> you could ask in #ffmpeg-devel :)
[13:04] <wolfgangw> you come and join me, like hold my hand and all? :)
[13:06] <burek> :)
[13:29] <wolfgangw> burek, i wonder what's good generic name for the trac ticket. i'm asking for 3x3 matrix transform via lut/lutrgb which is of course not exclusive to xyz->rgb, it involves referencing "the other" components when evaluating a component's expression ... idea?
[13:29] <wolfgangw> good name important :)
[13:30] <burek> I would name it conversion of xyz to rgb
[13:30] <burek> :)
[13:30] <burek> because there might be another (easier) way of doing the same :)
[13:31] <wolfgangw> not generic, and the feature would be generic. but yes, possibly easier way existing, you have a point there
[13:32] <burek> if they decide to add that feature in that filter, it would be great also :)
[13:32] <burek> but just for the sake of resolution of your problem
[13:32] <burek> I would do like that :)
[13:34] <wolfgangw> "X'Y'Z' -> RGB via lutrgb", how does that sound? the 's denote a gamma encoding, which is involved as well
[13:34] <burek> ok :)
[13:34] <wolfgangw> done, thanks
[13:35] <burek> :beer: :)
[13:35] <burek> http://en.wikipedia.org/wiki/Free_Beer
[13:36] <burek> Free Beer, an open source beer formerly known as Vores Øl, Danish for Our Beer
[13:36] <burek> :)
[13:38] <wolfgangw> not sure i want to see the changelog for that :)
[13:51] <_SKiTZO> when i want to stream over rtp directly from avconv/fmpeg, what ip adress should i put in the rtp:// path?
[13:51] <_SKiTZO> i tried all sorts but i just cant seem to find the rtp stream from another computer
[13:52] <_SKiTZO> my best guess is the "externally visible" ip address of the computer running avconv
[13:52] <_SKiTZO> (10.0.0.105 in my case)
[13:57] <_SKiTZO> while running avconv i cant se that it is listening on any ports whn i do netstat -l -n
[14:18] <wolfgangw> a build problem, maybe someone can help: want to build a version that can play jpeg2000 essence in an mxf container (digital cinema assets are like that). my default ubuntu 11.10 version of ffmpeg can do it (version N-34350-gf51e501, configuration: --enable-libopenjpeg --enable-parser=mxf). the version i'm building here with .configure --prefix=[some prefix] --enable-libopenjpeg --enable-parser=mxf can not. ldd ffplay-my-version sho
[14:18] <wolfgangw> ws that it picks up the right openjpeg lib. Messages for both versions are at http://pastie.org/pastes/3868257/text
[14:20] <wolfgangw> i'm trying to reproduce by following https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide now. but if someone has a hint or two i'd appreciate it. thanks
[14:25] <YellowOnion> how does one make a webm video that isn't fixed bitrate?
[14:25] <YellowOnion> or, why dose -crf do nothing
[14:25] <YellowOnion> does*
[14:26] <sacarasc> Can you paste your complete command line and output to a pastebin??
[14:26] <sacarasc> Ignore one of them question marks.
[14:27] <YellowOnion> http://pastebin.com/ggnjNGF8
[14:28] <sacarasc> And the output?
[14:29] <YellowOnion> http://pastebin.com/6MPkkGrj
[14:30] <APoulos> burek: hey mate, are you awake?
[14:30] <APoulos> burek: send me a pm when you can, I could use some more tutoring on ffmpeg
[14:30] <YellowOnion> sacarasc: missed a few starting output stuff, just looks like the compiled with info
[14:30] <YellowOnion> I can include if needed
[14:30] <APoulos> burek: just memoserv me when you get these
[14:31] <APoulos> burek: my sleep schedule's off, but as I said, memo me when you can
[14:31] <APoulos> for now I'm going dark
[14:33] <burek> :)
[14:33] <burek> take a beer :)
[14:34] <burek> YellowOnion
[14:34] <burek> can you please use pastebin.com, to show your command line and its output?
[14:34] <sacarasc> burek: He did.
[14:34] <burek> and please dont just cut some "interesting parts"
[14:34] <sacarasc> But there is that.
[14:35] <burek> we don't see ffmpeg's version
[14:35] <YellowOnion> got it today from the windows build site
[14:35] <YellowOnion> ffmpeg version N-34549-g13b7781, Copyright (c) 2000-2011 the FFmpeg developers
[14:35] <YellowOnion>   built on Nov  6 2011 22:02:08 with gcc 4.6.1
[14:36] <YellowOnion> burek: as I said I missed it, cmd.exe has a terrible cut and paste method
[14:37] <burek> ffmpeg.exe ... > output.txt
[14:37] <burek> notepad output.txt
[14:40] <burek> btw, YellowOnion, why option -ss twice?
[14:41] <YellowOnion> first one seeks the file, the second decodes then seeks, 2nd option is more accurate, but decoding 1h of video wont be fun
[14:42] <burek> so, you landed somewhere at -ss 1:31:37
[14:42] <burek> and fine tuned it with another -ss?
[14:43] <YellowOnion> yeah
[14:43] <burek> cool, I didn't know it can be done that way
[14:44] <YellowOnion> well output.txt is completely empty
[14:44] <burek> hm
[14:44] <burek> maybe 2> output.txt
[14:44] <burek> we need to redirect error :)
[14:45] <YellowOnion> that's if it works with cmd.exe :/
[14:45] <YellowOnion> ahh here we go
[14:45] <burek> and how do you run it
[14:45] <burek> ok :)
[14:45] <burek> well, is -crf option libx264 related
[14:46] <burek> or a general option?
[14:46] <burek> I thought it was libx264's option only
[14:46] <YellowOnion> http://ffmpeg.org/ffmpeg.html#libvpx
[14:46] <YellowOnion> it's a libvpx option as well
[14:47] <YellowOnion> wow, cmd.exe outputs unix text files
[14:48] <burek> no, it's not
[14:48] <burek> it's only redirecting ffmpeg's error output
[14:48] <burek> to a file
[14:48] <YellowOnion> ahh right
[14:48] <YellowOnion> gah, cmd's history is screwwy
[14:49] <YellowOnion> somehow ended up with the wrong command
[14:49] <burek> fun never ends :)
[14:49] <YellowOnion> it remembers your location after hitting enter >.>
[14:49] <wolfgangw> yep, following the build guide on https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide and then trying to playback a jpeg2000/MXF container gives me the same errors as before: http://pastie.org/pastes/3868351/text
[14:50] <YellowOnion> burek: http://pastebin.com/KjysdNpr
[14:52] <YellowOnion> burek: http://pastebin.com/NVUbnNqF
[14:53] <YellowOnion> oh wow, >.> that dosn't even have the -crf on it
[14:54] <YellowOnion> there must be some voodoo going on with cmd.exes scroll back buffer
[14:56] <YellowOnion> no wonder windows users hate the command line
[14:57] <YellowOnion> burek http://pastebin.com/pTddz6iL
[14:57] <YellowOnion> I think this is correct
[15:03] <cbreak> get yourself a bash and Console2
[15:06] <YellowOnion> know of any good windows bash versions, I tried cygwin once but it was causing strange errors
[15:07] <YellowOnion> but yeah I just want to know why -crf doesn't work, otherwise, how to do vbr with libvpx
[15:12] <cbreak> msys maybe
[15:12] <cbreak> -crf is an x264 option
[15:12] <NuxRo> hello, anyone knows if it's possible to build ffmpeg 0.8 statically against libvpx?
[15:12] <cbreak> not sure if other encoders support it
[15:13] <YellowOnion> cbreak: well it doesn't say unknown option and it's in the docs
[15:13] <sacarasc> libvpx should, but maybe not a 6 month old version of ffmpeg.
[15:13] <YellowOnion> I downloaded a fresh build today :/
[15:14] <sacarasc> [13:35:42] <YellowOnion> ffmpeg version N-34549-g13b7781, Copyright (c) 2000-2011 the FFmpeg developers
[15:14] <sacarasc> [13:35:42] <YellowOnion>   built on Nov  6 2011 22:02:08 with gcc 4.6.1
[15:14] <JEEB> <cbreak> -crf is an x264 option <- crf is specific to libx264
[15:15] <sacarasc> No it's not.
[15:15] <sacarasc> -crf               <float> E.V.. enables constant quality mode, and selects the quality (x264/VP8)
[15:15] <sacarasc> That's from ffmpeg -help
[15:16] <YellowOnion> wait installed imagemagick today, I vaguely remember it including ffmpeg
[15:16] <YellowOnion> it also added it's self to %PATH%
[15:18] <YellowOnion> somehow I deleted the '..\'ffmpeg from the switch >.>
[15:19] <YellowOnion> ffmpeg version N-40301-gc1fe2db Copyright (c) 2000-2012 the FFmpeg developers
[15:19] <YellowOnion>   built on May  3 2012 11:44:06 with gcc 4.6.3
[15:20] <YellowOnion> still doesn't work though :-\
[15:25] <NuxRo> hello, anyone knows if it's possible to build ffmpeg 0.8 statically against libvpx and how?
[15:31] <burek> built on Nov  6 2011
[15:31] <burek> YellowOnion, that's old
[15:33] <burek> NuxRo, it is: http://ffmpeg.gusari.org/static/
[15:34] <burek> oh you mean an old version, does it support libvpx.. well, that I don't know, but you can check the git to see when libvpx was introduced
[15:35] <YellowOnion> burek: did you even read my comment 4 above your one?
[15:35] <burek> indeed libvpx supports -crf http://ffmpeg.org/ffmpeg.html#libvpx
[15:35] <burek> YellowOnion, that's why we always ask for complete output
[15:36] <burek> to make sure that the command you run is the one you need
[15:36] <burek> :)
[15:36] <burek> if you tamper with the output, you're making us harder to help you
[15:37] <YellowOnion> never tappered with it, I just missed it, when someone asked for it I pasted it here, yet you still insistented I paste so I did, then you went afk, other people helped me, and the new version from may 3rd still has problems
[15:37] <burek> also, you can go to windows folder, and find cmd.exe, right click, properties and set the scroll buffer size, to be able to copy/paste everything you need
[15:38] <burek> YellowOnion,
[15:38] <burek> can you please use pastebin.com, to show your command line and its output?
[15:40] <YellowOnion> that's not the problems it's it's copy-paste method
[15:41] <burek> right click in the black cmd.exe window, choose Mark, Select All, then again right click, Mark - Copy
[15:41] <burek> simple as that
[15:41] <burek> you can also use properties on cmd.exe and select quick copy
[15:42] <burek> so you can just select what you want with your mouse and use right click to place in in the clipboard
[15:42] <YellowOnion> http://pastebin.com/bhTmJdMK
[15:42] <NuxRo> burek: what I want to do is to build ffmpeg 0.8 on Centos 6, but it requires libvpx1 whereas EL6 provides libvpx0.9.. hence my desire to build statically
[15:43] <YellowOnion> burek: I know how to copy paste, it's the box select and stupid newline problem it has
[15:43] <NuxRo> burek: thanks to that page, maybe i'll manage to do it like that guy does
[15:44] <burek> YellowOnion, there are no errors in your paste, what exactly doesn't work for you?
[15:45] <YellowOnion> it's still targetting 200kbps, I want vbr not cbr, and '0' is the highest quality it'll do otherwise I have to specify a fixed bitrate
[15:45] <burek> NuxRo, ok :) btw, it's my page :) and if you need any help, let me know :) but I think a simple git checkout <version> and usual ./configure ... --enable-libvpx && make && make install will do the job
[15:45] <burek> YellowOnion, add -b 800k
[15:46] <burek> 0 isn't the highest quality
[15:46] <burek> you're mixing x264 with vpx
[15:46] <YellowOnion> the docs are inaccurate
[15:46] <burek> no they're not
[15:46] <burek> docs are fine
[15:46] <YellowOnion> well i did 63 and it came out at 100kbps
[15:47] <burek> it says that -crf (for vpx) is mapping to -crf: VPX_CQ, VP8E_SET_CQ_LEVEL
[15:47] <burek> so, when you type -crf X
[15:47] <burek> that translates to specific VPX_CQ=... && VP8E_SET_CQ_LEVEL=...
[15:47] <burek> and you mixed that with x264's -crf=0 (lossless)
[15:48] <YellowOnion> yet, my file is still smaller and of worse quality when I follow the docs
[15:49] <burek> I give up, I got to go, I hope you'll find your answer :)
[15:49] <burek> :beer: :)
[16:05] <YellowOnion> Is anyone here actually able to help me get libvpx to do VBR?
[16:23] <NuxRo> burek: do you put the libvpx source in /root/ffmpeg-static/64bit when compiling?
[16:26] <burek> yes
[16:26] <burek> why
[16:26] <burek> actually it's in ffmpeg-static/64bit/obj/libvpx
[20:10] <PapaSmurf007> I'm trying to add some watermarking capabilities in p-frame motion vectors to the mpeg2video codec in FFMPEG.  After the picture type is determined to be a P-frame, I can modify the MV, but I'm having trouble determining how/what to update so that the residual is recomputed given the changed MV.   Is this something I should ask about in #ffmpeg-dev?
[21:08] <Mavrik> um, why does ffmpeg need x264 compiled to have yuv420p -> rgb24 accelerated conversion?
[21:33] <brontosaurusrex> can i copy/paste metadata using ffmpeg, say from flac to mp4?
[21:34] <burek> -map_metadata
[21:35] <AntumDeluge> I'm running Debian 6 with FFmpeg 0.10.2, and trying to figure out how to record from different microphones. Is there a way to tell FFmpeg to record from either my built-in mic or the mic jack?
[21:35] <brontosaurusrex> burek: thanks
[21:36] <cbsrobot_> Mavrik: I don not think i does ....
[21:36] <Mavrik> hmm, well, I'm getting "No accelerated conversion..." warnings and some online sources say I need x264 O.o
[21:37] <Mavrik> AntumDeluge, that's wholly dependant on the system you're grabbing audio from
[21:37] <Mavrik> AntumDeluge, ffmpeg just takes audio from a source, it's up to you to point to the right one
[21:37] <brontosaurusrex> burek: uhmm, how would that work if both files allready exist?
[21:38] <AntumDeluge> Yeah, I noticed that from my Volume Control Options I could switch between Digital Mic (built-in) and Front Mic (mic jack). So FFmpeg can override this setting?
[21:38] <Mavrik> AntumDeluge, there's nothing to "override"
[21:39] <Mavrik> you told ffmpeg which microphone to grab the audio from
[21:39] <Mavrik> and without actually showing us the command line as the guidelines tell you we can't tell you more.
[21:39] <Mavrik> since we don't know which audio subsystem you're using.
[21:39] <cbsrobot_> Mavrik: do you have yasm installed ?
[21:39] <AntumDeluge> I'm using alsa "ffmpeg -f alsa -i hw:0,0"
[21:40] <Mavrik> cbsrobot_, yep, the other conversions work
[21:40] <Mavrik> AntumDeluge, "hw:0,0" is the name of ALSA capture deive
[21:40] <Mavrik> *device
[21:41] <Mavrik> where you tell ffmpeg from which device to grab audio from.
[21:41] <cbsrobot_> Mavrik: show an example and file a trac issue ....
[21:42] <Mavrik> cbsrobot_, example of what exactly? (I'm using libav through the API in my own binary - so I'll know what would help)
[21:42] <AntumDeluge> Okay, my alsa config shows two subcapture devices and I thought that maybe if I used something like "hw:0,1" that might work, but It doesn't.
[21:43] <AntumDeluge> It's not a big deal. I guess I'll just have to set up the desktop for recording from my preferred device each time I capture.
[21:43] <cbsrobot_> ah ok - but when converting it with ffmpeg you get the same error ?
[21:43] <aphid> any good examples of adding arbitrary files into mkv container with ffmpeg? (images, json metadata, &etc)
[21:44] <Mavrik> cbsrobot_, yeah, if I'm using that same build (I have ffmpeg 0.10.2 with all encoders, devices etc. stripped out)
[21:44] <Mavrik> AntumDeluge, "aplay -l" should give you list of devices
[21:45] <Mavrik> AntumDeluge, and "setting your desktop" won't help, because you're explicitly telling ALSA to grab from that microphone.
[21:45] <Mavrik> it's like turning a stove knob to switch TV channel -_-
[21:47] <brontosaurusrex> how would i : "FFmpeg is able to dump metadata from media files into a simple UTF8encoded INI-like text file and then load it back using the metadata muxer/demuxer."?
[21:47] <AntumDeluge> Yeah, that's what I mean. When I want to record from the built-in I'll just set my "desktop" (gnome/alsa) to capture from "Ditital Mic". And when I want the mic jack, "Front Mic". I was just hoping there was a way to point ffmpeg to each device without touching my Volume Control in Gnome.
[21:50] <Mavrik> AntumDeluge, yeah, what exactly is unclear?
[21:50] <Mavrik> run "aplay -l", get list of audio devices with addreses
[21:50] <Mavrik> put proper "hw<num>,<num>" into -i parameter
[21:51] <AntumDeluge> "Subdevice #0: subdevice #0" and "Subdevice #1: subdevice #1"... Okay, I think I understand what that means. Let me give something a try...
[21:52] <burek> brontosaurusrex, you can use filter metadata to create ini file
[21:52] <burek> but only if the metadata is in the container
[21:52] <burek> not if it's in one of the streams
[21:53] <burek> but, to be honest, it's best to use -map_metadata to copy all meta into a dummy output file (set -t 1)
[21:53] <brontosaurusrex> artists = burek, composer = brontosaurusrex is probably container no?
[21:53] <burek> and then when you need that info, to use -vn -an -d copy
[21:53] <burek> brontosaurusrex,
[21:53] <burek> can you please use pastebin.com, to show your command line and its output?
[21:54] <AntumDeluge> Okay, I've tried "-i hw:0,0", "-i hw:0,1", "-i hw:1,0", and "-i hw:1,1". But the only one that works is "-i hw:0,0".
[21:54] <burek> AntumDeluge, use -arecord -l
[21:54] <burek> (small L)
[21:54] <brontosaurusrex> burek: i don't have anything yet to show
[21:54] <burek> and further use -f alsa -i hw:<blah>
[21:54] <burek> where blah is one of the items from the list you got with arecord
[21:55] <burek> arecord (not -arecord).. it's a cmd line tool
[21:55] <AntumDeluge> $ arecord -l
[21:55] <AntumDeluge> **** List of CAPTURE Hardware Devices ****
[21:55] <AntumDeluge> card 0: Intel [HDA Intel], device 0: STAC92xx Analog [STAC92xx Analog]
[21:55] <AntumDeluge>   Subdevices: 2/2
[21:55] <AntumDeluge>   Subdevice #0: subdevice #0
[21:55] <AntumDeluge>   Subdevice #1: subdevice #1
[21:55] <brontosaurusrex> burek: but the idea is: original = flac, use ffmpeg to make aif and store tmp metadata, use afconvert to get arc from that aif, use ffmpeg to copy/paste metadata from flac to than new aac
[21:55] <brontosaurusrex> *arc=aac
[21:55] <burek> AntumDeluge, please use pastebin
[21:56] <burek> brontosaurusrex, just type ffmpeg -i original
[21:56] <burek> and paste it to pastebin
[21:56] <AntumDeluge> pastebin? I am using pidgin, I don't see a "pastebin" option.
[21:57] <burek> pastebin.com
[21:57] <burek> did you ever use it?
[21:58] <AntumDeluge> Oh! Then I just paste the URL?
[21:58] <burek> yes :)
[21:59] <AntumDeluge> Thanks.
[22:11] <brontosaurusrex> ffmpeg -i input.flac  -f ffmetadata meta.txt < seems to do something
[22:14] <brontosaurusrex> burek: how to use -vn -an -d ?
[22:14] <burek> ffmetadata works for metadata that is in the container
[22:14] <burek> but not for metadata that is in the encoded streams
[22:22] <brontosaurusrex> that would be good enough
[22:22] <burek> ok, then use it :)
[22:22] <brontosaurusrex> i can't read it back
[22:23] <brontosaurusrex> neither take it from the original stream, paste: http://pastebin.com/hKx39d85
[22:24] <brontosaurusrex> Invalid metadata type 1
[22:25] <burek> why not
[22:25] <brontosaurusrex> dunno
[22:25] <burek> ffmpeg -f ffmetadata -i bla.ini -i input.avi -c copy output.avi
[22:25] <burek> no
[22:25] <burek> actually yes :)
[22:25] <burek> that's correct line
[22:27] <brontosaurusrex> yes!!!!
[22:29] <brontosaurusrex> -c copy < is copy everything?
[22:29] <burek> yes
[22:30] <brontosaurusrex> so this would work in a lot of cases? http://pastebin.com/0bUE0RUC
[22:32] <burek> if the output format supports metadata
[22:32] <burek> and if the input media has got metadata in the container, not in the audio/video stream
[22:32] <burek> then yes
[22:32] <burek> it would work
[23:39] <scalability-junk> hey I'm trying to convert a ovg video size 3.7 gb to an avi one, but the output video seems to have some quality issues. there are blurring pixels. any idea how to prevent that?
[23:42] <NuxRo> scalability-junk: try to use -sameq switch
[23:44] <Mavrik> no, no, no, no
[23:44] <Mavrik> you practically NEVER want to use "-sameq"
[23:44] <Mavrik> ESPECIALLY when transcoding.
[23:44] <NuxRo> he doesnt?
[23:45] <NuxRo> ok, that might explain why my vids quality is so shit usually hehe
[23:45] <Mavrik> "sameq" doesn't mean what you think it means :P
[23:45] <Mavrik> it's a really poor choice of name for a parameter -_-
[23:45] <NuxRo> Mavrik: if i wanted to transcode and dont lose much quality, what should i do?
[23:45] <scalability-junk> didn't work anyway
[23:45] <Mavrik> scalability-junk, well... maybe you should post your command-line, output as the guideline tells you
[23:45] <Mavrik> and then MAYBE we can help
[23:46] <Mavrik> if you can tell us what you want to achieve... there is a slightly bigger chance we can help.-
[23:46] <Mavrik> (sorry, forgot crystal ball at home)
[23:46] <scalability-junk> Mavrik, sorry I'm using winff with the standard conf, reading the man for ffmpeg right now
[23:46] <Mavrik> NuxRo, well... sadly the only solution for that is to find the right quality setting for your encoder
[23:47] <scalability-junk> I want to acheive not having pixels, which have nothing todo in the video, in my video after transcoding to avi
[23:47] <Mavrik> NuxRo, sameq just sets the same quantizer, which means something very different from encoder to encoder... and you practically never want to tamper with the quantizer settings
[23:47] <Mavrik> scalability-junk, videos are made from pixels.
[23:47] <Mavrik> avi is a container.
[23:48] <Mavrik> you're basically telling us "I don't want water in my car. It's red. What's wrong?"
[23:48] <scalability-junk> I'm looking for an example ;)
[23:48] <Mavrik> NuxRo, using "-crf" for quality control if you're encoding with x264 is usually the best choice
[23:48] <lemonzest> grrr, i want theora to be multi-threaded :(
[23:49] <lemonzest> getting a whole 25fps, mpeg4 i get like 200
[23:50] <aphid> lemonzest - vp8 any better? it's also royalty-free
[23:50] <Mavrik> yuck, theora :)
[23:50] <NuxRo> Mavrik: what I usually end up doing is transcoding to ogv/ogg for the web browsers, any advice for this?
[23:50] <lemonzest> aphid: not actaullt tried iut yet, its going to be next
[23:50] <Mavrik> NuxRo, well, my testing has shown that Theora is really bad for streaming content, so we switched to webm
[23:50] <lemonzest> NuxRo: you the same guy who has the el6 repo?
[23:50] <aphid> serves the same purpose, unless people are using ancient firefox
[23:50] <Mavrik> that's mostly it... Theora encoder doesn't really give you much choice
[23:51] <lemonzest> theroa is ok with 2 pass, or quality, but 1 pass bitrate? blah
[23:52] <NuxRo> lemonzest: i do have a repo for el6, yes
[23:52] <scalability-junk> Mavrik, see the flickering pixels in the right top corner? http://dl.dropbox.com/u/5581/out.ogv
[23:52] <NuxRo> Mavrik: EL6 firefox doesnt support webm afaik
[23:52] <Mavrik> lemonzest, I've seen theora miss target for more than 60% of bitrate even in 2-pass mode
[23:52] <lemonzest> my test encode is with big bang theery, itunes rip, h264 at 4.6Mb/s, and trans coding down to 1.5Mb/s and its still looking nice with 2 pass
[23:53] <Mavrik> and those spikes weren't really acceptable
[23:53] <lemonzest> NuxRo: i use the firefox from remi's repo, it uses the system libvpx for webm
[23:53] <Mavrik> at least not for web streaming applications
[23:53] <Mavrik> NuxRo, what's EL6 firefox?
[23:53] <lemonzest> Mavrik: enterprise linux 6, aka RHEL
[23:54] <NuxRo> Mavrik: EL = Enteprise Linux, usually referrs to RHEL and clones (centos, scientific linux etc, EL6 = RHEL6 and clones)
[23:54] <lemonzest> or in my case, Scientific Linux
[23:54] <Mavrik> ah, I see
[23:54] <Mavrik> funny that, is it still shipping with 3.5? or is it stripped?
[23:54] <lemonzest> Mavrik: its shipping 10 ESR, but they stripped webm out, god knows why
[23:54] <aphid> isn't 3.6 not even supported anymore?
[23:54] <NuxRo> Mavrik: 10 now
[23:54] <Mavrik> lemonzest, we decided to rather support users on mobile connections instead of market with old browsers
[23:55] <NuxRo> lemonzest: maybe patents issue, who knows
[23:55] <lemonzest> i'm using 12 from remi's repo
[23:55] <NuxRo> hence my tendency to use ogg
[23:56] <lemonzest> NuxRo: it just brings me to tears, that my cpu can turbo upto 3 threads at 4Ghz (overclocking FTW) and i still only get 25fps for libtheora encoding through ffmpeg
[23:56] <lemonzest> might see if ffmpeg2theora is faster
[23:56] <aphid> webm 's been in firefox nightlies for 2 years
[23:57] <NuxRo> well.. can't assume most EL users use remi's repo, ogg is fine with me, not a professional video guy anyway
[23:57] <lemonzest> its theora thats slowing the encode down tho, i did a null encode and it goes at 700fps decoding h264
[23:58] <aphid> unlikely to get better, as far as i can tell most of the #theora guys are working on audio stuff, daala or webm (or some combo of the above)
[23:59] <Mavrik> mhm, theora development seems defunct
[00:00] --- Mon May  7 2012