[Ffmpeg-devel-irc] ffmpeg.log.20120924

burek burek021 at gmail.com
Tue Sep 25 02:05:01 CEST 2012


[07:59] <reactormonk> I want to store the audio stream of my video file separatly -c:a copy -c:v null ?
[08:01] <reactormonk> ... apparently nope
[08:02] <grepper> try -vn
[08:04] <ubitux> or just -map it
[09:29] <ameeth> hi i am using gstreamer for streaming  an web cam video on  to a rtmp their is an audio error can any one tell me hoe to get an audio from webcam
[10:57] <lserrno> hi, anyone know why the libstagefright codec needs muxed inputs ? if it's a codec, it should accept raw h264 inputs no ?
[12:02] <arpu> hello
[12:02] <arpu> is it possible always selct the same audio stream Stream #0:1[0x1c0] ?
[12:02] <arpu> so always [0x1c0]
[12:03] <arpu> some times the audio order is diferent
[12:03] <arpu> different
[13:02] <junctions> Hi guys, I'm trying to compile ffmpeg without root and I want to use x264. So I compiled x264 into a folder (call it $FFMPEG-LIB-DIR) using
[13:02] <junctions> % ./configure --prefix=/$FFMPEG-LIB-DIR --enable-static
[13:02] <junctions> Now I read that to compile ffmpeg with this I should run:
[13:02] <junctions> % PKG_CONFIG_PATH="$FFMPEG-LIB-DIR/usr/lib/pkgconfig" \
[13:03] <junctions> ./configure --prefix=/$HOME/bin/ffmpeg --enable-libx264 --enable-gpl
[13:03] <junctions> But when I do that I still get ERROR: libx264 not found
[13:03] <junctions> How do I tell ffmpeg where to look for a library?
[13:04] <JEEB> --extra-cflags="-I${FFMPEG-LIB-DIR}/include" --extra-ldflags="-L${FFMPEG-LIB-DIR}"
[13:04] <JEEB> should do it
[13:04] <JEEB> as I think the x264 check is not done with pkg-config :)
[13:04] <JEEB> also, it's PREFIX/lib/pkgconfig
[13:04] <JEEB> not prefix/usr/lib/pkgconfig
[13:05] <JEEB> (no usr)
[13:05] <JEEB> and I of course forgot the /lib out of the ldflag
[13:05] <JEEB> but yeah
[13:05] <JEEB> you get it, right?
[13:05] <JEEB> add the prefix/include to include path and prefix/lib into library search path
[13:05] <junctions> woops, right
[13:06] <junctions> awesome, hehe, thanks again JEEB
[13:06] <JEEB> no problemo
[13:06] <JEEB> also here you are being rescued by the fact I recommended static libraries, shared and non-standard prefixes are a pain in the butt
[13:06] <JEEB> because the system by default only looks for them in specific locations
[13:07] <JEEB> and you would have to set LD_LOAD_PATH or whatever it was when launching the app
[13:09] <rainmaker1> Hi, I am trying to capture image from live video source every 1 minute, now the problem is that ffmpeg does not allow me to use -f image2 -r 1/60 somepic.png rather it fires the error
[13:09] <rainmaker1> "Could not get frame filename number 2". I know can use  -frames:v 1 but this is a live stream and I want just to capture one image every minute and that
[13:09] <rainmaker1> image should overwrite the image from a miniute ago. Is there any help?
[13:09] <rainmaker1> I mean a live stream that ffmpeg is transcoding
[13:19] <ubitux> rainmaker1: try with vf select
[13:20] <ubitux> -vf select 'not(mod(n\,60/TB))' or something like this
[13:21] <ubitux> ah and overrite the next image mmh
[13:21] <ubitux> you can have a new image incremented
[13:24] <ubitux> (-vf 'select=not(mod(n\,60/TB))' sorry)
[13:42] <rainmaker1> ubitux: huh, I will take a look at doc as I don't understand select=not(mod(n\,60/TB)) part of the command :)
[13:50] <rainmaker1> hmmm does not work either... :( nevermind I guess I can start ffmpeg from cron every minute to capture the image
[13:56] <microchip_> or use fswebcam
[13:57] <rainmaker1> fswebcam?
[13:57] <microchip_> oh i see, you want from live source
[13:57] <microchip_> are you capturing from /dev/video0 ?
[13:58] <rainmaker1> nope, it's a live stream in mepg2ts with h264 inside
[13:58] <rainmaker1> it's a multicast
[13:58] <microchip_> ah i see, nvm my fswebcam suggestion
[13:58] <rainmaker1> :)
[13:59] <rainmaker1> tnx, btw one thing bothers me also, soe anyone know I can be sure to capture the full image?
[14:00] <rainmaker1> I mean, I have GOP frame every 10 seconds for example
[14:00] <rainmaker1> how can I be sure not to get only I frames in output image?
[14:02] <ubitux> rainmaker1: ffmpeg -f v4l2 -i /dev/video0 -vf 'select=not(mod(n\,60/TB)' -f image2 'out%04d.png' doesn't work?
[14:02] <ubitux> (i can't test)
[14:02] <ubitux> for i-frame only
[14:02] <ubitux> i'd say select='eq(pict_type\,I)'
[14:04] <rainmaker1> ubitux: actually I am using ffmpeg -i udp://230.11.0.23:5000 -vf 'select=not(mod(n\,60/TB)' -f image2 'out%04d.png'  rtl_2.png
[14:04] <ubitux> remove rtl_2.png
[14:05] <ubitux> it will create out*.png images
[14:05] <ubitux> every minutes
[14:05] <rainmaker1> ubitux: and it now works, here is pastebin with error http://pastebin.com/v57HMN9s
[14:05] <ubitux> at least it should.
[14:05] <ubitux> lemme try.
[14:05] <rainmaker1> ubitux: I would like just to have one image rtl_2.png which is overwritten every minute with the new one
[14:05] <ubitux> yeah i know
[14:06] <ubitux> but i don't think that's supported
[14:06] <ubitux> at least through the cmd line
[14:06] <ubitux> you should be able to write a program to do that easily though
[14:06] <rainmaker1> ubitux: a bash/perl script or I should be using some ffmpeg api?
[14:07] <ubitux> c program with ffmpeg api
[14:07] <ubitux> add a ')' to the vf select
[14:07] <ubitux> at the end of the expression
[14:08] <rainmaker1> ubitux: you are right now I have and old error  Could not get frame filename number 2 from pattern 'rtl_2.png'
[14:09] <rainmaker1> ubitux: which means that I should find someone who knows C I guess :)
[14:09] <ubitux> remove rtl_2.png
[14:09] <ubitux> and try at first generating n files
[14:09] <ubitux> with my cmd line
[14:11] <ubitux> ffmpeg -i ... -vf 'select=not(mod(n\,1.0/TB))' -vsync vfr -f image2 out%04d.png
[14:11] <ubitux> this works for me
[14:11] <ubitux> for 1 image per second
[14:11] <ubitux> (replace 1 with 60 for one minute)
[14:12] <ubitux> ’ -vf 'select=not(mod(n\,60/TB))' -vsync vfr -f image2 out%04d.png
[14:12] <rainmaker1> heh strange
[14:12] <rainmaker1> I get only one file
[14:12] <rainmaker1> out0001.png
[14:12] <ubitux> did you wait 1 minute?
[14:13] <ubitux> also, what is the version of your ffmpeg?
[14:13] <ubitux> anyway, if you need some examples for the API: http://git.videolan.org/?p=ffmpeg.git;a=tree;f=doc/examples;hb=HEAD
[14:13] <ubitux> see decoding_encoding.c or just demuxing.c
[14:14] <rainmaker1> ffmpeg version N-43112-g39b0d40 Copyright (c) 2000-2012 the FFmpeg developers built on Aug  1 2012 15:34:18 with gcc 4.4.6 (GCC) 20120305 (Red Hat 4.4.6-4)
[14:14] <ubitux> oh and maybe filtering_video.c actually, if you want to directly use the select filter
[14:14] <ubitux> kind of old but should be ok
[14:15] <ubitux> (~2 months)
[14:15] <rainmaker1> nope, it only creates one file and seems like it stops working
[14:15] <rainmaker1> I will try to upgrade to the laster version from git
[14:16] <ubitux> can you pastebin your cmd line and full output?
[14:16] <rainmaker1> sure
[14:17] <rainmaker1> http://pastebin.com/BxuwSHqZ
[14:18] <ubitux> does it work if you lower the value?
[14:18] <ubitux> to, let's say, 3 or 4 seconds
[14:18] <rainmaker1> let me try
[14:19] <rainmaker1> nope
[14:19] <rainmaker1> always the same
[14:19] <rainmaker1> using 1,3 or 4 seconds
[14:19] <ubitux> strange
[14:20] <rainmaker1> are you using the latest version from git?
[14:20] <ubitux> try with a more reliable source (a video file) instead of a udp maybe
[14:20] <ubitux> yes
[14:20] <rainmaker1> ok, I will try, but first I think I should upgrade to the latest version :)
[14:21] <Velmont> burek: Thank you so much for the static binaries :-)
[14:21] <rainmaker1> thank you ubitux :)
[14:21] <ubitux> np :P
[14:30] <Velmont> So, I just want to save /some/ space, so not pcm, - what to use? I thought about Opus, but it's not compiled in, so will try something else.
[14:31] <Velmont> mp3 et al made the fps go from 1000 to 119, so it's a bit slow. (I have huge amount of files to merge (h264 + pcm audio))
[14:31] <ubitux> vorbis is nice as well
[14:31] <Velmont> Yes, vorbis was a bit quicker, but still on the slowpoke side :-) -- Unless it has a fastness option
[14:32] <ubitux> you can use a native lossless codec otherwise
[14:32] <ubitux> like flac
[14:32] <ubitux> you won't be able to save as much space as with a lossy one though
[14:33] <Velmont> Hm, flac does something strange, it starts very fast 1000fps, but then slower slower, down to 200-300 fps.
[14:34] <Velmont> Oh, pure copy does in fact the same, only not as quick. So maybe because it's getting more io bound.
[14:34] <Velmont> Okay, can do with flac then, -- it'll get a bit smaller at least :-) Thanks
[15:30] <Velmont> Since I'm in a talking mood today -- I see that someone made ffmpeg able to read video from Panasonic MXF files from P2 cards. Thanks very much for that, however, it doesn't find/extract the audio.
[15:32] <Velmont> I guess the video file I uploaded also had sound. -- It works when I run mxfsplit on the video first, but that extra mxfsplit is slow.
[15:33] <Velmont> Wonder why Panasonic thought it a smart idea to embed a MXF file inside an MXF file... To make decoding hard I guess. Secret sauce "interopable" format.
[15:53] <elxa> is someone working  on h.264 mvc (3d bluray) support? I am not sure if those people talking about it on the mailing list are actually working on it.
[16:00] <Cobra_Fast> Hi! How do I resize a video while keeping codecs and quality? I want a very simple way that works automatically without someone actually writing the codec names in the command.
[16:03] <pappen> hello, I don't know how to set the quantizer (-q) exactly. I want to have an constant quantiser of 3.2 for xvid encoding. but if I set -q 3.2 it shows -q 2-31. How do I set it accurately?
[16:04] <Cobra_Fast> My current attempt is: ffmpeg -i 'test.mov' -s 1000x566 -f 'mov' pipe:1 > resized.mov - but that's giving me an arror saying my parameters might be incorrect althought i can see it detect everything correctly in the debug output.
[16:10] <Cobra_Fast> Oh apparently MOV isn't pipeable.
[16:27] <tr33fall> Output pad "default" for the filter "Parsed_null_0" of type "null" not connected to any destination -> anyone seen this before?
[16:27] <tr33fall> i just added "-s 640x480"
[16:27] <tr33fall> to my command
[16:45] <teratorn> hi! I'm having trouble with complex filtergraphs!! everyone's favorite subject I'm sure! using a command like this, ffmpeg -i sample.mp4 -i sample2.mp4 -filter_complex '[0:0] [0:1] [1:0] [1:1] concat=n=2:v=1:a=1 [v] [a1]' -map '[v]' -map '[a1]' -strict -2 -acodec aac -vcodec libx264 -crf 23 output.mkv, I want to also add an overlay PNG to the output result of the concat filter, but how??
[16:46] <MiteshShah> I'm following this guide to install ffmpeg and everything is ok but last comamnd hash show me ffmplay is not installed :(
[16:46] <MiteshShah> Any Idea
[16:47] <MiteshShah> Setup Guide: https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide
[16:47] <tr33fall> you need to --enable-ffplay perhaps?
[16:48] <tr33fall> i dont know if it is default or not
[16:49] <MiteshShah> its not specified in official guide and should i need to --enable-ffplay in ffmpeg configure?
[16:50] <MiteshShah> tr33fall, i'm new to ffmpeg can you tell me what the use of ffplay or any link where i can update my knolwdge
[16:51] <tr33fall> you can add --enable-ffplay
[16:51] <tr33fall> are you trying to build for what operating system btw?
[16:54] <MiteshShah> tr33fall, trying... Ubuntu 12.04.1 Server
[16:54] <MiteshShah> tr33fall, thanks for the helping me :)
[16:56] <MiteshShah> tr33fall, actually i'm making script  to install ffmpeg easily in ubuntu 8.04 10.04 10.10 11.04 11.10 12.04  (https://github.com/rtCamp/media-node/blob/master/setup/ubuntu.sh)
[17:01] <ubitux> MiteshShah: you need the sdl libs
[17:01] <ubitux> % grep HAVE_SDL config.h
[17:01] <ubitux> #define HAVE_SDL 1
[17:01] <ubitux> you must have this after configure
[17:02] <ubitux> teratorn: try to append "; [v][your-overlay-source] overlay [outv]" at the end of your filtergraph
[17:03] <ubitux> and -map '[outv]'
[17:03] <ubitux> (instead of [v])
[17:03] <teratorn> ok I don't understand what the -map args do :(
[17:03] <teratorn> but let me see here
[17:05] <denny--> Hi. I'm trying to write a Perl script which uses ffmpeg to process a chunk recorded off a live TV stream (I'm working for the rights-holder), cut out some bits their legal team advised them to remove before rebroadcasting (swearing etc), cut out the ad breaks, then patch the whole thing back together and produce an XML file detailing where the ads should be inserted when rebroadcasting it.  The first phase of the project was just re
[17:06] <denny--> ... and that went fine, but when I started on phase 2, removing the profanity-etc cuts, I found that my cuts are inaccurate (between half a second and one and a half seconds, roughly).  Any idea what could be causing that?
[17:07] <denny--> the original file is an MP4, it's converted to a TS before I start hacking away at it.  The process is more-or-less detailed in this debug output: http://pastebin.com/cGq2BHD3
[17:08] <denny--> I'm given the cut points as frame numbers and have to convert them to timestamps for the ffmpeg commands, but we've triple-checked the maths to death there, it's good.
[17:08] <denny--> I tried bumping keyframes up to one per second, that didn't make any difference
[17:08] <denny--> I'm stuck  :)
[17:08] <ubitux> teratorn: the -map options selects the outputs from your filtergraph you want in your output
[17:09] <ubitux> +file
[17:09] <denny--> oh, in case it's not clear from the commands - to cut the censored bits out, what I'm actually doing is extracting everything except those bits, and then stitching that back together
[17:09] <teratorn> ubitux: success!!!1
[17:09] <teratorn> ffmpeg -i sample.mp4 -i sample2.mp4 -i template2.png -filter_complex '[0:0] [0:1] [1:0] [1:1] concat=n=2:v=1:a=1 [v] [a1]; [v][2:0] overlay [outv]' -map '[outv]' -map '[a1]' -strict -2 -acodec aac -vcodec libx264 -crf 23 output.mkv
[17:09] <ubitux> :)
[17:10] <ubitux> you might want to use a better aac encoder though
[17:10] <teratorn> wow ffmpeg makes me look like a genius
[17:10] <ubitux> like fdk-aac or something
[17:10] <teratorn> ok will do thx
[17:10] <teratorn> how let me ask one more tricky question
[17:11] <teratorn> i have a batch script, which takes a video and a list of time segments to keep. it makes temp files for each segment to keep, then combines then, with an overlay, as in the above command
[17:11] <teratorn> how could i do the cuts, concat, and overlay in one filtergraph?
[17:11] <ubitux> denny--: that's because of the -codec copy; it's kind of hard to make a proper selection
[17:12] <denny--> is there a workaround or better way to approach the problem?
[17:12] <denny--> I need to maintain the quality of the file so that it can be rebroadcast
[17:12] <ubitux> teratorn: with a sendcmd filter (very) recently added i guess
[17:12] <ubitux> denny--: not that i know of
[17:13] <teratorn> ubitux: documented?
[17:13] <ubitux> yup
[17:13] <ubitux> http://ffmpeg.org/ffmpeg.html#asendcmd_002c-sendcmd
[17:13] <ubitux> it *might* be possible
[17:13] <teratorn> much thanks
[17:13] <ubitux> but it was comitted a few hours ago
[17:13] <ubitux> so still experimental ;)
[17:13] <teratorn> bahaha
[17:14] <ubitux> feel free to try it :)
[17:14] <teratorn> hold hold
[17:14] <teratorn> im still compiling ;)
[17:14] <ubitux> i'm not even sure the filters are yet ready
[17:15] <ubitux> saste_: so can we make some kind of edit list with sendcmd? :)
[17:15] <MiteshShah> ubitux, Don't understand :(  How to install sdl libs
[17:15] <teratorn> well, I will definitely want to try it at some point
[17:15] <teratorn> for my, my temp files with vcodec copy get the job done :)
[17:15] <teratorn> *for now
[17:15] <ubitux> MiteshShah: ubuntu?
[17:15] <MiteshShah> ubitux, yes
[17:16] <saste_> ubitux, we need to extend the overlay
[17:16] <ubitux> MiteshShah: the something like apt-get install libsdl or something similar
[17:16] <saste_> with a reinit command we could change the source of overlay
[17:16] <ubitux> then*
[17:16] <saste_> *assuming* it has the same size
[17:16] <ubitux> saste_: i think the point is doing a selection first
[17:17] <saste_> and all the stuff related to reconfiguration
[17:17] <saste_> ubitux, what kind of selection?
[17:17] <ubitux> like doing various scene selection (and the overlay will just be appended on that output)
[17:17] <ubitux> well just like multiple -ss and -t concatened
[17:17] <ubitux> i guess
[17:17] <ubitux> at least that's how i understand it
[17:18] <saste_> ah an edit list...
[17:18] <ubitux> yup.
[17:18] <saste_> not simple at the moment
[17:18] <ubitux> saste_: maybe we should write a dedicated filter for this actually
[17:18] <MiteshShah> ubitux, when i enter ffplay apt-get suggest me to install libav-tools should i install libav-tools or libsdl
[17:18] <ubitux> who acts a bit like vf select but with a playlist syntax file
[17:19] <ubitux> install libsdl, so the configure will detect the sdl libs on your system, and you'll be able to build ffplay
[17:19] <saste_> ubitux, that sounds more like a demuxer
[17:19] <ubitux> oh right a demuxer could do the trick
[17:20] <saste_> currently you could create a movie for each source, connect them all to concat, and connect concat to overlay
[17:20] <saste_> (i don't know why overlay is needed)
[17:20] <ubitux> ah right movie source has some start & end options
[17:20] <saste_> has it?
[17:20] <ubitux> hasn't it?
[17:21] <saste_> it can seek, but iirc can't specify the end time
[17:21] <ubitux> sp
[17:21] <ubitux> seek_point
[17:21] <saste_> i also thought about a seteof filter which cut a stream at some point
[17:21] <ubitux> no duration though
[17:21] <saste_> or we could add a duration option to movie
[17:21] <ubitux> yeah would be nice to have duration
[17:21] <saste_> ubitux, patch?
[17:22] <ubitux> but it also means for each selection you decode everything until the sp for the same input
[17:22] <ubitux> :)
[17:22] <ubitux> btw, about the timing
[17:22] <ubitux> i wonder if that won't be an issue
[17:22] <saste_> it is
[17:22] <saste_> seeking is not accurate in ffmpeg
[17:22] <saste_> movie is no different
[17:22] <ubitux> no i mean about the gap
[17:23] <ubitux> like if you cut 3 seconds at the begining and 3 seconds one hour later
[17:23] <ubitux> the ts will have a huge hgap
[17:23] <ubitux> -h
[17:23] <saste_> yes
[17:23] <ubitux> won't that be corrected or something?
[17:23] <saste_> that's why god created the setpts filter
[17:23] <ubitux> :D
[17:23] <ubitux> \o/
[17:23] <saste_> btw
[17:23] <saste_> sometimes you *want* to keep the gap
[17:24] <ubitux> oh and btw, would it be possible to use setpts with vcodec copy?
[17:24] <saste_> so i thought a timepad filter (both for audio and video) may be useful at some point
[17:24] <saste_> setpts + copy? no sense
[17:24] <ubitux> in what situation would you want that?
[17:24] <ubitux> why non-sense?
[17:25] <ubitux> isn't pts only useful at demux/mux time?
[17:25] <saste_> to fill the gap: for example to create *continuous* streams in case more filters (or demuxers) don't cope very well with gaps
[17:26] <saste_> setpts+copy: that's impossible, setpts works at the filtering level
[17:26] <saste_> a bitstream filter doing PTS mangling may be possible though I suppose
[17:26] <saste_> or maybe even something at the lavf level
[17:27] <ubitux> maybe we should continue this discussion on the -devel channel :p
[17:32] <denny--> looking at the docs again... I could use -vframes to specify the duration of my extracts, which might work better, but I don't see a frame-based way to specify the starting point?
[17:37] <ubitux> vf select
[17:38] <ubitux> but filtering ’ re-encode
[17:39] <denny--> ah
[17:41] <denny--> -vframes isn't a filter though, right?  so I can use that and maybe cut out my drift on the endpoints at least.
[17:42] <ubitux> it just counts frame
[18:06] <teratorn> ugh OK new problem. when I make my temp cut files, I use vcodec copy. which *seems* like it might be leaving PTS/DTS values unmodified...... this causes problems it seems when I try to use the concat filter on them
[18:06] <teratorn> in particular, [mov,mp4,m4a,3gp,3g2,mj2 @ 0x16aa240] max_analyze_duration 5000000 reached at 5007302
[18:06] <teratorn> [mov,mp4,m4a,3gp,3g2,mj2 @ 0x16aa240] Could not find codec parameters for stream 0 (Video: h264 (avc1 / 0x31637661), 1024x768, 619 kb/s): unspecified pixel format
[18:06] <teratorn> Consider increasing the value for the 'analyzeduration' and 'probesize' options
[18:07] <teratorn> does my theory make sense?
[19:00] <teratorn> hi guys, I'm using -ss <seek> and -t <duration> to "cut" a video. but it only cuts to nearest keyframe. How can I get it to do key-frame splitting and make the cut precise?
[19:02] <ChannelZ> Are you using 'copy' to try to do it without recompressing?
[19:03] <teratorn> I was, then I tried with full transcoding
[19:03] <teratorn> same result :/
[19:03] <ChannelZ> hmm
[19:04] <teratorn> oh crap it makes a diff it -t and -ss go BEFORE The input, or AFTER
[19:04] <teratorn> http://superuser.com/questions/458761/accurately-cut-video-files-from-command-line
[19:04] <teratorn> makes perfect sense now!
[19:06] <ChannelZ> nice
[19:29] <denny--> ChannelZ: I was asking a similar question about cut accuracy earlier... is there no way to get it accurate while using the copy codecs?  Regardless of file format etc?  Someone here claims they managed it by doing the cuts on mp4 files, but I can't replicate that.
[19:29] <denny--> by 'here' I mean where I'm working, not this channel
[19:30] <denny--> I tried setting every frame to be a keyframe, that didn't help either
[19:53] <mkrufky> i want to add a feature to my TV app to transcode broadcast television (mpeg2) into h.264 on-the-fly ... does anybody know of any sample source code that i can look at for transcoding a live stream?
[19:57] <saste_> mkrufky, and doc/examples if you want to start with something simpler
[19:57] <mkrufky> thanks saste_, i saw your message in the other channel
[19:58] <mkrufky> sometimes the obvious answer is the last that comes to someone's mind ;-)
[19:58] <saste_> mkrufky, ffmpeg.c is (really) hard to read
[19:58] <mkrufky> i waould assume so :-/
[19:59] <saste_> i'd suggest to start from doc/examples
[19:59] <mkrufky> meh.  im sure this would be extremely easy to do form the command line
[19:59] <mkrufky> yes, ive been looking at doc/examples so far
[19:59] <saste_> well you can always do a system(...) from your program
[19:59] <mkrufky> yes i could, but it would be horribly inefficient
[20:00] <mkrufky> imagine the amount of extra unneeded memcpy's
[20:04] <mkrufky> it looks like ffmpeg.c has what i was looking for ...  i should have just looked there to begin with ;-)
[20:05] <saste_> mkrufky, -re
[20:05] <mkrufky> (now im embarassed cuz i dont know what '-re' means :-P )
[20:06] <saste_> mkrufky, ffmpeg -re -i INPUT ... OUTPUT
[20:06] <mkrufky> oh, lol
[20:06] <saste_> -re == real time
[20:06] <mkrufky> thanks
[22:43] <Mista_D> I'm looking at upping file system RW block size of FFmpeg. Want to test performance with various NFS/TCP frames. Where can I modify it?
[23:39] <JCDenton> Any opinions on WebM from anyone here..? I had a bit of difficulty getting support for it to compile into ffmpeg on my server box a week or so ago, but I do have it working on my laptop. I haven't ever bothered trying to encode anything in it, but I've seen a lot of mxied opinions and varied results.
[23:40] <JEEB> JCDenton, it is somewhat on par to H.264 baseline profile with relatively quality-related settings
[23:41] <JEEB> and this is spec-wise (VP8 and H.264)
[23:41] <JEEB> encoder-wise libvpx has gotten better than it was, but it's still slower and harder to use than libx264
[23:42] <JEEB> quality-wise I think if you set libx264 to baseline and thunk libvpx up to eleven it might be on par, otherwise libx264 is going to just fly by
[23:45] <JCDenton> Thanks, very helpful response. I was largely curious about the comparison quality wise too. I probably won't bother caring about it much right now, maybe I'll play with it sometime just for fun, but I'm still happy using libx264 right now.
[00:00] --- Tue Sep 25 2012


More information about the Ffmpeg-devel-irc mailing list