[Ffmpeg-devel-irc] ffmpeg.log.20120704

[burek]

[00:16] <idodeisuke> any1 knows, where to find ffmpeg binary tarballs?
[00:32] <beandog> FF|taqattack: http://ffmpeg.org/legal.html
[00:35] <FF|taqattack> beandog, those only pertain to LGPL version.
[00:47] <mxmx> I get this error PHP Warning:  PHP Startup: Unable to load dynamic library '/usr/lib64/php/modules/ffmpeg.so' - libswscale.so.0: cannot open shared object file: No such file or directory in Unknown on line 0
[00:48] <mxmx> libswscale.so.0 doesnt exist in the lib directory, only libswscale.so.2
[00:48] <mxmx> how can i fix this?
[01:05] <iive> mxmx: the current library version of swscale is 2.1.100
[01:06] <iive> I suspect you may be using ffmpeg-php, that despite the name is completely separate project.
[01:07] <iive> btw, in case the package you got comes with the needed libraries. run ldconfig as root.
[01:08] <mxmx> i ran ldconfig as root but didnt help
[01:09] <mxmx> i also raon ldconfig -v
[01:10] <mxmx> i located libswscale.so.0 in another lib, made a sym link but then the error becase that the symlink had too many levels
[01:11] <iive> i've seen too many levels only when it is recursive, aka forms a curcle.
[01:11] <iive> find the real one.
[02:14] <a|3x> hi
[02:15] <a|3x> when i specify -vcodec libmpeg4 my video shows up fine but when i use libx264 all i see is a gray screen, what could be wrong?
[02:17] <burek> FF|taqattack, if you have read "legal" documentation for ffmpeg (the link you were given) then you can conclude that as long as you respect those mentioned points, you're on the right side
[02:18] <burek> i.e. if you clearly show that your project includes ffmpeg and provide complete source code to the end users, letting them know that you used ffmpeg, etc. you're good to go (and of course don't charge people for ffmpeg stuff)
[02:18] <burek> if you try to "embed" ffmpeg into your application, obfuscating the fact that you used it, pretending like you have written all that, then you might be in a legal trouble
[02:18] <burek> that's for short
[02:19] <burek> idodeisuke, what kind of binary tarbals?
[02:20] <idodeisuke> burek, any kind of, for linux.
[02:20] <burek> mxmx, it would be easier if you would ffmpeg directly, using shell_exec()
[02:20] <burek> avoiding all the mess with dependencies
[02:21] <burek> or go to the download page of ffmpeg
[02:38] <idodeisuke> when i tried to run it on my shell account it said the following: FATAL: kernel too old Segmentation fault: 11
[02:38] <idodeisuke> uname -a: FreeBSD shellmix.com 8.2-RELEASE-p9 FreeBSD 8.2-RELEASE-p9 #1: Sat Jun 30 19:14:12 UTC 2012     spaj at shellmix.com:/usr/src/sys/amd64/compile/kernel  amd64
[02:39] <sacarasc> You can't use a Linux static build on FreeBSD...
[02:45] <iive> a|3x: never heard of libmpeg4, i had no idea ffmpeg supports it, is it asp or avc?
[02:54] <a|3x> not sure, i typed it in and it seemed to work
[02:57] <iive> are you sure the gray picture doesn't come from it?
[02:59] <a|3x> no, in fact, when i am using libmpeg4 it shows up fine, but when i am replacing it with libx264 then it becomes gray
[02:59] <a|3x> burek, iive: here is that pastebin: http://pastebin.com/03tk3gv6
[03:00] <idodeisuke> ar6e there any static freebsd biuld around?
[03:02] <iive> a|3x: hum "Incompatible pixel format 'uyvy422' for codec 'libx264', auto-selecting format 'yuv422p' "
[03:02] <iive> your input is a raw.
[03:03] <a|3x> ya so?
[03:03] <a|3x> doesn't seem to bother other video codecs
[03:03] <burek> a|3x, can you update your ffmpeg?
[03:03] <burek> and your x264
[03:04] <a|3x> i updated x264 by downloading and compiling source
[03:05] <iive> the problem is most likely in the swscale
[03:05] <iive> but yes, there is 10.4 and 11.2 releases.
[03:05] <a|3x> how do i fix this problem?
[03:06] <iive> yours is 10.3 , the configure line looks like it is from gentoo ;)
[03:07] <iive> a|3x: the swscale converts image size and color format.
[03:08] <iive> for some reason it may not be working, e.g. returning error. then the buffers that are fed to x264 would be empty and they are by default gray.
[03:08] <iive> so, first thing is try a newer release, as i remember some swscale fixes few months ago.
[03:09] <iive> in theory you may be able to workaround it by requesting another set of convertion.
[03:09] <iive> e.g. yuyv422 into yuv420
[03:10] <iive> probably by placing -pix_fmt yuv420 next to the -vcodec
[03:13] <a|3x> ok i upgraded to 0.11.1 and now it just says pipe:: Invalid data found when processing input
[03:14] <a|3x> oh never mind
[03:16] <a|3x> nope, same gray screen with new version of ffmpeg
[03:17] <a|3x> Unknown pixel format requested: yuv420.
[03:19] <a|3x> oh its yuv420p
[03:20] <a|3x> amazing, it seems to work
[03:25] <iive> a|3x: can you make a small sample with the raw 'uyvy422'  in it and fill bugreport at the tracker?
[03:25] <iive> this is bug and it should be tracked and fixed.
[03:35] <a|3x> i wish i could but i got so many things to do it's not possible
[03:36] <a|3x> does anybody know of a good rtmp streamer that works like ffmpeg except allows to specify authentication parameters?
[03:39] <burek> im afraid we have so many things to do so it's not possible to answer your questions no more
[03:43] <a|3x> damn
[03:44] <a|3x> by the way, that static ffmpeg build doesn't work for me
[03:44] <a|3x> ffmpeg: ../sysdeps/unix/sysv/linux/getpagesize.c:32: __getpagesize: Assertion `_rtld_global_ro._dl_pagesize != 0' failed.
[03:44] <burek> which OS you use
[03:44] <a|3x> gentoo
[03:45] <burek> is that linux? :D
[03:45] <burek> then probably your kernel is old or something, I dunno..
[03:45] <a|3x> Linux host 2.6.32.59 #1 SMP Mon Jul 2 13:47:52 PDT 2012 x86_64 Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz GenuineIntel GNU/Linux
[05:09] <a|3x> so it seems rtmp streaming doesn't work damn it nothing ever works
[07:12] <lake> hi all, i am trying to stream "/dev/video0" from the v4l2 driver. I am unable to capture audio.
[07:13] <lake> ffmpeg -f alsa -i hw:2,0 -f v4l2 -i /dev/video0 foo.mpg
[07:13] <lake> cat /proc/asound/pcm returns this:
[07:13] <lake> 02-00: USB Audio : USB Audio : capture 1
[10:51] <damascene> Hi, Is there any command to check if an audio or video file is corrupt or not?
[12:15] <mxmx> fflogger
[12:52] <jameshowe> hello, is there a reference somewhere of CLI changes between 0.9 and 0.11 ?
[13:03] <jameshowe> like how trying to do -s or -vf or -vframes before -i doesn't work anymore
[13:08] <brocatz> can't you do a diff on the code base
[13:08] <brocatz> all that shit is in one big .c file
[13:22] <atlithorn> hi all, I'm combining 2 live streams via overlay and there is a time lag between them, I'm pretty sure it's just the time difference between ffmpeg connecting to stream1 and then stream2.
[13:22] <atlithorn> The result is that stream1 is about 7 seconds behind in the output. Does anyone know how I can get rid of that?
[13:25] <jameshowe> brocatz: which c file? Providing documentation of breaking changes is more usual
[13:26] <atlithorn> I tried setpts=0.8*PTS on stream1 so it would catch up and it does but that effects stream2 slowing it down, any idea how I could negate that or is there a better way to do this?
[13:31] <brocatz> ffmpeg.c
[13:31] <brocatz> in the source code
[13:31] <brocatz> it'd be easy enough to diff both versions of it
[13:31] <brocatz> the code isn't very complicated
[13:36] <jameshowe> it's a 6000-line file
[13:36] <jameshowe> any clue as to where CLI flags are parsed?
[13:40] <jameshowe> and there's no mention of the -er flag (which now doesn't work) in either that or cmdutils_common_opts.h
[13:40] <jameshowe> in either version
[13:52] <burek> atlithorn, use -itsoffset
[13:52] <atlithorn> thanks burek
[13:53] <jameshowe> for example any command line with "-er 4" in it gives the error "Unrecognized option 'er'"
[13:53] <jameshowe> anything with a -s before -i gives "Option video_size not found"
[13:53] <burek> jameshowe, please give us pastebin
[13:53] <burek> to figure out what do you need
[13:53] <burek> and you'll get your answers faster
[13:53] <jameshowe> i need to know what the breaking changes are
[13:53] <burek> ok
[13:54] <jameshowe> this isn't for a specific thing
[13:54] <jameshowe> i have read that, yes
[13:55] <burek> well then you already know your answers
[13:55] <jameshowe> know, I can look up what the CLI currently is
[13:55] <jameshowe> assuming it's up-to-date
[13:56] <jameshowe> most software projects explicitly tell their users what they've broken, and how the equivalent things are done in the new version
[13:57] <jameshowe> so it's possible to upgrade without having to manually cross-reference a whole code-base against two massive sets of documentation
[13:57] <burek> I agree, but if you are looking for feature suggestions, then use ffmpeg trac system to suggest it
[13:57] <burek> if you are looking for the answers, please do like suggested
[14:30] <varaderoguy> afternoon chaps
[14:31] <varaderoguy> wonder whether somebody could tell me what RED Camera Support is like at the moment
[14:31] <varaderoguy> known as r3d I think in ffmpeg terms
[14:33] <varaderoguy> I've been following some chat threads about this, although it seems to be okay for 'original' RED camera support, it seems that when they introduced encryption, then the decoder fell apart
[14:33] <varaderoguy> Where are we now?
[14:36] <varaderoguy> I understood that RED camera was based on JPEG2000
[14:37] <burek> you are probably asking this in a wrong channel
[14:37] <varaderoguy> burek: where do I need to go then?
[14:38] <burek> for a beer :D
[14:38] <varaderoguy> okay - #ffmpeg-devel
[14:38] <leandrosansilva> Hello to all. I have just updated ffmpeg and the avfilter_poll_frame doesn't exist anymore. I couldn't find in the APIChanges any message about it being depreciated. Which function replaces that one?
[14:40] <burek> leandrosansilva, do just a quick grep on ffmpeg src folder
[14:40] <burek> and see how has it been replaced
[14:41] <leandrosansilva> burek, I couldn't find any occurency of this function in ffmpeg source tree.
[14:41] <burek> well that's expected if it's been replaced :D
[14:42] <leandrosansilva> there's just a message in APIChanges about it's addition
[14:42] <leandrosansilva> doc/APIchanges:  Add av_buffersink_poll_frame() to buffersink.h.
[14:42] <burek> you didn't understand me maybe
[14:43] <burek> take a look at the ffmpeg's code that used that function before
[14:43] <burek> and see what does it use now
[14:43] <leandrosansilva> ops
[14:43] <leandrosansilva> I think I found
[14:43] <leandrosansilva> av_buffersink_poll_frame
[14:46] <leandrosansilva> Is it normal removing a function without mark it as deprecated before?
[14:46] <burek> :) ask in #ffmpeg-devel
[14:46] <leandrosansilva> of course :-)
[14:46] <burek> this channel is for ffmpeg usage
[14:50] <leandrosansilva> burek, of course. thx
[14:52] <burek> :beer: :)
[15:45] <rainmaker1> Hi, it seams that new version of ffmpeg need extra args if I specify yadif as a filter. I got a this message Error parsing options string: video_size=0x0:pix_fmt=-1:time_base=1/90000:pixel_aspect=0/1:sws_param=flags=2
[15:45] <rainmaker1> I only have -filter yadif
[15:46] <rainmaker1> in my command line as a filter, when I remove -filter yadif ffmpeg works, but I need to deinterlace picture
[15:47] <rainmaker1> any thoughts?
[15:59] <ubitux> rainmaker1: -filter:v, or -vf
[16:06] <rainmaker1> just a sec to try
[16:09] <rainmaker1> Here it is
[16:09] <rainmaker1> http://pastebin.com/3BSWMH0N
[16:11] <burek> what is "overrun_nonfatal"
[16:11] <rainmaker1> ubitux: -filter yadif I use, but the same result with -filter:v
[16:13] <burek> try "-vf yadif" instead of "-filter yadif=0:-1:0"
[16:13] <rainmaker1> burek: to overcome the problem with curcular buffer overun error we were hitting
[16:13] <burek> rainmaker1, where did you read about that option "overrun_nonfatal"
[16:13] <rainmaker1> same error
[16:13] <rainmaker1> burek: here ofcourse
[16:14] <burek> and where exactly is "here" :)
[16:14] <rainmaker1> on #ffmpeg
[16:14] <rainmaker1> but if you ask who told me... I don't know
[16:14] <rainmaker1> however
[16:14] <rainmaker1> I can try to findout if that matters
[16:14] <rainmaker1> :)
[16:15] <burek> try this ffmpeg -f mpegts -y -i udp://230.0.0.58:5000 -vf yadif -c:v libx264 -an out.ts
[16:15] <burek> no, I'm just curious
[16:15] <burek> because there is no such option in docs
[16:15] <rainmaker1> well well well :)
[16:15] <burek> and I'd like to know is it missing in the docs
[16:15] <rainmaker1> 22:49 < burek> Circular buffer overrun. To avoid, increase fifo_size URL option. To survive in such case, use overrun_nonfatal option
[16:15] <burek> or it is non-existent
[16:16] <burek> I never said that?
[16:16] <burek> where did you copy that from
[16:16] <rainmaker1> yes you did
[16:16] <rainmaker1> from my log files
[16:16] <burek> man I don't even know such option exists
[16:16] <burek> so I'm pretty sure I didn't
[16:16] <rainmaker1> well... :)
[16:18] <burek> ./libavformat/udp.c:262: *         'overrun_nonfatal=1': survive in case of circular buffer overrun
[16:18] <burek> you are right, it exists
[16:18] <burek> it's a docs bug
[16:22] <rainmaker1> :D
[16:22] <rainmaker1> at last :)
[16:23] <rainmaker1> hmmmmm no
[16:23] <rainmaker1> it still do not work
[16:43] <rainmaker1> sorry my mistake, -filter does not work -vf works
[16:43] <rainmaker1> thank you all
[16:45] <ubitux> -filter:v should work as well
[16:46] <rainmaker1> strange that -filter was working till I upgraded to git version (the old one was few months old)
[16:48] <burek> rainmaker1, some changes happened with ffmpeg's syntax to better specify each option
[16:48] <burek> so :v and :a were also added to specify video/audio stream
[16:48] <burek> because there are audio filters too, and it might make ambiguity using just -filter
[16:50] <rainmaker1> ubitux: yes it works
[16:50] <rainmaker1> burek: nice, thank you for info :)
[16:51] <burek> :beer: :)
[16:51] <rainmaker1> :beer: :)
[16:54] <cmon> Hi - am trying to compress 1.4 gb video to 500 mb. Is there a good way of using ffmpeg for the job?
[16:55] <sacarasc> Work out the bitrates and use them!
[16:56] <burek> :)
[16:56] <burek> cmon, can you type ffmpeg -i your_input_file
[16:56] <burek> and use pastebin.com to show the output
[16:56] <cmon> I tried out using  -acodec command and got it to 157 mb but with great loss of quality
[16:57] <atlithorn> hi burek, you helped me an hour or so ago with itsoffset, my problem is that it only seems to offset the video...
[16:57] <atlithorn> i have a complex filter that is mixing video via overlay and audio via amix
[16:57] <atlithorn> itsoffset did the trick for the video but the audio is still out of sync, shouldn't it work for both?
[16:59] <burek> atlithorn, I'm not sure, but I think -itsoffset works for video only.. maybe you could use -async -isync or -vsync combined with -itsoffset to reach the solution..?
[17:00] <atlithorn> thanks again, I'll have a look
[17:02] <damascene> Hi, Is there any command to check if an audio or video file is corrupt or not?
[17:03] <jameshowe> does anyone maintain Windows builds of ffmpeg with --enable-libfaac ?
[17:03] <burek> damascene, try to convert it?
[17:04] <burek> jameshowe, did you check Zeranoe's builds
[17:04] <jameshowe> yes, those don't
[17:04] <JEEB> it's pretty hard to maintain such libfaac being nonfree
[17:05] <JEEB> libvo-aacenc and the internal aac encoders aren't nonfree, but IIRC libvo-aacenc fails at >2ch and the internal one isn't exactly nice at lower bitrates :)
[17:07] <JEEB> basically it was found a few years ago that faac contained non-GPL-compatible sources from the reference implementation, and thus it had to be added to the list of "nonfree" [non-redistributable] libraries
[17:07] <jameshowe> a dynamic-link one would be fine - I already have the dll
[17:07] <JEEB> well, to enable faac you have to --enable-nonfree
[17:07] <JEEB> you can't really sidestep it with any kinds of autosearch or whatever because ffmpeg doesn't have that :P
[17:08] <damascene> burek: there is a lot of video and audio files. How much time that will take?
[17:08] <parceval> hello
[17:09] <damascene> how about this solution? http://superuser.com/questions/100288/how-to-check-the-integrity-of-a-video-avi-mpeg-file
[17:09] <JEEB> jameshowe, for 2ch stuff you can use libvo-aacenc, and for >2ch you can use the internal aac encoder (or you can compile ffmpeg yourself). Or you just encode the AAC track separately.
[17:09] <parceval> how can i extract a list of frames at specific timecodes with ffmpeg in one go instead of starting it for each frame i want to extract
[17:10] <JEEB> (there's encoders like nero's or multiple that use QT's aac encoder backend)
[17:10] <burek> damascene, you can use lossless raw "encoding" and output it to /dev/null
[17:10] <JEEB> you can basically pipe the decoded audio into those and then mux it together with the video after both have been finished
[17:11] <burek> like this ffmpeg -y -i file -vcodec rawvideo -acodec pcm_s16le /dev/null
[17:11] <burek> it should be pretty fast
[17:11] <JEEB> the situation might get better when licensing etc. gets set up with fraunhofer's aac encoder
[17:11] <JEEB> *gets settled
[17:11] <damascene> ah, thanks
[17:12] <burek> damascene, on that link, a guy is basically doing the same
[17:12] <jameshowe> and finally, there's no mention in the documentation of -preset or how the old -vpre files map to it
[17:12] <burek> just, he uses additional encoding, after the decoding, which you don't really need, to save some time
[17:13] <damascene> thanks then I will try your solution
[17:14] <burek> :beer:
[17:14] <JEEB> jameshowe, I think -vpre still works as it did, if it doesn't its features were just completely moved to -preset. Also, for libx264 you are better off using the internal presets via -preset anyways
[17:14] <burek> parceval, I think  you can't so far
[17:15] <jameshowe> yes, but the default preset files were all removed
[17:15] <JEEB> for libx264, yes
[17:15] <jameshowe> and -preset isn't documented
[17:15] <burek> jameshowe, x264 --help
[17:15] <burek> it's mapped directly to x264
[17:15] <JEEB> or http://mewiki.project357.com/wiki/X264_Settings#preset
[17:17] <jameshowe> thankyou for the link
[17:17] <jameshowe> I don't have a program called x264
[17:18] <JEEB> >_>
[17:18] <burek> hm :) they actually don't have man page :)
[17:18] <JEEB> yes
[17:18] <JEEB> just link what I linked when needed
[17:19] <burek> http://pastebin.com/ZKkemcG5
[17:19] <burek> :)
[17:22] <jshanab> I have been using libavcodec in my apps on desktops so I am trying to use it in IOS on iPhone. I compiled libavcodec and wrote a player, but I have trouble on higher quality video in the decode time is longer than play time so it runs-pauses-runs-pauses...  I have all scaling and colorspace conversion offloaded to the GPU and profiling shows 49% cpu on the call to avcodec_decode_video2
[17:24] <jshanab> I am trying to get it to use more CPU to get done faster but other I/O like streaming texture to card are also needed. Are there settings I can put in my context to speed up decodeing. This stream is 640x480 30fps H264 on the high quality setting
[17:25] <JEEB> not really, you could try cutting off in-loop deblocking but that will lead to breaking of the picture
[17:25] <JEEB> I really recommend you use the dedicated hardware decoding on such mobile things
[17:25] <JEEB> because as you can see using the CPU isn't exactly efficient on any scale of the word
[17:26] <damascene> is this a problem "Unable to find a suitable output format for '/dev/null'" ?
[17:26] <burek> add -f null before /dev/null
[17:27] <damascene> now it's working
[17:27] <sacarasc> Or -f mkv, -f mp4
[17:27] <burek> http://ffmpeg.org/ffmpeg.html#null-1
[17:27] <sacarasc> Doesn't really matter, as it's going to /dev/null anyway. :D
[17:28] <burek> you can use other formats too, but if you specify -f null then ffmpeg won't waste cpu on creating the output
[17:28] <jshanab> JEEB. on the iPhone the specific requirements to use the hardware decodeing are a problem for our setup. I was trying to find out what is the loop filter, is that the deblocking? If so do I disable by setting pCodecContext->skip_loop_filter to 1?
[17:28] <JEEB> yeah
[17:28] <JEEB> I think so
[17:29] <damascene> burek: it's working but it's not extremely fast
[17:29] <JEEB> of course all the quality goes off with that of course :P
[17:29] <JEEB> because you are skipping a part of decoding that is part of the specification
[17:29] <burek> damascene, it uses the least time needed for at least decoding your video
[17:29] <JEEB> jshanab, basically "welcome to how powerful ARM CPUs are"
[17:29] <burek> that's the only way I can think of right now to check if your videos are valid
[17:29] <jshanab> You said "break picture" does the deblocking filter just handle the interpolation when the image is scaled up? or is it part of the macroblock smoothing
[17:30] <damascene> burek: no worries.
[17:30] <JEEB> it's a part of the encoding and decoding
[17:30] <JEEB> thus you will get macroblocks and other fun as stuff won't match up properly
[17:30] <JEEB> s/macroblocks/blocking artefacts/
[17:31] <damascene> what kind of errors I should expect from a corrupted video? Should I examine manually? or the list at the bottom of this solution is sufficient? http://superuser.com/questions/100288/how-to-check-the-integrity-of-a-video-avi-mpeg-file
[17:31] <jshanab> JEEB. I am right on the edge with this stream. Most of our streams will use only 17% CPU for decode.
[17:32] <JEEB> also you might just encode all of your videos with --tune fastdecode in x264
[17:32] <JEEB> that turns off CABAC and in-loop deblocking etc.
[17:33] <jshanab> Somewhere I saw something about skipping if not needed. I could imagine the keyframe, which is the one who's decoding blows me out of the water, does not depend on the previous frame. Is that settable?
[17:33] <jshanab> JEEB. I do not have control over the encoding. ... yet
[17:33] <burek> damascene, usually an exit code check would suffice
[17:33] <burek> if ffmpeg encountered an error it will return non-zero value
[17:33] <JEEB> I'm not sure deblocking is related to whether or not the frame is an IDR frame or not
[17:34] <JEEB> it's either used or not used
[17:34] <damascene> burek: cool
[17:35] <jshanab> JEEB. Thanks. I will turn it off just to see if that is my issue. It may be that I just move my limit in my "supports iphone up to {fill in specs}"
[17:36] <JEEB> yeah, that might be a not-so-bad idea
[17:36] <jshanab> It amazes me how the difference in quality setting effects bitrate and decode time.
[17:38] <JEEB> what really affects decode time are the settings mentioned in the --tune fastdecode's help
[17:38] <JEEB> otherwise it's also a matter of "how much data there is" (bitrate)
[17:39] <jshanab> JEEB the tune params are on encodeing? not build or decodeing right
[17:39] <JEEB> yes
[17:40] <jshanab> When do I set skip_loop_filter to 1? it does not seem to make a difference
[17:41] <JEEB> I don't know how you exactly set it programmatically but if it doesn't help then it doesn't help >_>
[17:41] <JEEB> (you might want to check something in various libavcodec-using applications that let you set that setting)
[17:42] <jshanab> Well my code sets up the context then opens the codec. I was never sure which settings have to be set before open and which may get wiped out.
[17:46] <jshanab> I just want to check one thing. I have H264 just I-frames and P-frames. My understanding is that each p frame depends on the accumulated state since the last keyframe. Therefore. I cannot drop any of the p frames. (well except at end.  I jsut want to make sure there isn't a choice I am missing
[17:52] <jameshowe> hello, me again: http://pastebin.com/vEpE2nvh
[17:53] <JEEB> I have a feeling the preset file is outdated
[17:54] <JEEB> which iphone are you aiming this at anyways?
[17:54] <jameshowe> it's only got 4 lines, you want it here or pb?
[17:55] <jameshowe> vcodec=libx264 vprofile=baseline
[17:55] <jameshowe>  level=30
[17:55] <jameshowe> maxrate=10000000
[17:55] <jameshowe> bufsize=10000000
[17:55] <jameshowe> sorry, i thought i put those all on the same line...
[17:56] <JEEB> -profile:v baseline -level 30 -maxrate 10000k -bufsize 10000k
[17:56] <JEEB> should be an equivalent
[17:57] <JEEB> -vcodec libx264 -preset slow -profile:v baseline -level 30 -maxrate 10000k -bufsize 10000k
[17:57] <jameshowe> also, it seems to treat the input stream as 25 fps, whereas it's supposed to be 16 fps
[17:57] <JEEB> also I kind of would guess that using crf instead of bitrate would be a good idea :)
[17:58] <JEEB> you seemingly set -r 16 before -i so it _should_ get read as 16fps, and then converted to 25fps from that
[17:58] <JEEB> (since you then set -r after -i)
[17:58] <JEEB> although I'm not sure if stuff around that has changed
[17:58] <jameshowe> is that stderr output lying then?
[17:59] <JEEB> no idea
[17:59] <JEEB> also that ipod preset is basically something aimed at the first H.264-capable things I guess
[17:59] <jameshowe> yep
[18:00] <jameshowe> we have broad device support :)
[18:00] <JEEB> ok, if you specifically need broad support then baseline/level 3.0 is a relatively good idea
[18:00] <JEEB> just wanted to make sure you knew
[18:03] <jameshowe> yep, setting those options instead of the vpre sorts it out
[18:03] <jameshowe> still doubtful about the framerate though
[18:40] <jameshowe> despite what stderr says, it does appear to work correctly RE input framerate
[18:40] <sacarasc> \o/
[19:38] <automatical> hey all, i've got a video that is registering at 24 frames a second (h264), i'm hoping to convert this file to ogv, but slowed down to 15 frames a second. I tried the command: "ffmpeg -i infile.h264 -r 15 outfile.ogv", but the video is still as fast as it was before - any ideas?
[19:59] <ubitux> :D
[19:59] <ubitux> oups.
[20:03] <burek> automatical, update your ffmpeg
[20:03] <automatical>  i'm using version 0.8.1-4, was it fixed in a later release?
[20:04] <burek> try 0.11.2
[20:04] <automatical> kk
[20:07] <automatical> i'm guessing that's current trunk?
[20:07] <automatical> as the last release was 0.11.1
[21:00] <marcosmlopes> Hello guys, anyone have some resources explaining how video streaming works? I have created an node.js app that gets ffmpeg webm video and outputs in an URL, but the video doesn't play. I think that there is something underneath that tells the server how to sync the video with the source. Is that right?
[21:21] <a|3x> lets say i need ffmpeg to write to two files at the same time (with same codecs and data and everything), how do i make it encode once, write twice?
[21:22] <marcosmlopes> burek: http://pastebin.com/56BMrx7V
[21:23] <marcosmlopes> burek: i have build an app to receive this video stream with node.js. Do you want to see?
[21:23] <marcosmlopes> its 50 LoC
[21:25] <burek> a|3x, try ffmpeg -i input -vcodec <coding options> f mpegts - | ffmpeg -f mpegts - -vcodec ... output1 -vcodec ... output2
[21:25] <burek> -f mpegts*
[21:26] <burek> i.e. let the first ffmpeg process does the encoding and 2nd just multiply the input to 2 outputs
[21:26] <burek> actually something like this: ffmpeg -i input -vcodec <coding options> -f mpegts - | ffmpeg -f mpegts - -vcodec copy output1.ts -vcodec copy output2.ts
[21:26] <burek> actually something like this: ffmpeg -i input -vcodec <coding options> -f mpegts - | ffmpeg -f mpegts -i - -vcodec copy output1.ts -vcodec copy output2.ts
[21:26] <burek> :)
[21:27] <burek> that's the final one :)
[21:27] <burek> marcosmlopes, what do you mean to "receive" this video?
[21:27] <burek> you are taking input file and re-encoding it to output file
[21:27] <burek> what's there to receive?
[21:28] <marcosmlopes> burek: i want to stream this video on internet via http
[21:28] <marcosmlopes> and use the url in html5 <video> tag
[21:28] <burek> marcosmlopes, use vlc
[21:29] <marcosmlopes> burek: i want to build my own media server because i need this app to scale on a large number of viewers
[21:29] <burek> well, ok.. then take a look at ffserver's source code
[21:29] <burek> to get the idea how to do it
[21:32] <marcosmlopes> burek: http://stackoverflow.com/questions/11334568/how-to-stream-an-webm-video-through-ffmpeg-using-node-js
[21:32] <marcosmlopes> burek: i will have a look! Thanks
[21:34] <burek> :beer: :)
[21:40] <catdude_> Anyone feel like helping me transcode a live stream? I've got a live stream coming from an IP camera (MPEG2 over UDP). I can capture the stream bytes to disk and get the characteristics of the stream (http://pastebin.com/WmMqwbTe).  I'm trying to transcode it to something I can deliver to Wowza.
[21:40] <catdude_> My most recent command line was:  ffmpeg -i udp:@:5000 -acodec aac -ac 2 -b:v 8000k -ar 48k -async 1 -bufsize 3M -s 720x480 -r 31.67 -vcodec libx264 -y output.mp4. Results are at http://pastebin.com/xbPEBBqs (basically "Error while opening encoder for output stream #0:1 - maybe incorrect parameters such as bit_rate, rate, width or height)
[21:43] <a|3x> burek, thanks, it worked though not with mpegts
[21:48] <burek> a|3x, what did you use instead of mpegts
[21:48] <burek> catdude_, why don't you feed wowza directly with that udp stream
[21:49] <catdude_> I was under the impression that wowza doesn't understand mpeg2, but it wouldn't hurt to try.
[21:50] <a|3x> burek, what i am using in the second ffmpeg instance, flv
[21:50] <burek> yeah flv is ok too
[21:51] <a|3x> burek, it worked but for some reason it's broke now, hangs, strange
[21:52] <burek> usually updating ffmpeg solves most of issues :)
[21:52] <a|3x> i got the latest version yesterday and compiled, this is something else
[22:01] <catdude_> Meanwhile, though, I'd still like to figure out what I'm doing wrong just so that I'll have that information. I'd like to use this as a learning experience.
[00:00] --- Thu Jul  5 2012