[Ffmpeg-devel-irc] ffmpeg.log.20120905

[burek]

[00:00] <njbair> I see a lot of MPEG-4/AAC videos that seem to be encoded with the goal of fitting in under 700MB. I'm assuming this is to fit on a CD-ROM, but how do the encoders nail that size so precisely?
[00:02] <JEEBsv> libx264 will fit it to a certain size if you tell it to. 2pass is recommended to get optimum output with such rate control mode
[00:02] <JEEBsv> other encoders will too, more or less
[00:02] <JEEBsv> at the moment I think libx264 is the one that will try the hardest to keep to your set overall average bit rate
[00:03] <njbair> 2pass is pretty much all related to quality of compression, right?
[00:04] <JEEBsv> yes, libx264 will otherwise just try to abide to your limitations without too much information about how demanding the stuff in the future will be
[00:04] <JEEBsv> I personally use the crf rate control mode because I have no specific size in mind, I'd just rather set the highest rate factor that usually gives me the quality I need and enjoy the compression
[00:04] <JEEBsv> single-pass too
[00:05] <njbair> I'm not particularly worried about size...but I am satisfied with the 700mb quality and i don't see any reason to be wasteful with the size.
[00:05] <JEEBsv> if you are not particularily worried about size I recommend using crf
[00:06] <JEEBsv> because it has the same possible quality of a 2pass average bit rate encode, but instead lets you set something closer to "constant quality"
[00:08] <JEEBsv> (you usually encode a few thousand frames with one and check how it looks, default is 23, and then move up/down the scale to find the highest that still looks good -- that is the highest compression you can take without losing too much information from the source)
[00:08] <JEEBsv> better than setting an arbitrary bit rate :)
[00:08] <njbair> yeah I didn't know about that at all
[00:08] <JEEBsv> -crf for crf rate control mode of libx264
[00:09] <JEEBsv> also -preset sets the preset (compression features/speed)
[00:09] <JEEBsv> http://mewiki.project357.com/wiki/X264_Settings#preset
[00:09] <njbair> is that 23 or whatever generally consistent from source to source, or will it vary for each movie?
[00:09] <njbair> I've used -preset medium before
[00:10] <JEEBsv> which is the default :)
[00:10] <njbair> actually that's normally what I use
[00:10] <JEEBsv> also crf 23 is the default, then you move up/down depending on the resolution/source. Similar stuff generally works well enough with the same crf.
[00:10] <njbair> ok thanks
[00:10] <njbair> will have to do some more reading but this is a good start
[00:11] <JEEBsv> basically with SD stuff you usually want a bit lower crf value because you'll be blowing up that picture onto a bigger screen, and with HD there's less of that so you can go a bit higher
[00:12] <JEEBsv> the usual way is to just encode a few thousand frames worth of content
[00:12] <JEEBsv> (2500 or so)
[00:12] <JEEBsv> and then check that. With Avisynth you could easily select areas so I usually picked 500 frames out of five different parts of the source
[00:13] <JEEBsv> and then you just use the highest crf that still looks good, and the preset you want and you're done
[00:14] <JEEBsv> (extra settings might be needed for plastic toy compatibility, but that's a whole separate piece of discussion)
[02:18] <rud> hmm how would you broadcast a live event for non-HLS, non-RTMP clients ?
[02:18] <rud> i mean, i have the whole chain from broadcaster > RTMPD > RTMP Clients,  the same generates HLS streams, fine.
[02:19] <rud> but some people might still have restricted port 1935 (so, no RTMP) and HLS only works on iOS..
[02:19] <rud> now, RTMPT is an interesting concept, but i haven't found any RTMPD that currently support the feature
[02:20] <rud> thus the question, how do i server my live video stream (coming from my RTMPD) to these clients & ?
[02:20] <rud> just encode a .mp4 and provide a link to it while it's being written ? (might work ?) but than, hdd space is an issue
[02:21] <rud> i don't think HTTP-Pseudo-Streaming can be of any help, since most docs i read on the topic assume you are using a static file, not a live video feed ..
[02:21] <rud> hmm
[08:55] <Kroz> what music format support 24bit/96khz/5.1 ?
[09:10] <ylluminate> running the following command: `ffmpeg -r 6 -i overlays_%02d.jpg -c:v libx264 -preset slow -crf 18 -threads 0 -s hd1080 output.mp4`
[09:10] <ylluminate> Seeing an issue with uploading a vid created from some snapshots where there's a grey overlay that appears to be some kind of corruption. it seems to happen regularly in flash player on os x safari, and flash on fullscreen on windows with chrome (but not in smaller sized players on windows)
[09:10] <ylluminate> A quick screencast of it to illustrate: http://screencast.com/t/QOv6THaD
[09:11] <ylluminate> I'm not 100% convinced it comes from ffmpeg as the left and right vertical boxes on the side are pure black
[09:11] <ylluminate> oddly though facebook uploads look just fine and so do regular playing of the resultant original vid file uploaded to youtube
[09:11] <ylluminate> any ideas on what may be going on here?
[11:05] <mdsh> Hi #ffmpeg. I've got a TIFF that contains n RGB image that actually has 16-235 image levels. How can I make ffmpeg convert RGB to YUV without also doing the 0->16  & 255 -> 235 conversion? Could I re-save as DPX, or something like that?
[11:18] <cbreak-work> mdsh: http://en.wikipedia.org/wiki/Yuv
[11:18] <cbreak-work> search on that page for studio swing
[11:20] <JEEB> mdsh, you'll probably want to check if there's a pixel format for limited range RGB
[11:20] <JEEB> and set that as the -pix_fmt
[11:21] <cbreak-work> http://en.wikipedia.org/wiki/Rec._709 too
[11:21] <mdsh> Hi cbreak-work - I'm well aware of studio levels ;-) (I've worked in BBC TV for >20 years)
[11:21] <cbreak-work> ok, then you know about 20 years more than me :)
[11:23] <mdsh> JEEB, the file is a TIFF - I don't think that setting a -pix_fmt for the input file will override the TIFF decoder setting the levels to full range
[11:24] <mdsh> I think I'll go and find an Avid or Quantel seat where I can set the input range and re-export from there. Shame - it means I have to get out of my seat ;-)
[11:25] <JEEB> mdsh, well -- it might or might not help
[11:28] <JEEB> hmm, seems like there's a color_range option
[11:29] <mdsh> JEEB, Thanks - I'll take another look at colour_range, last time I looked it was mostly ignored
[11:29] <JEEB> although knowing swscale it might fail even if you set it :D
[11:30] <mdsh> that
[11:30] <JEEB> at least it seems like the "J" YCbCr pixel formats have been deprecated
[11:31] <JEEB> so it should do "something"
[11:31] <mdsh> ;-)
[11:34] <mdsh> On a different topic - does anyone have an Avid Airspeed, and if so, does it play FFmpeg generated MPEG2 correctly?
[11:44] <mdsh> cbreak-work, you might find http://mdsh.com/RGB-YUV/ interesting ;-)
[12:24] <brogol> hello
[12:27] <brogol> someone know how to stream a file using ffmpeg libraries ?
[12:28] <brogol> in C language
[12:30] <DerLola> Hi ppl, I'm new here and have a question.
[12:30] <DerLola> I found out FFMPEG automatically trims silence of an audio track when I extract it from a video
[12:30] <DerLola> is there a way to stop this behaviour
[12:30] <DerLola> I couldnt find it in the docs
[12:37] <LaserDragon> I have what is no doubt a simple problem to do with cutting short videos with ffmpeg, I'm sorry to say I'm stumped and out of my depth. I posted on the user forum but havent recieved a response (http://ffmpeg.gusari.org/viewtopic.php?f=16&t=660). Is there anyone here willing to have a look at my forum post and see if they can help me? I'm sure it's quite simple for anyone who knows their stuff...
[12:50] <burek> LaserDragon
[12:50] <burek> you can't use -vcodec copy
[12:50] <burek> consider your video like a zip file
[12:50] <burek> you can't just cut in the middle of it
[12:50] <burek> and just copy the remaining bytes
[12:50] <LaserDragon> i see
[12:51] <burek> although
[12:51] <burek> videos are not exactly like zip files
[12:51] <burek> they are more like chunks of zip file
[12:51] <burek> s
[12:51] <burek> so to speak :)
[12:51] <LaserDragon> :)
[12:51] <burek> which is determined by keyframe positions
[12:51] <JEEB> so you can cut on a border where the other part doesn't need the other part to decode correctly
[12:51] <burek> if you cut on a keyframe, then you can safely use -vcodec copy
[12:52] <JEEB> I don't think ffmpeg even lets you cut at !keyframe :)
[12:52] <JEEB> with audio you have just audio frames in many cases, and you can cut those very well, but with video you may even have a group of pictures that's 10 seconds long
[12:52] <LaserDragon> is there a way I can reencode from the point I want but minimizing quality loss and keeping the same format?
[12:53] <burek> LaserDragon can you paste your "ffmpeg -i input.avi" output
[12:53] <burek> using pastebin.com of course
[12:53] <cbsrobot-> burek: hi
[12:53] <burek> cbsrobot- hi :)
[12:54] <cbsrobot-> can you add the !help command to fflogger so he shows all available commands ?
[12:54] <LaserDragon> burek http://pastebin.com/RWJ6wFZ1
[12:55] <burek> cbsrobot- sure
[12:55] <cbsrobot-> maybe just a one liner
[12:55] <cbsrobot-> like:
[12:55] <LaserDragon> btw, the videos in question have no audio
[12:56] <burek> LaserDragon try: ffmpeg -i full.mp4 -ss 3 -t 2 -crf 23 -preset medium output.mp4
[12:56] <cbsrobot-> available commands: !pb [nick], !google [term] &.
[12:56] <cbsrobot-> or similar
[12:56] <LaserDragon> great, thanks, i'll give that a go
[12:56] <burek> if you are not satisfied with the quality, then lower the -crf value (0 is true lossless)
[12:56] <burek> yes, and remove -acodec
[12:57] <burek> since there's no audio
[12:57] <burek> cbsrobot-, I'll see to add it now
[12:57] <cbsrobot-> cool
[12:57] <cbsrobot-> free beer from me the whole week then &.
[12:58] <burek> :)
[12:58] <DerLola> hey guys, couldnt someone have a look at my question
[12:58] <DerLola> I would really appreciate it
[12:58] <LaserDragon> burek, that does work, although I suspect a great deal of quality has been lost, the filesize was 6949kb at 7 seconds, and 270kb after cutting
[12:59] <DerLola> fflogger, great idea. will do right now
[12:59] <burek> LaserDragon, then try -crf 15
[13:00] <burek> that will keep a little bit more details and raise the quality
[13:00] <burek> also, if you just want a better compression (and to wait longer) use -preset slower
[13:00] <burek> or slow
[13:02] <LaserDragon> time's probably not an issue, we can leave this running as long as it takes
[13:02] <DerLola> fflogger please see http://pastebin.com/vTqXfLr2
[13:05] <cbsrobot-> DerLola: flogger is a bot :-)
[13:05] <cbsrobot-> but read again what he said
[13:05] <DerLola> I thought so, but it kinda does respond to what I'm saying :D
[13:06] <cbsrobot-> read the last 3 words he said &.
[13:07] <DerLola> Yeah, I got you. Here's the new link http://pastebin.com/czca7Zmq
[13:07] <DerLola> It seemed irrelevant since it does work
[13:09] <cbsrobot-> DerLola: hahaha this morning I converted a file with the same messy codecs
[13:10] <cbsrobot-> some adobe generated files i guess
[13:10] <DerLola> Yeah, unfortunately yes
[13:10] <DerLola> through FLash Media Server
[13:10] <DerLola> They're webcam recordings from a website
[13:11] <cbsrobot-> FMS sucks
[13:11] <DerLola> I totally agree :D
[13:11] <DerLola> well actually, this is done with Wowza, but that doesn't make it any better :D
[13:12] <cbsrobot-> so
[13:12] <cbsrobot-> FFmpeg version 0.6.5 is kinda from the stone age
[13:12] <LaserDragon> burek, you've solved all my problems, i think i love you :-)
[13:13] <cbsrobot-> LaserDragon: send him some beer - I guess he likes it more !
[13:13] <DerLola> okay, I think I installed it trough yum on CentOs
[13:13] <cbsrobot-> DerLola: and as a workaround you can try to convert it to a video with audio in another format
[13:13] <cbsrobot-> and then strip the audio
[13:13] <LaserDragon> if he wasnt very probably on another continent I just might :-)
[13:14] <DerLola> Sounds like a plan B indeed.
[13:14] <cbsrobot-> LaserDragon: he's from southern europe I guess
[13:14] <DerLola> So if I understand you correctly, the automatic trim of silence is not expected FFMPEG behaviour
[13:14] <cbsrobot-> so it's not so far away
[13:15] <burek> LaserDragon :beer: :)
[13:15] <cbsrobot-> DerLola: hop, but I my experiences the FMS streams tend to not have audio at the beginning
[13:15] <cbsrobot-> I didn't look into it too much
[13:15] <LaserDragon> fair enough, send me your address and I'll see what I can do :-)
[13:16] <cbsrobot-> but I converted such a file to another format and then I guess I could just strip the audio out of it and it would have the same length as the video
[13:16] <DerLola> OKay, thanks cbsrobot- I will try to encode it to a different format.
[13:16] <DerLola> It gives me a pretty big performance penalty (it's a website), but I think you might be right
[13:17] <DerLola> And I dont see any other option
[13:17] <DerLola> I was hoping for a -keep-silence param :)
[13:17] <cbsrobot-> you need live transcoding ?
[13:18] <DerLola> not really live, but asap encoding
[13:18] <DerLola> like Youtube
[13:18] <cbsrobot-> ah ok
[13:18] <cbsrobot-> and why do you need to strip the audio then (sorry for this dumb question)
[13:19] <DerLola> I strip the audio, then mix it with a different track using sox
[13:19] <DerLola> and then remix it with the video again
[13:20] <burek> cbsrobot-, try !help now
[13:20] <DerLola> brb
[13:20] <cbsrobot-> didn't see it working on ffmpeg-devel
[13:21] <cbsrobot-> ^burek
[13:21] <burek> it sends notices to you only
[13:21] <burek> not to spam the channels
[13:21] <cbsrobot-> doesn't work for me
[13:21] <cbsrobot-> but he's sometimes a bit bitchy
[13:21] <cbsrobot-> talking bout fflogger
[13:21] <burek> maybe you blocked notices
[13:22] <cbsrobot-> nope
[13:23] <burek2> works here too
[13:24] <cbsrobot-> ah - i see them now
[13:24] <burek> :)
[13:24] <cbsrobot-> showing up in growl
[13:25] <cbsrobot-> cool
[13:26] Action: cbsrobot- hands burek a cold beer and some freshly roasted french fries
[13:26] <burek> :excited:
[15:37] <DerLola> cbsrobot-, i found the culprit
[15:37] <DerLola> it's Flash specific and described in the Actionscript docs
[15:37] <DerLola> Microphone.setSilenceLevel() is designed to optimize bandwidth. When an audio stream is considered silent, no audio data is sent. Instead, a single message is sent, indicating that silence has started.
[15:38] <DerLola> so thanks for your help!
[15:38] <cbsrobot-> nice - can you change it ?
[15:42] <flori_> hey guys, i recently started working with ffmpeg. Now i am stuck and want to ask you for help: I am able to decode mp3-audio and save the decoded file. For that i am using avcodec_decode_audio4. I get the decoded AVFrames iterative in an AVFrame called decoded_frame. Now i want to use the decoded_frames to get encoded AVPakets and save them as encoded file. This way i could compare if the...
[15:42] <flori_> ...audiofile i decoded in first place is still the same after decoding&encoding. Therefore i have called ret = avcodec_encode_audio2(c, &encodedPkt, decoded_frame, &got_output);. By calling the function i get an access violation at address 0x00000000. I'm quite sure it's a very stupid mistake, but i'm pretty new to programming. Can anyone give me a hint?
[15:49] <DerLola> well, you can disable the gate. That fixes it since there is never silence. The downside - obviously - is the noise you don't suppress anymore.
[15:54] <saste> DerLola: or you could ask your users never stop talking
[15:56] <burek> or Microphone.setSilenceLevel(-1)
[15:56] <tr33fall> hi all im kind with an issue after following the ubunto guide with minor changes since im building to android
[15:56] <tr33fall> http://pastebin.com/AwqixjA0
[15:57] <tr33fall> that is my configure command
[15:57] <tr33fall> and it keeps saying that libfdk-aac not found, if i remove it complains about libopencore-amrnb
[15:57] <tr33fall> and so on
[15:57] <tr33fall> however i followed the guide
[15:58] <tr33fall> can you help me out?
[15:58] <tr33fall> please
[15:58] <relaxed> don't run configure as root
[15:58] <relaxed> ever
[15:59] <tr33fall> oh ok
[15:59] <tr33fall> sorry
[15:59] <tr33fall> is that the problem
[15:59] <tr33fall> ?
[16:00] <relaxed> no, it's just not wise
[16:00] <tr33fall> thanks for the heads up on that
[16:00] <tr33fall> kinda newbie
[16:02] <divVerent> Hi... what am I doing wrong in my code when I hit the message from this commit: https://lists.ffmpeg.org/pipermail/ffmpeg-cvslog/2012-March/048382.html
[16:02] <relaxed> you only need sudo/root when you're installing (make install)
[16:02] <divVerent> or rather: I know I am sending a pts 0 for the first audio frame
[16:02] <divVerent> but why is that wrong, and what pts should I send instead?
[16:02] <relaxed> tr33fall: look at config.log for clues
[16:03] <divVerent> is pts 0 somehow invalid and I have to offset all by like 0.1 seconds?
[16:03] <divVerent> and if yes, why?
[16:03] <tr33fall> relaxed: i will do that and report back soon. thanks
[16:07] <JEEBsv> divVerent: man I wish I remembered how PTS works in ffmpeg
[16:07] <JEEBsv> and/or knew
[16:09] <tr33fall> the error is because "error: fdk-aac/aacenc_lib.h: No such file or directory"
[16:09] <tr33fall> but inside a directory inside fdk-aac/ i found that .h
[16:11] <relaxed> tr33fall: is the correct path to it ../fdk-aac ?
[16:12] <tr33fall> afraid not
[16:12] <tr33fall> the correct path to that specific .h is ../fdk-aac/libAACenc/
[16:12] <relaxed> That's what your ./configure states.
[16:13] <relaxed> adjust --extra-cflags="-I../fdk-aac"
[16:13] <tr33fall> ok thanks
[16:14] <tr33fall> ill try that
[16:14] <DerLola> @saste, that would be... hard :)
[16:14] <DerLola> and burek, that's what I ended up doing
[16:24] <brogol> does someone know how to stream a file using libav* over RTP in C language  ?
[16:25] <divVerent> 16:07:38    JEEBsv | divVerent: man I wish I remembered how PTS works in ffmpeg
[16:25] <divVerent> JEEBsv: what I do know, is that this code tries to detect codecs that don't set pts
[16:25] <divVerent> and then replaces their pts by a bad guess
[16:25] <divVerent> in my specific case, the frame with pts 0 is the very first one
[16:25] <divVerent> so this replaces 0 by 0 and thus does no harm
[16:25] <divVerent> other than that it prints the warning
[16:26] <divVerent> my question is, why is pts 0 deemed invalid, and how am I supposed to fix it
[16:26] <divVerent> like, is the pts for audio maybe not the START of the frame, but the END, and thus the solution is to use the time of the last sample, and not the first?
[16:28] <JEEBsv> not sure, and I'm not sure why 0 is invalid
[16:28] <JEEBsv> possibly used as an invalid (or loldunno) value before
[16:30] <tr33fall_> relaxed: still giving error --> http://pastebin.com/NFSb56dM
[16:31] <relaxed> did you use --extra-cflags="-I../fdk-aac/libAACenc" ?
[16:31] <tr33fall_> --extra-cflags="-I../fdk-aac/libAACenc/include" and still gives error
[16:32] <tr33fall_> so i assume the error was not from there
[16:32] <relaxed> it's -I/path/to/headers  are the headers where you're pointing?
[16:33] <tr33fall_>  path /fdk-aac/libAACenc/include  where the headers are
[16:35] <JEEBsv> divVerent: you could try asking on the -devel side just to attract knowledge'able people, if this is not documented anywhere
[16:35] <tr33fall_> i added -I/home/esteves/fdk-aac/libAACenc/include
[16:35] <tr33fall_> and still /tmp/ffconf.ubqNpH8Y.c:1:32: error: fdk-aac/aacenc_lib.h: No such file or directory
[16:35] <relaxed> copy them to fdk-aac/
[16:36] <JEEBsv> did you compile and install fdk-aac to some prefix?
[16:36] <JEEBsv> or are you just trying to use the headers out of the source folder?
[16:36] <JEEBsv> because many projects don't have the same folder structure in their source distro
[16:37] <JEEBsv> you can always add a --prefix to the configure of the library to set a specific prefix to install to (can be in your home folder)
[16:37] <tr33fall_> i followed the ubunto guide
[16:37] <tr33fall_> autoreconf -fiv \ ./configure --disable-shared
[16:37] <JEEBsv> then you should just be able to do --extra-cflags="-I/your/prefix/include" and --extra-ldflags="-L/your/prefix/lib"
[16:37] <JEEBsv> on ffmpeg's side
[16:38] <JEEBsv> tr33fall_: so you didn't install it but just compiled after configuring?
[16:38] <JEEBsv> you should re-configure with a --prefix=/your/prefix and make -> make install
[16:39] <JEEBsv> I usually used to just do --prefix=/home/username/ownapps
[16:39] <tr33fall_> i did a make but not make install
[16:39] <tr33fall_> after the make i did checkinstall
[16:39] <JEEBsv> yeah, so you're just trying to use the headers from the source folder
[16:39] <JEEBsv> checkinstall I don't know
[16:40] <tr33fall_> sudo checkinstall --pkgname=fdk-aac --pkgversion="$(date +%Y%m%d%H%M)-git" --backup=no \   --deldoc=yes --fstrans=no --default
[16:40] <JEEBsv> did that actually install that package?
[16:40] <JEEBsv> or just made it?
[16:40] <tr33fall_> so your are saying that i put that in a prefix and then add the prefix/lib and prefix/include on the extras?
[16:41] <JEEBsv> yes, but I'm trying to first see what exactly you did
[16:41] <tr33fall_> ah ok
[16:41] <JEEBsv> check your /usr/lib and /usr/opt/lib for related stuff if it got installed
[16:41] <tr33fall_> ill explain then
[16:41] <JEEBsv> nah, I'm trying to understandf what exactly checkinstall does
[16:41] <tr33fall_> i found the lib on /usr/local/lib
[16:41] <JEEBsv> ok
[16:42] <JEEBsv> so you can just use that as the -L and -I is /usr/local/include
[16:42] <relaxed> dpkg -L fdk-aac
[16:42] <tr33fall_> damm that shows where is the include and lib
[16:43] <tr33fall_> thanks a lot
[16:43] <relaxed> you're welcome
[16:43] <tr33fall_> never remember to check for that
[16:45] <JEEBsv> I started with university machines where I had no root so I just got used to using --prefix and installing to my home
[16:45] <relaxed> JEEBsv: I still do that, it's cleaner and I can have multiple version installed at one time.
[16:46] <tr33fall_> i thought it will work but still gives error
[16:46] <tr33fall_> -L/user/local/lib/
[16:46] <rud> hmm i asked that during insomnia a couple hours ago, so i had no answers, thus me trying again now& theoretically speaking, if you have a video&audio stream over RTMP, would you know any way/method/codecs allowing me to generate an mp4 file i could stream using http-pseudo-streaming (using nginx here) ..? my problem is that RTMPT is not an option (I do not know any RTMPD that actually does RTMP-Transversal), and I do need to server that feed to people who's po
[16:46] <rud> 1935 is blocked &
[16:46] <tr33fall_> -I/usr/local/include/fdk-aac/
[16:46] <JEEBsv> leave the fdk-aac out
[16:46] <relaxed> tr33fall_: omit fdk-aac
[16:47] <JEEBsv> just add your prefix's include and lib folders to -I and -L
[16:47] <rud> i thought about running my RTMPD on port 80, but I guess (hope?) most intelligent content filters would figure this is no HTTP traffic and thus block it.
[16:47] <JEEBsv> for some things pkg-config is used so you can do PKG_CONFIG_PATH=/my/prefix/lib/pkgconfig ./configure --my-options
[16:48] <tr33fall_> ok
[16:48] <JEEBsv> depends on the lib etc. naturally, but it works for some stuff :)
[16:48] <JEEBsv> the extra cflag and the ldflag are the sure way
[16:51] <tr33fall_> yeah that solve the library issue
[16:51] <tr33fall_> but now i have -> cannot find -lfdk-aac
[16:52] <tr33fall_> isnt that the library
[16:52] <JEEBsv> yes
[16:52] <JEEBsv> --extra-ldflags="-L/your/prefix/lib"
[16:52] <JEEBsv> -I is a cflag, and -L is an ldflag
[16:53] <tr33fall_> in the ffmpeg i have this error -> "cannot find -lfdk-aac collect2: ld returned 1 exit status ERROR: libfdk_aac not found"
[16:53] <tr33fall_> now i confused
[16:53] <tr33fall_> lol
[16:53] <JEEBsv> I just told you to add the ldflag of -L to your prefix's lib
[16:54] <JEEBsv> you have made it find the header, but you haven't told it where the linker finds the library
[16:55] <tr33fall_> i already had --extra-ldflags="-L/user/local/lib"
[16:55] <JEEBsv> hmm
[16:56] <JEEBsv> post full config.log
[16:56] <JEEBsv> on pastebin
[16:56] <JEEBsv> naturally
[16:56] <tr33fall_> ok
[16:58] <tr33fall_> there it is -> http://pastebin.com/AiX4qHwi
[16:59] <JEEBsv> uhhhhhh
[16:59] <JEEBsv> are you sure your fdk-aac is compiled for arm as well?
[17:00] <JEEBsv> it now sounds like you're trying to link a x86 fdk-aac to arm ffmpeg
[17:00] <tr33fall_> hum
[17:00] <tr33fall_> that is indeed what im trying
[17:00] <tr33fall_> but seems i have to compile to arm then
[17:01] <tr33fall_> didnt know that
[17:01] <tr33fall_> so any library i want to add to ffmpeg for android also needs to be compiled to arm?
[17:01] <JEEBsv> yes
[17:02] <tr33fall_> ok
[17:02] <JEEBsv> also, man those settings
[17:02] <JEEBsv> yasm with arm?
[17:02] <JEEBsv> that all reads more like "What on earth were you smoking?"
[17:02] <tr33fall_> :P
[17:02] <tr33fall_> thats why im newbie
[17:03] <tr33fall_> ill go then and try to compile and understand what i can and cannot compile
[17:03] <tr33fall_> thanks for all help
[17:04] <JEEBsv> first of all you take like 99% of that out, and leave the arch, cc, cross-prefix there
[17:04] <JEEBsv> then you compile a simple ffmpeg
[17:04] <JEEBsv> then you start adding/removing things
[17:12] <tr33fall_> ok thanks
[17:13] <JEEBsv> no install needed naturally when you're testing
[17:13] <JEEBsv> just have it compile
[17:24] <tr33fall_> im now compiling it to arm
[17:26] <rschmidt> hi guys... I'm trying to optimize transcoding performance in a live streaming setting. The input I'm getting is way too big (1920x1080 at 60fps), and I need to drop it down to 720x480 @ 30fps (maximum). Is using a filter to scale the input and drop half the frames a good idea, performance-wise?
[17:33] <relaxed> rschmidt: what other option do you have? try -filter:v scale=720:-1,fps=30
[17:34] <relaxed> scratch that, use -filter:v fps=30,scale=720:-1
[17:34] <rschmidt> relaxed: currently I have almost no options... I'm passing the feed to ffserver, so my command line is very basic... just -async 2 -vsync 1. The rest (output size, framerate, etc) is specified in ffserver.conf
[17:35] <rschmidt> it works, but it uses 99% of one of my CPU cores constantly... so I'm trying to figure out a way to lessen the load a bit
[17:35] <rschmidt> relaxed: ok, I'll give that a shot, thanks
[17:35] <rschmidt> relaxed: I'm also looking at framestep:2... my input source gives me too many frames
[19:01] <FunnyLookinHat> Anyone here used the Akamai CDN w/ their OSMF plugin to build a video player?  I know it's a strange place to ask, but figured I might find someone.
[20:00] <saliak> Is there an interface to get the status of an encoding job that's running in the background by sending a signal to the ffmpeg process?
[20:16] <divVerent> saliak: no
[20:17] <divVerent> just grepped the source
[20:17] <saliak> divVerent: that sucks.  so no way I could have a process encoding videos in the background, and another process that can display the status and progress of encoding a queue?
[20:18] <divVerent> you can, by using a tool to turn the stdout/stderr of ffmpeg into something you can query
[20:18] <divVerent> http://cr.yp.to/daemontools/readproctitle.html
[20:18] <divVerent> this is one of the tools that can do it
[20:18] <divVerent> there's more of these
[20:18] <divVerent> I just remember that one, and don't actually endorse djb's stuff ;)
[20:21] <saliak> ok
[20:24] <divVerent> saliak: another idea is to just run ffmpeg in tmux or screen
[20:24] <divVerent> but that one you can't just programmatically query
[20:24] <divVerent> or the really stupid one: ffmpeg ... >logfile 2>&1 &
[20:24] <divVerent> and then
[20:24] <divVerent> tail logfile
[20:24] <divVerent> to see the status
[20:25] <saliak> yeah.  i'm trying tosetup some web scripts to let me encode and view my digital movie library vai web interface
[20:25] <saliak> can get a php script to kick off ffmpeg, but would be nice to have a status display of an encoding job
[20:25] <divVerent> hm... for web stuff, I'd probably recommend the "tail" method
[20:25] <divVerent> just make sure you ignore the last line if incomplete
[20:29] <saliak> ok
[20:30] <JodaZ> is there some option to just have ffmpeg place a short stco atom at the start of mp4 so one could use it for live transcoding ?
[21:21] <burek> JodaZ, did you try -flags +global_header
[21:21] <burek> or qt-faststart
[21:22] <JEEB> JodaZ, for live stuff and mp4 you need to use the fragments feature
[21:22] <JEEB> otherwise it's not possible to start decoding an mp4 file that hasn't finished yet
[21:22] <JEEB> ffmpeg should have support for it, but I'm not sure what the option is \o/
[00:00] --- Thu Sep  6 2012