[Ffmpeg-devel-irc] ffmpeg.log.20120713

burek burek021 at gmail.com
Sat Jul 14 02:05:02 CEST 2012


[01:09] <JPeterson> is there a ffdshow channel?
[01:09] <JPeterson> wheres the best channel to ask a ffdshow question?
[01:25] <killown> somebody here could help me with recording internal sound system?
[01:25] <killown> I had tried pavucontrol and no success
[01:37] <llogan> burek knows more about that than i
[01:37] <llogan> i uploaded a fdk-aac to Arch Linux AUR if anyone wants to try it with ffmpeg
[01:37] <llogan> i'll update ffmpeg-git too
[01:50] <newl>  /17
[02:00] <davidOmatic> hi there.
[02:00] <davidOmatic> is it possible to encode dvcpro100 with fmpeg?
[02:01] <davidOmatic> ffmpeg**
[02:04] <llogan> davidOmatic: i'm not sure but see http://ffmpeg.org/pipermail/ffmpeg-user/2011-October/002758.html
[02:04] <davidOmatic> tnx llogan
[02:04] Action: davidOmatic reading...
[02:44] <mtyson> Hey Guys, I'm getting "warning: first frame is no keyframe" here's the info: http://pastebin.com/aiAks46f
[02:45] <llogan> mtyson: are you still getting a useful output?
[02:45] <mtyson> llogan: such a ffmpeg newbie, I'm not sure
[02:45] <mtyson> I put my output in that pastebin
[02:47] <mtyson> it does create the file, but its empty
[02:48] <llogan> what do you mean by empty?
[02:50] <newl> please don't change the meaning of the word empty .. life is tough enough
[02:50] <mtyson> hah... empty in the conventional sense: its a 0k file
[02:50] <newl> whew
[02:51] <mtyson> existentially empty as well, like all phenomena :)
[02:52] <mtyson> Does this look right? ffmpeg -i "M2U00001.MPG" -f mjpeg -t 0.001 -ss 3 -y "newTest.jpg"
[02:55] <llogan> you probably want: ffmpeg -i "M2U00001.MPG" -vframes 1 -ss 3 -y "newTest.jpg"
[02:58] <mtyson> llogan: Thanks, gave it a shot... still have that error
[02:59] <llogan> but is the output ok?
[02:59] <mtyson> Ah, but the jpg is there!
[02:59] <mtyson> Very cool
[02:59] <mtyson> llogan: Was the frame time wrong?  1 instead of 0.001?
[03:00] <mtyson> Anyway, that is much appreciated
[03:01] <llogan> you wanted a single frame output, not a timespan of frames...if that makes sense
[03:01] <mtyson> ok, yeah, makes sense
[03:12] <mtyson> thanks again
[06:10] <md_5> I have a maktroska stream which ffmpeg handles just fine when fully done, but how might I make ffmpeg keep reading the stream to see if more data is being added
[06:15] <md_5> seems -re is what I wanted
[09:17] <killown> somebody here can help me providing me an example of how record internal audio from sound system OS?
[09:23] <iive> killown: really don't know how to help you.
[09:24] <iive> old stand alone card had a loopback recording/capturing and it was elementary
[09:24] <iive> ac97 seem to lack it.
[09:25] <iive> probably as anti-piracy measure. Intel is known to have such "technologies".
[09:25] <iive> so I guess you may try to look up if alsa provides something to replace it.
[09:32] <rainmaker1> I want to save multicast stream to a file and I am using this command
[09:32] <rainmaker1> /usr/local/bin/ffmpeg -y -i udp://230.0.0.3:5000 -c:v copy -c:a copy -map 0 -f segment -segment_time 60 -segment_format mpegts -segment_list test.m3u8 stream%05d.ts
[09:32] <rainmaker1> now, when I play with vlc it plays fine
[09:32] <rainmaker1> but when I try to stream that file via rtmp I get audio/video out of sync
[09:54] <iive> killown: http://ubuntuforums.org/archive/index.php/t-914405.html this looks like a good start.
[09:54] <killown> iive, thank you
[10:55] <smokeyrat> I am doing "-b:v 48k" to convert some MP3 files (sound effects) of varying bitrates to 48,000 bits per second, as Adobe Premiere Pro CS6 wants for the video footage I am editing. This appears to be working, but when I look in Winamp or in the properties in Premiere, it turns out that the resulting files have different bitrate than what I specified. What's going on?
[11:12] <smokeyrat> Ahem.
[11:43] <rainmaker1> Anyway I can insert creation_time in mpegts? I use ffmpeg -y -i udp://230.0.0.3:5000 -metadata creation_time="1:11:11" -c:v copy -c:a copy -map 0 -f segment -segment_time 3600 -segment_format mpegts stream0000%d.ts
[11:43] <rainmaker1> but can find 1:11:11 in output file
[11:45] <burek> smokeyrat, it's not -b:v but -b:a
[11:46] <smokeyrat> :S
[11:46] <smokeyrat> Then what is -b:v?
[11:46] <burek> bitrate for video
[11:47] <smokeyrat> But...
[11:47] <smokeyrat> Oh...
[11:47] <smokeyrat> Aaaaaaah! Works!
[11:47] <smokeyrat> Thanks.
[11:47] <burek> :)
[11:49] <smokeyrat> Not sure what the hell it did when I fed it -b:a on a music file.
[11:49] <smokeyrat> *audio
[11:50] <smokeyrat> One might consider this something that Premiere should support inside itself.
[11:52] <burek> well, can you use pastebin to show your command and the output
[11:52] <burek> so we might see what went wrong
[11:59] <smokeyrat> I just meant when it didn't work. When I used -b:v.
[11:59] <smokeyrat> I wonder what it actually did to the "video" of an audio-only file.
[11:59] <smokeyrat> Just a general curiosity thing.
[12:21] <burek> smokeyrat, it did nothing
[12:21] <burek> if your input was mp3
[12:21] <burek> -b or -b:v gets ignored
[12:21] <burek> since no video stream is inside
[12:21] <burek> so, your audio bitrate gets some default values
[12:22] <burek> depending on the output format and encoder used
[12:23] <smokeyrat> ffmpeg is weird and scary. And powerful.
[12:47] <smokeyrat> Just when I thought I had "got it", I realized that I'm confused like hell.
[12:47] <smokeyrat> "48 kbps, 16 frames
[12:47] <smokeyrat> 11025 Hz Mono"
[12:47] <smokeyrat> What does this mean? Why is "kbps" different from Hz, and what are 16 frames?
[14:37] <mindfire6482> hello all, if i may request some assistance with ffmpeg commands and error codes, ive only just started using ffmpeg and for the most part im pretty impressed but ive come across a hurdle.
[14:43] <mindfire6482> i am converting to WMV format with presets to make the resulting file compatible with xbox 360, nowo i am able to do this with the code on http://pastebin.com/nSzf18bx
[14:46] <mindfire6482> now this is fine if i want to leave the conversion running over night for multiple files but im looking at a quicker solution really. i read the just changing the container might work so i did a vcodec copy acodec copy command to change to WMV, this resulted in a file that works on the computer but not on xbox, i have tried vcodec wmv2 acodec wmav2 without the trailing commands as per the pastebin and ffmpeg throw
[14:46] <mindfire6482> s out an error
[14:51] <dougle> When running an mp4 through ffmpeg i get a video stream of 4.2 seconds and an audio stream of 4.272 seconds, concatenating a few of these together with MP4Box produces progressively offset audio, is there a way to chop the audio off to the length of the vid stream or vice versa?
[14:51] <dougle> the shortest switch doesn't seem to make any difference
[14:52] <dougle> the bit rates of each stream are set high to try to get them the same length but .072 seconds is the best i could manage with this method
[14:53] <dougle> ffmpeg  -y -i invid.mp4 -threads 4 -r 25 -shortest -vcodec libx264 -b:v 2m -acodec aac -ac 2 -b:a 512k -strict experimental outvid.mp4
[14:55] <mindfire6482> sorry lost my connection for some reason
[14:55] <mindfire6482> if anyone posted a reply about my issue can they please repost
[14:56] <JEEB> I'm afraid I don't think the wma/wmv encoders will be any faster
[14:56] <JEEB> there could be some settings, but they most probably are quite obscure
[14:57] <JEEB> also, didn't the X360 support H.264 just fine? I only know that you IIRC can't do >2ch audio with non-wmv, but you're setting audio channels to two anyways
[15:00] <mindfire6482> yeah JEEB ive had to set it to 2ch audio because if i dont set that ffmpeg throws out and error about audio encoding, it been the only way i can get it to convert anything so far
[15:01] <undercash> hello
[15:02] <undercash> i have a git ffmpeg compiled in january, any major changes since then?
[15:02] <JEEB> if you really need WMV I kind of encourage you to use one of MS's own things for that, although I must say I have no idea how easy or hard that would be to do (the MS Expression Encoder series seems to be their newest)
[15:02] <JEEB> undercash, check the changelog file in current trunk
[15:02] <JEEB> although I bet there's been plenty of changes
[15:02] <undercash> or better, wil i need to modify my ffmpeg command paramaters?
[15:03] <JEEB> most probably not
[15:03] <undercash> last time i upgraded, i had to change -vpre to -preset etc etc
[15:03] <undercash> ok
[15:03] <JEEB> mindfire6482, although in your case I'd probably just use libx264 and encode H.264 in mp4 for the X360, as I think it supports a certain subset of H.264
[15:06] <mindfire6482> JEEB, its wierd, i have some mp4 files which i have tried to play on the 360 but they arent supported, it seems to me, (although ive only been using ffmpeg for about 3 days now) that the preset i have for the 360 conversion works perfectly its just slow as hell, i thought maybejust changing the container for the video would be much faster which it was but the file was unusable on the console
[15:07] <JEEB> well, unsurprisingly :P
[15:07] <JEEB> because you can in theory stuff whatever you want into asf (wmv)
[15:07] <JEEB> whether or not something actually supports <random stuff> in asf (wmv) is a whole separate thing
[15:07] <mindfire6482> yeah just read that back... *kicks ones own butt*
[15:08] <JEEB> also, it might just be that those mp4 files are just encoded in a way or contain streams that the X360 doesn't like
[15:08] <mindfire6482> there has to be a quicker way i can covert though surely... maybe... possibly? im kiding myself arent i?!
[15:08] <mindfire6482> BTW im on a linux machine so MS not really an option
[15:09] <burek> just read the link
[15:09] <JEEB> burek, uhh that one just tells a person to remux the video and re-encode audio to AAC
[15:09] <JEEB> that doesn't _always_ work
[15:09] <burek> says you, without even asking him did he try
[15:10] <JEEB> well, I just know of o9k files that are encoded every day over the limitations of X360, so unfortunately it's not just me. Of course, remuxing and only re-encoding audio would be the perfect way to do it if the H.264 stream was compatible
[15:10] <JEEB> so yes, it's worth a try, sure
[15:11] <JEEB> I'm just saying that it's not a perfect solution and hitting any file with that just won't work
[15:12] <dougle> @burek  http://gist.github.com/3104814
[15:12] <JEEB> mindfire6482, btw can you show either ffmpeg -i derp.input or ffproble derp.input's output in a pastebin of that input file?
[15:13] <mindfire6482> Hang on a sec JEEB
[15:14] <burek> dougle, -shortest works for 2 (or more) inputs, not for a single input
[15:14] <burek> if you need to sync your a/v streams inside 1 input file, use -itsoffset
[15:15] <dougle> ah right of course, not between streams
[15:15] <burek> also, don't use constant bit rate with -b:v 2m when using libx264
[15:15] <burek> it's far better to use vbr with -crf and -preset
[15:16] <dougle> ok will look into vbr settings
[15:16] <burek> smaller file / better image quality
[15:16] <JEEB> note: -b:v is most probably not CBR, you need vbv for real CBR
[15:16] <JEEB> if it sets the same thing as --bitrate with x264
[15:17] <burek> ok, it's not cbr, it's a targeted bit rate, floating around the middle value
[15:17] <dougle> with the offset it's not a constant delay, but due to each vid having slightly longer audio, they shunt each other along a bit
[15:17] <JEEB> (with two passes average bitrate is more or less the same thing as crf of that same output bitrate, with just one pass it's worse off than crf naturally)
[15:17] <JEEB> (crf is one-pass only)
[15:18] <dougle> enough to put lip sync off noticeably
[15:18] <burek> but it's still better to use -crf and -preset to fine tune it
[15:18] <burek> rather than just hint the bitrate
[15:18] <JEEB> depends on what you need
[15:18] <JEEB> if you need a set file size for a specific length of a clip
[15:18] <JEEB> then 2pass + bitrate
[15:19] <burek> well, think a little bit, do you get better control using 2 things or using 1 thing for control
[15:19] <JEEB> uhh, the preset does the same thing with whichever
[15:19] <JEEB> if you just want good quality and have no file size restraints then crf naturally
[15:19] <burek> dougle, what exactly is the issue?
[15:19] <burek> JEEB, ok
[15:20] <JEEB> crf and abr are basically two things using the same algorithm while setting it for two different things
[15:20] <JEEB> for two different use cases, of course
[15:21] <burek> dougle, you want exactly the same length of audio and video, right?
[15:21] <dougle> yes
[15:21] <dougle> in that gist, the bottom output (from MP4Box) shows video 4 seconds and audio track of 4.063  when combining loads of mp4s together that difference puts the audio out of sync
[15:22] <burek> can you type ffmpeg -i invid.mp4
[15:22] <burek> and use pastebin.com to show the output
[15:22] <dougle> am wondering if there is anyway to control that in ffmpeg (rather than the concat stage)
[15:22] <dougle> ffmpeg -i is in that gist too
[15:22] <dougle> oh sorry infile, will update
[15:23] <burek> ah yes, sorry
[15:23] <burek> i was ignorant
[15:23] <burek> well
[15:23] <burek> first of all you are not using ffmpeg
[15:23] <yetifoot> Does anyone know if the image2pipe output has no filters like deblocking etc
[15:23] <burek> you are using something that's named avconv
[15:24] <burek> "Copyright (c) 2000-2012 the Libav developers"
[15:24] <burek> which was based on old version of ffmpeg
[15:24] <burek> so, it would be wise to uninstall that thing
[15:24] <burek> and compile ffmpeg from source
[15:24] <dougle> http://gist.github.com/3104814
[15:24] <dougle> ok
[15:25] <dougle> is there much difference between the two?
[15:26] <dougle> i'm on debian and it was from the repos
[15:26] <dougle> have been using ffmpeg manual too :s
[15:26] <yetifoot> are you using the version from debian-multimedia?
[15:26] <dougle> errrrrrr
[15:26] <dougle> no
[15:27] <dougle> don't have that repo
[15:28] <yetifoot> i think they are usually more up to date, and have more codecs enabled than the main repos
[15:29] <dougle> another pain in the arse is that this will end up on centos
[15:33] <burek> dougle, you really wanna fix that
[15:33] <burek> libav is not ffmpeg
[15:33] <burek> and if you decide to stick with it, you need to accept some pain :)
[15:33] <burek> you can try static builds though
[15:34] <burek> http://ffmpeg.org/download.html
[15:34] <dougle> i'm just reading that ffmpeg will be remove on future debian releases?
[15:34] <dougle> saying it's deprecated
[15:35] <yetifoot> i'm reading about this too, never knew this before
[15:35] <yetifoot> so debian are going with the libav version
[15:36] <JEEB> yes, the libav project rewrote parts of the ffmpeg command line app under the name avconv
[15:36] <JEEB> and then after that release that you have, removed the ffmpeg command line tool from their side
[15:36] <dougle> do you know the reason for that? falling out with ffmpeg devs or..?
[15:36] <JEEB> it's a long story but nowadays people just do development on both sides
[15:36] <dougle> k
[15:37] <JEEB> ffmpeg's ffmpeg command line app has most of the stuff that was rewritten as avconv
[15:37] <JEEB> as merging happens often
[15:37] <dougle> hmmmm
[15:39] <JEEB> but yeah, the debian/ubuntu maintainer is more active on the libav side so it was switched there
[15:39] <JEEB> both projects are still very much active
[15:39] <dougle> so just personal preference then
[15:41] <dougle> "FFmpeg Static Builds by Burek"   cheers
[15:41] <dougle> (and thanks relaxed)
[15:43] <JEEB> also if you're trying to merge AAC streams in mp4 files with mp4box, it could be a problem with mp4box not reading the encoder delays right
[15:43] <JEEB> but that's just one of many possible things
[15:44] <JEEB> of course it could also be that the older version of libav (and an older version ffmpeg from similar time frame) wouldn't be able to write the encoder delay correctly
[15:44] <JEEB> all I know is that GPAC (of which mp4box is a part) has always been a clusterderp
[15:44] <dougle> yes been trying to force mp4box to comply for a while, decided to attack from other end and see what luck i ge
[15:44] <dougle> t
[15:45] <dougle> i haven't been able to find an alternative for mp4 container though, which surprised me
[15:45] <yetifoot> hmm, i've been using MP4Box and it seemed ok, but what would you recommend using instead?
[15:46] <JEEB> L-SMASH is easier to to build and is generally saner (GPAC tries to be everything from MPEG-4 or so), but of course it's still relatively young so it lacks features like appending files
[15:46] <dougle> (an alternative for mp4box for the mp4 container)
[15:47] <JEEB> the boxdumper app from L-SMASH f.ex. is a rather nice troubleshooting tool when you need to take a closer look at the insides of a mp4/3gp/mov file
[15:47] <yetifoot> ok i will check it out thanks
[15:47] <JEEB> http://code.google.com/p/l-smash/
[15:47] <JEEB> and yeah, muxer and remuxer are still separate apps in L-SMASH
[15:48] <JEEB> (also if you don't need appending and such, ffmpeg/avconv's muxing capabilities should be fine in most cases as well
[15:53] <JEEB> anyways, personally if I wanted to be completely sure I'd just encode the thing in one pass with current ffmpeg/libav
[15:53] <dougle> i'm considering converting to mpg and just cat-int the data
[15:54] <burek> or extract everything as raw video/audio and use the simplest video/audio editor to concat :)
[15:54] <burek> after that save again as raw/lossless and use ffmpeg to finally encode to what you need
[15:54] <dougle> has to be command line
[15:54] <dougle> it's being automated
[15:54] <mindfire6482> well... im reconverting using the details on that webpage, it seems to be going a bit faster which is a plus, lets just hope the files are xb friendly
[15:54] <dougle> yes i think it might be ffmpeg for that last step to mp4
[15:55] <burek> I think concat: protocol might work well
[15:55] <burek> just don't expect -c:v copy to be successful
[16:02] <dougle> JEEB in gpac is there no way to sync on a stream other than #1
[16:02] <dougle> ?
[16:02] <JEEB> no idea
[16:07] <undercash> "flv does not support that sample rate, choose from (44100, 22050, 11025)"
[16:08] <undercash> i have -ar 44100
[16:08] <undercash> so why this error
[16:13] <LexSfX> is it possible that this is a matter of incorrect ordering of the ffmpeg command line?  most of the problems i've seen with the ffmpeg cli program are caused by this, since the ffmpeg cli program uses command line arguments in a particular order.
[16:17] <undercash> hm
[16:18] <undercash> ffmpeg -re -i 'Thats Poker.avi' -vcodec libx264 -preset fast -crf 27 -acodec libfaac -ab 128k -ar 22050 -ac 2 -s hd480 -aspect 16:9 -f flv rtmp://live.justin.txxxxxx
[16:18] <undercash> seems normal to me
[16:23] <undercash> actually when i select only one output, it works, when i put 2 rtmp output , i m getting that error
[16:32] <undercash> really it s weird, when i remove one of the rtmp output it works fine for each of them , when i combine the 2 output get this samplerate error
[16:34] <undercash> if i replace one of the rtmp output to output to file , no problem either
[16:35] <undercash> ok
[16:40] <undercash> http://pastebin.com/biNG02fq
[16:40] <undercash> and this is the command http://pastebin.com/hsTVUNYi
[16:41] <burek> well that's normal what you're experiencing
[16:41] <burek> since you didn't specify the 2nd output encoding options
[16:41] <burek> so ffmpeg used defaults
[16:41] <undercash> i dont want 2 encoding, i want one encoding 2 output
[16:41] <undercash> not to load too much the server
[16:42] <burek> The generic syntax is: ffmpeg [global options] [[infile options][‘-i’ infile1]] [[infile options][‘-i’ infile2]] [[infile options][‘-i’ infile3]]... {[outfile options] outfile1}... {[outfile options] outfile2}... {[outfile options] outfile3}...
[16:42] <burek> I'm not sure ffmpeg can do that
[16:42] <burek> you might try to do the encoding and pipe it to another instance of ffmpeg
[16:42] <undercash> hmm
[16:42] <burek> which will produce 2 copy of the input
[16:42] <undercash> that would explain why on ustream it was in a weird format , like 4/3 maybe ffmpeg default
[16:43] <burek> ffmpeg ... (encoding) ... | ffmpeg -i - -c copy out1 -c copy out2
[16:43] <undercash> was working though but a lot worst quality than on justin
[16:43] <undercash> i think you told me once about that -c copy command, but i dont know how to use it
[16:45] <burek> undercash, only your 1st output was encoded the way you wanted
[16:45] <burek> the 2nd one used default options (since you provided none)
[16:45] <undercash> exactly, and it worked coz the file was simple
[16:45] <undercash> the second file, with 48 samplerate audio wouldnt work
[16:46] <burek> of course
[16:46] <burek> that's expected behavior
[16:46] <undercash> i got it now
[16:46] <burek> try using pipes
[16:46] <undercash> i m sure the output to file, also didn't encode using my encoding options
[16:46] <burek> 1st process should just encode and the 2nd should just copy the input to multiple outputs
[16:47] <undercash> yes i think i understand what you mean, but in concrete, i dont know how to write the command
[16:52] <undercash> actually the output to file is perfect
[16:54] <undercash> but  yea as you said, it wasn't encoded though http://pastebin.com/Ha1uesCZ
[16:58] <undercash> thx anyway for your help :)
[16:59] <burek> :beer: :)
[17:27] <dougle> in the end i converted my source mp4s to mpeg2 versions, then cat and pip them to ffmpeg saving to final mp4
[17:27] <dougle> whole process has tripled in duration, but audio is spot on
[19:43] <pyoor> Hi all.  I was wondering if someone here might be able to help me with a quicktime, mp4 question.  I'm seeing a number of files containing and atom structure similar to: <size:1500,type:mdat><wide><size:0000,type:mdat>
[19:44] <pyoor> The size for the first mdat is bogus as it ends mid stream.  and The second mdat extends to the end of the file
[19:44] <pyoor> I'm just trying to figure out where in the spec this behavior is defined.
[20:05] <burek> pyoor, and what does your question have to do with ffmpeg in particular?
[20:07] <pyoor> burek: sorry.  any idea where would this be defined in the ffmpeg source?
[20:07] <pyoor> maybe better suited for #ffmpeg-devel
[20:08] <burek> to be honest, I have no idea what did you just ask :)
[20:09] <pyoor> heh
[20:10] <pyoor> I think I may have actually answered my own question.  It looks like it's listed here: /ffmpeg/libavformat/movenc.
[20:10] <pyoor> */ffmpeg/libavformat/movenc.c
[20:39] <lnb> Hi, when i run ffmpeg under winff, the default of 320x180 works fine. but if i change it to 640x360 it keeps failing with bitrate too small
[20:40] <lnb> can someone tell me what the following line should be changed to:
[20:40] <lnb> -vcodec flv -f flv -r 29.97 -vf scale=640:360 -aspect 16:9 -b:v 300k -g 160 -cmp dct -subcmp dct -mbd 2 -flags +aic+cbp+mv0+mv4 -trellis 1 -ac 1 -ar 22050 -b:a 56k
[20:46] <iive> lnb: well.. i see a lot of nice options in there, so i guess the bitrate is too small for that resolution. have in mind, 640x360 is 4 times bigger than 320x180.
[20:46] <lnb> iive: hi, so what must i change?
[20:47] <iive> there is a trick that might give you 30% boost in compressibility, but you need 300% :)
[20:47] <iive> lnb: i would say, increase the video bitrate.
[20:47] <lnb> what part is that in the line line?
[20:48] <sacarasc> -b:v
[20:48] <sacarasc> Also, the -r should be -r 30000/1001, not 29.97.
[20:48] <sacarasc> But, if you're playing it on a computer, it could really be whatever you wanted.
[20:49] <lnb> its movies i make on trips and upload them to our server for family to watch
[20:49] <lnb> hmm -b:v is at 300k
[20:49] <iive> so the bitrate is not a big issue.
[20:49] <lnb> change it to what?
[20:50] <iive> try with 1200k
[20:50] <lnb> iive: ok trying it now
[20:50] <lnb> i wonder how big the file will be
[20:50] <iive> 4 times bigger than before :)
[20:50] <lnb> heh, when it worked before it was like 535mb
[20:51] <lnb> tooo big
[20:51] <iive> how many hours is it?
[20:52] <iive> 4hours?
[20:59] <lnb> no no
[20:59] <lnb> actually i am not sure until i upload it
[21:00] <lnb> i can see its still too big in size
[21:00] <iive> lnb: few tips. you may want to consider using h264 as video codec. It gives better compression but needs relatively new flash and normal cpu for decoding.
[21:00] <iive> try 512x384
[21:01] <lnb> use h264 as in ?
[21:02] <lnb> -vcodec flv
[21:02] <iive> actually, 512x288 in your case.
[21:03] <iive> -vcodec libx264 -preset veryslow
[21:03] <iive> but... it may need experimenting on both ends. so try something shorter first.
[21:04] <iive> and look up some howto... or ffmpeg documentation.
[21:04] <lnb> i went to ffmpeg and to be honest, most of this is for pro's
[21:05] <lnb> ok so i should change -vcodec flv  to -vcodec libx264 -preset veryslow
[21:06] <llogan> lnb: what are you trying to do? (i missed the initial question)
[21:06] <lnb> trying to make a flv from mov
[21:06] <lnb> just the file size is tooo big
[21:06] <llogan> to play in a flah player on a web site?
[21:06] <iive> llogan: to convert video so it could be watched by flash.
[21:06] <llogan> *flash
[21:06] <lnb> 44070kb
[21:07] <llogan> if you want to target a particular output file size then you can use two pass encoding
[21:07] <lnb> thats not too big
[21:07] <lnb> sorry
[21:07] <lnb> its gtood
[21:07] <lnb> s/gtood/good
[21:08] <lnb> but i am going to try -vcodec libx264 -preset veryslow to see what happens
[21:08] <llogan> (although I usually just use one-pass crf and adjust it for a good quality/file size balance)
[21:08] <llogan> in fact i can't remember the last time i used two passes
[21:08] <lnb> hey iive, thanks for the help to get it to work!
[21:09] <lnb> before iive helped, every run of winff produced failure bitrate too small
[21:09] <lnb> you need to be a pro to make quality movies
[21:09] <iive> i'm not quite experienced with ffmpeg, so I kind of learn in motion.
[21:09] <lnb> well, you're pretty good :)
[21:11] <lnb> i wonder if 50 or 100 people send up movies like this, what the bandwidth is going to be
[21:11] <lnb> i will certainly find out
[21:11] <copper> hi
[21:11] <lnb> hi
[21:11] <tdj-br> Hi, where can i get ffmpeg for windows?
[21:12] <lnb> you mean winff ?
[21:12] <lnb> just put wiff in search
[21:12] <lnb> you will get it right away
[21:12] <lnb> oops
[21:12] <copper> I'm trying to compile ffmpeg-git with --enable-libfaac. I compiled and installed fdk-aac, but ffmpeg make reports: "ERROR: libfdk_aac not found"
[21:12] <tdj-br> does it have ffplay?
[21:12] <lnb> winff
[21:12] <lnb> try it
[21:12] <copper> sorry
[21:12] <lnb> its free
[21:12] <copper> I meant --enable-libfdk-aac
[21:13] <lnb> i use the same sort of app in ubuntu also
[21:13] <tdj-br> i want ffmpeg compiled to windows... for use ffmpeg.exe and ffplay.exe
[21:13] <JEEB> copper, config.log into pastebin
[21:13] <tdj-br> i want to use the rtmp
[21:13] <tdj-br> publish to rtmp
[21:13] <JEEB> tdj-br, there are windows builds of the ffmpeg package available
[21:13] <tdj-br> do you know where?
[21:14] <tdj-br> FFmpeg Windows Builds are available at Zeranoe FFmpeg Builds.
[21:14] <JEEB> http://ffmpeg.zeranoe.com/
[21:14] <JEEB> yes
[21:14] <tdj-br> sorry, i didn't read the site
[21:14] <JEEB> zeranoe's place
[21:14] <tdj-br> so sorry about that
[21:14] <JEEB> no problem
[21:14] <copper> JEEB: http://pastebin.com/raw.php?i=SRPEGpAS
[21:14] <lnb> hey maybe you guys know, when i see a tutorial movie, they have typing in there or red circle around key words etc. How do they do that?
[21:15] <copper> I installed fdk-aac-0.1.0
[21:16] <llogan> lnb: most decent video editors should allow you to do that sort of thing.
[21:16] <JEEB> copper, that looks like it's not finding stuff from the c/c++ runtime
[21:16] <JEEB> or not runtime, but the standard headers
[21:16] <lnb> ahh ok
[21:16] <lnb> is there any free for windows or ubuntu?
[21:16] <copper> JEEB: it's all installed in /usr. Do I have the wrong version or something?
[21:17] <llogan> lnb: for linux kdenlive and openshot are two that come to mind
[21:17] <tdj-br> Unable to find a suitable output format for 'rtmp://19"
[21:17] <tdj-br> anybody know if rtmp is in it build?
[21:17] <JEEB> copper, ok that could also be a problem with the tarball release it seems
[21:17] <JEEB> https://github.com/mstorsjo/fdk-aac
[21:17] <JEEB> I would personally build from here
[21:17] <llogan> note that fdk-aac in ffmpeg currently seg faults (unless Derek's patch has been applied)
[21:18] <JEEB> (note that you will have to use autoreconf with the git repo)
[21:18] <JEEB> (autoreconf -fiv f.ex.)
[21:18] <llogan> tdj-br: ffmpeg -protocols
[21:18] <copper> ok, I'll wait until it's sorted out
[21:18] <tdj-br> yes, rtmp
[21:18] <tdj-br> thanks
[21:18] <copper> thank you
[21:19] <tdj-br> should be my command line
[21:19] <llogan> the tarball installed and ffmpeg recognized it for me
[21:19] <llogan> what's your distro?
[21:19] <copper> Arch Linux
[21:19] <llogan> it's in AUR
[21:20] <llogan> and ffmpeg-git (also in AUR) should have no problem with it
[21:20] <llogan> other than the current seg fault, but that's a simple fix
[21:20] <lnb> llogan: thanks -> p   kdenlive   - a non-linear video editor
[21:20] <copper> llogan: what's the fix?
[21:21] <copper> ok it compiles with fdk-aac from AUR
[21:22] <copper> I made my own PKGBUILD before, without --enable-shared=o
[21:22] <thephilmcc> Howdy.
[21:22] <copper> no*
[21:23] <llogan> copper: http://ffmpeg.org/pipermail/ffmpeg-devel/2012-July/127559.html
[21:23] <llogan> care to paste your PKGBUILD?
[21:23] <copper> llogan: of fdk-aac or ffmpeg-git?
[21:23] <llogan> fdk
[21:23] <copper> I deleted it
[21:24] <copper> the only difference was, I didn't use --enable-shared=no
[21:24] <copper> everything else was the same
[21:25] <llogan> you even added the license file?
[21:25] <copper> hum I guess not
[21:25] <copper> no matter
[21:28] <llogan> i wonder how many people will bitch about that --enable-shared=no
[21:30] <ubitux> copper: are you a ffmpeg maintainer on arch?
[21:32] <copper> ubitux: no, just a regular joe
[21:34] <copper> I read about the fdk-aac release on hydrogenaudio.org and Google revealed ffmpeg already had support for it :)
[21:34] <JEEB> yeah, it was pushed to libav and then ffmpeg rather quickly merged it :)
[21:35] <JEEB> we have a non-nonfree, usable AAC encoder again
[21:35] <copper> it's compiling
[21:35] <JEEB> (the libavcodec aac encoder is "OK" at high'ish rates but nothing to write home about with low ones)
[21:35] <llogan> ubitux: the arch maintainer, ionut (or wonder) comes here with quesitons on occasion
[21:36] <copper> damn, it failed because it didn't find pod2Man
[21:36] <llogan> is this ffmpeg-git from AUR?
[21:37] <copper> yes
[21:37] <copper> it's missing something in makedepends I guess?
[21:39] <llogan> copper: perl provides pod2man, IIRC
[21:40] <copper> I just saw that
[21:40] <copper> it's not in my $PATH
[21:40] <llogan> and perl is in base, so it does not need to be in makedepends
[21:40] <copper> /usr/bin/core_perl/pod2man
[21:40] <llogan> worksforme
[21:41] <copper> probably my fault
[21:44] <copper> yeah, I reset my PATH in /etc/bash.bashrc.local
[21:44] <copper> bad idea I guess
[21:44] <copper> /usr/lib/gcc/x86_64-unknown-linux-gnu/4.7.1/../../../../lib/libfdk-aac.a: could not read symbols: Bad value
[21:44] <copper> collect2: error: ld returned 1 exit status
[21:44] <copper> :(
[21:45] <copper> /usr/bin/ld: /usr/lib/gcc/x86_64-unknown-linux-gnu/4.7.1/../../../../lib/libfdk-aac.a(aacenc_lib.o): relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC
[21:45] <copper> does that mean recompile fdk-aac or ffmpeg with -fPIC?
[21:45] <copper> fdk-aac I guess
[21:46] <JEEB> everything you're linking to what you're building atm, yeah
[21:50] <llogan> did you modify ffmpeg-git?
[21:52] <JEEB> it's most probably a shared libavcodec + static fdk-aac, in which case he needs fPIC
[21:52] <JEEB> (on the static fdk-aac)
[21:52] <llogan> people love to --enable-shared
[21:53] <JEEB> and then forget to run ldconfig and friends, or forget to set a special LD_LIBRARY_PATH if they installed it outside of the system library folders :)
[22:09] <BW-> hi guys! i had a clue of this before but can't find it now: i have some quicktime movies on my kodak camera, they
[22:09] <BW-> are almost uncompressed, veery large. want to compress them down to an archive quality but with much smaller sizes, suppose x264 is suitable for this.
[22:10] <BW-> how do i specify to the codec that it should apply a particular quality level (as opposed to a kbit/sec)?
[22:10] <BW-> with specifying a bitrate per second i suppose there'd be the risk that the image quality across movie compressions would get uneven, so, i want just to tell it a constant i.e. i want quality number 20 etc.
[22:11] <BW-> how do i do this?
[22:12] <beandog> -crf
[22:12] <beandog> 20 is default
[22:12] <beandog> I think
[22:12] <beandog> or 22
[22:12] <JEEB> 23
[22:12] <beandog> ah okay
[22:12] <beandog> I was close
[22:12] <bw^-> aha
[22:13] <bw^-> super!
[22:13] <bw^-> would any other config argument make any contribution to this archive material? like, turning on two-pass or sth
[22:13] <JEEB> 2pass is only usable for bitrate-based encoding modes for libx264
[22:13] <bw^-> super, so i'll try with ffmpeg -i G:\DCIM\103KM753\103_6957.MOV  -vcodec libx264 -crf 20 103_6957_c.AVI
[22:13] <bw^-> ok
[22:13] <JEEB> you can set a preset
[22:13] <JEEB> speed vs compression
[22:13] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset
[22:13] <beandog> ugh.  avi?
[22:14] <JEEB> and yeah, something not avi
[22:14] <JEEB> for H.264
[22:14] <bw^-> aha
[22:14] <beandog> do mp4
[22:14] <JEEB> be it mp4/mkv/flv
[22:14] <bw^-> any suggestions, mv4?
[22:14] <bw^-> mp4 ah right
[22:14] <JEEB> also you probably want to copy the audio
[22:15] <JEEB> because the default will encode it into <something>
[22:15] <beandog> -c:a copy
[22:15] <JEEB> acodec or c:a copy, yes
[22:15] <beandog> I could have sworn there was an x264 encoding guide in the /topic
[22:15] <beandog> maybe that was somewhere else.
[22:16] <bw^-> ah right - the camera produced audio stream is: Stream #0:1(eng): Audio: aac ([64][0][0][0] / 0x0040), 11025 Hz, mono, s16, 128 kb/s
[22:16] <bw^-> right, so probably i just want to leave that as is, it does not occupy any notable amounts of space anyhow
[22:16] <beandog> yah, just let it copy straight through
[22:16] <JEEB> ok, yeah -- aac is nicely compressed in most cases
[22:17] <JEEB> so yeah, do copy that
[22:17] <JEEB> also the case with crf is to find the highest value that still looks good for you
[22:17] <JEEB> :)
[22:17] <bw^-> -c:a gives: [mp4 @ 01ed9960] track 1: could not find tag, codec not currently supported in container
[22:17] <JEEB> o_O
[22:17] <copper> Thanks for your help, ffmpeg-git compiled fine and I managed to encode a file with -acodec libfdk_aac
[22:17] <bw^-> JEEB: aha, so the higher crf the lower quality
[22:18] <JEEB> bw^-, yes
[22:18] <JEEB> 23 is something that compresses nicely but what most people will see as tolerable
[22:18] <JEEB> or even good
[22:18] <bw^-> aha very nice
[22:18] <bw^-> what about the issue with -c:a copy here, any way around it?
[22:18] <JEEB> you probably want to encode a few thousand frames with various crf values
[22:18] <JEEB> I'm not sure
[22:18] <beandog> yah that's odd.
[22:19] <JEEB> it looks as if AAC is not good
[22:19] <JEEB> while it surely is for mp4
[22:19] <bw^-> uh, wait, on this file i have here now the source is: Stream #0:1(eng): Audio: pcm_u8 (raw  / 0x20776172), 11025 Hz, mono, 88 kb/s
[22:19] <JEEB> yeah
[22:19] <JEEB> raw pcm is not supported in mp4
[22:19] <beandog> oh there you go
[22:19] <JEEB> mov/mkv/flv for that
[22:19] <beandog> I didn't know mov supported pcm
[22:19] <bw^-> mhm. hm.. perhaps just some high-quality other format would be nice to compress to
[22:19] <beandog> except ... derp .. his source is mov and has it. -_-
[22:19] <bw^-> there's no sense in having this lowquality stream stored uncompressed
[22:19] <JEEB> mov is like the AVI of Macs
[22:20] <bw^-> mm
[22:20] <JEEB> then mp4 was based on it
[22:20] <beandog> is it?  I usually ignore it, but I thought it was closer to mp4
[22:20] <JEEB> yes, it is _used_ like AVI
[22:20] <bw^-> any suggestion for an audio codec setting that should provide a high archive quality and that's it?
[22:20] <beandog> ah, gotcha
[22:20] <beandog> bw^-: defaults are fine
[22:20] <bw^-> supr
[22:20] <JEEB> bw^-, I personally really wouldn't start compressing that stuff
[22:20] <JEEB> I would just use a container that can deal with it
[22:20] <beandog> the audio?
[22:20] <beandog> yah
[22:20] <JEEB> yeah
[22:20] <JEEB> it's small
[22:20] <beandog> that's a better idea
[22:21] <beandog> bw^-: dump it to output.mkv instead of output.mp4
[22:21] <JEEB> mov, mkv, flv -- whatever you want
[22:21] <beandog> or one of those
[22:21] <bw^-> mhm. ok.   hmm.. mov i
[22:21] <JEEB> although I'm not sure if flv supported raw pcm :P
[22:22] <JEEB> the flv specs are at least short
[22:22] <bw^-> m not so fond of.. then mkv or flv.. which ought to be the most supported?
[22:22] <JEEB> commercial solutions might not support matroska as much, but personally of those two I'd probably select matroska
[22:27] <copper> ffmpeg -i foo.wav -acodec libfdk_aac -flags +qscale -global_quality 2 -afterburner 1 foo.m4a
[22:27] <copper> -global_quality 1 doesn't work
[22:28] <beandog> if you don't want that, just do mp4 and reencode the audio to aac
[22:32] <JEEB> copper, "VBR quality %d out of range, should be 1-5\n"
[22:32] <JEEB> or wait
[22:32] <JEEB> 1 _should_ work
[22:32] <JEEB> at least looking at the libfdk-aacenc code
[22:33] <JEEB> are you trying with the tarball or mstrosjo's git repo?
[22:33] <JEEB> I haven't had the time to actually build fdk-aacenc yet
[22:34] <copper> argh
[22:34] <copper> forgot about mstrosjo's git
[22:34] <copper> doing several things at once
[22:34] <JEEB> but I don't think it really changed
[22:35] <JEEB> https://github.com/mstorsjo/fdk-aac/commits/master
[22:35] <JEEB> only automake-related changes
[22:35] <JEEB> as far as I can see
[22:37] <llogan> copper: did you apply the patch a linked to?
[22:37] <copper> llogan: yes
[22:38] <copper> http://ffmpeg.org/pipermail/ffmpeg-devel/2012-July/127559.html
[22:38] <llogan> someday i'll read what i type before pressing enter
[22:38] <copper> a wouldn't worry about it ;)
[22:38] <llogan> it's ironic that i made a comment on a patch about readability but had a similar typo
[22:39] <copper> gotta walk the dog, brb
[22:40] <llogan> burek: is that your first patch that got committed?
[22:40] <llogan> i guess i could just look myself </lazy>
[22:45] <BW^-> great, thanks a lot guys =D
[22:46] <beandog> BW^-: have fun :)
[22:50] <GA-Flix> Good day everyone
[22:50] <GA-Flix> Does anyone know how I'm able to create a 32bit FFMPEG executable that runs under Windows 7 x64 ?
[22:53] <beandog> there's gotta be some prebuilt windows binaries
[22:53] <beandog> just download a 32bit one
[22:54] <GA-Flix> but i need to compile it myself (license and so on)
[22:54] <GA-Flix> :(
[22:54] <llogan> you need non-free?
[22:54] <GA-Flix> i need it to pack it into an own executable (free, but not public)
[22:55] <GA-Flix> maybe i'll publish if it works 100% bugfree ^^, but i think i've read that i have to compile it myself if i want to use it in another executable
[22:55] Action: beandog has no idea
[22:55] <beandog> I've heard compiling it on windows is a major pain, though
[22:56] <beandog> and I don't doubt it
[22:56] <GA-Flix> i compiled the sourcecode for windows in ubuntu
[22:57] <beandog> did it run?
[22:57] <GA-Flix> on the 32bit windows xp it did, but in windows 7 64bit not :(
[22:57] <JEEB> that's weird
[22:57] <GA-Flix> true :(
[22:58] <GA-Flix> wait a sec, i can post the error message
[22:58] <JEEB> mingw should by default only need the msvcrt and that should be available on everything win2k+
[22:59] <GA-Flix> damn, the error-message is in german ...
[23:00] <GA-Flix> translated, it means something like: """The version of "ffmpeg.exe" is not compatible with the executed Windows-version. Open the computer's system information to check whether an x32 or an x64 version of the program is required and contact the author of the software."""
[23:01] <JEEB> ok, I'm not sure what exactly you've done wrong but you've done something wrong indeed
[23:01] <JEEB> you could use zeranoe's scripts to make a mingw toolchain
[23:01] <JEEB> and build the windows binary with that
[23:02] <JEEB> http://ffmpeg.zeranoe.com/blog/?p=96
[23:03] <GA-Flix> ok, thanks, i'll try
[23:04] <beandog> wow
[23:04] <beandog> that's a pretty impressive feature list
[23:05] <JEEB> not really that impressive, but it's probably the sanest script available to build a toolchain like that
[23:06] <JEEB> you can then edit it a bit to even built a mingw toolchain for win32
[23:06] <beandog> freak this thing is huge, too
[23:06] <JEEB> you haven't seen other people's scripts for mingw/mingw-w64 toolchains then
[23:06] <beandog> I guess not
[23:07] <beandog> I'd rather not. :)
[23:12] <lake> I notice this ticket: http://ffmpeg.org/trac/ffmpeg/ticket/442
[23:12] <lake> how can i find out where that was implemented? and how to use it?
[23:15] <llogan> lake: http://git.videolan.org/?p=ffmpeg.git;a=commit;h=72868144e5862168a4f829557b7a41f686b8b12d
[23:15] <JEEB> lake, it's a video filter and it seems like some documentation was added for it :)
[23:16] <llogan> "git log" is useful for finding commits
[23:16] <JEEB> as well as git show
[23:16] <JEEB> (show does show the code changes, too, though)
[23:19] <lake> llogan: thanks for the commit link.
[23:19] <lake> when i run the example command "ffmpeg -i destination.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png"
[23:20] <lake> Error initializing filter 'select' with args 'gt(scene,0.4)'
[23:20] <lake> that's what i get
[23:23] <GA-Flix> how dafuq do i install subversion correctly?
[23:30] <GA-Flix> oh okay, i got a runnable linux executable (using the terminal, i can convert audio from one format to another with new bitrates etc, as needed); how do i get the windows executable now? :)
[23:32] <sacarasc> If you want a Windows executable, you'll either have to compile on Windows, cross compile on another OS or download one from the net somewhere.
[23:34] <lake> i'm going to grab HEAD and try again
[00:00] --- Sat Jul 14 2012


More information about the Ffmpeg-devel-irc mailing list