[Ffmpeg-devel-irc] ffmpeg.log.20121204

burek burek021 at gmail.com
Wed Dec 5 02:05:01 CET 2012


[00:13] <Yulth> Hi everyone!
[00:14] <Yulth> I'm trying to convert flac audio to 96k aac_he_v2, but maximum bitrate achieved is 64kbps. Is there a limitation?
[00:16] <JEEB> that's where I generally see HE-AAC finishing itself, as in most encoders don't let you go higher because the gain you get from the rape that goes into HE-AAC(v2) gets smaller and you might as well be using a better profile
[00:16] <Yulth> mmm
[00:16] <Mavrik> JEEB, what exactly would be a "better" profile?
[00:18] <JEEB> if you are going to use that high of a bit rate you could start moving towards LC-AAC
[00:18] <JEEB> depends on how many channels you have naturally
[00:19] <Yulth> here the pastebin: http://pastebin.com/wa8cfHMX
[00:24] <JEEB> if it doesn't change with -b:a 96k then I will guess that fdk-aac doesn't want to let you code HE-AAC(v2) with a too high a bit rate
[00:30] <Yulth> JEEB: efficiency reasons?
[00:31] <JEEB> well, you do understand that HE-AAC(v2) is a mode that tries to use various methods instead of just using the normal compression algorithms to try and re-enact something similar to what was cut off to make the source easier to encode
[00:32] <JEEB> this is mostly useful with lower bit rates, where the extra removal etc. is needed
[00:32] <JEEB> if you are going to have a higher bit rate used, you might as well use a profile that does less things like that
[00:34] <JEEB> also the without-v2 profile might let you encode a bit higher, but in general >64kbps or so is where many AAC encoders switch to LC
[00:39] <Yulth> I understand
[00:42] <Yulth> however, at least in my case, encoding from FLAC to HE-AACv2 64kbps sounds really bad. Nothing like this "high audio quality" said by some AAC defenders.
[00:44] <JEEB> well, HE-AACv2 isn't for "high quality" >_>
[00:44] <JEEB> it's for "low bandwidth use cases where the audio has to be more or less like the source"
[00:44] <Yulth> mmm
[00:45] <JEEB> try LC-AAC at around 96kbps-128kbps or so
[00:45] <JEEB> there is a cutoff at some rate (which can be configured methinks), but otherwise it should be nicer
[00:47] <Yulth> I'm going to try it and perform several "listening" tests, because in some info sources like Wikipedia, it's shown that, for example, HE-AACv2 at 48kbps are able to confuse a listener about what he is listening: the original file or the coded...
[00:48] <Yulth> and in real practices, it seems othert hing
[00:49] <JEEB> also there was an afterburner setting that might or might not make the result better
[00:49] <JEEB> Yulth, I'd be quite surprised btw if that 48kbps case would happen with all sources :D
[00:52] <Yulth> JEEB: wikipedia shows only uncompressed audio as source :)
[00:53] <JEEB> anyways, HE-AACv2 is useful where it's generally used -- 32kbps-64kbps encoding :)
[00:53] <JEEB> for low-bandwidth needs
[00:54] <Yulth> ok understood :)
[00:54] <Yulth> thanks for your time!
[01:04] <teratorn> anyone can tell me how to set pts values on audio packets correctly? my code for making a webm/vorbis file mostly works, but the sound doesn't sound right... seems to play back too fast in mplayer. http://hastebin.com/luputuxilo.c
[01:04] <teratorn> ffplay can't identify the file's duration...
[02:57] <llogan> teratorn: might want to try libav-user mailing list (that's a FFmpeg mailing list).
[03:58] <dericed> is there any reason why adding -filter_complex "[0:1] [0:2] amerge" should make an mpeg output fail in dvdauthor?
[04:00] <dericed> dvdauthor gives "ERR: Cannot infer pts for VOBU if there is no audio or video and it is the first VOBU" if the input mpeg used -filter_complex "[0:1] [0:2] amerge" but otherwise it is fine.
[04:26] <James123_> Hey guys
[04:53] <kcm1700> Hello
[06:08] <kcm1700> does ffmpeg native h264 decoder support svc(scalable video coding) video?
[06:30] <escortkeel> hello again! would there be a way for me to modify the timestamp of frames written to the output file? such that, say, the first frame starts at time 50 seconds "into" the video?
[06:30] <escortkeel> thanks so much!
[08:51] <zap0> yes
[12:07] <AlecTaylor> hi
[12:10] <AlecTaylor> How do I get MP3 encoding with FFmpeg?
[12:13] <JEEB> you build ffmpeg with the LAME library
[12:17] <Miesco> Hi
[12:17] <Miesco> ffmpeg -ss 3:31:03 -t 3 -i /srv/ftp/public/dal/jam_sessions/dec1/dec1_basement_garage.mp3 bitch_slap.wav
[12:17] <Miesco> This works when I make it bitch_slap.mp3, but when I try to convert the audio it doesn't work at all
[12:19] <Miesco> wait NVM it did work
[12:30] <Miesco> Im trying to combine a .gif and a .wav
[13:00] <Miesco> I got all these images and im trying to convert them to a .mp4
[13:00] <Miesco> The command works but its only getting like half the images...
[13:01] <Miesco> ffmpeg -i image%02d.png aresult.mp4
[13:01] <Miesco> Theres 28 images
[13:21] <bartus> how to build static ffmpeg libs which has no undefined symbols ?
[13:22] <JEEB> by making sure all libraries you're linking against are in reach of the compiler and linker?
[13:29] <bartus> I'm on debian, have all static library installed with -dev packages. And still got libavcodec.a full of undefined references :/
[13:31] <JEEB> then you are doing it wrong, pastebin all of the compilation log into a pastebin provider of your choice and link here
[14:33] <Miesco> FUCK
[14:39] <burek> the most common word among developers :)
[16:15] <alphis> anyone familiar with issues compiling ffmpeg-1.0 on gentoo?
[16:16] <cbsrobot> ubitux: do we support youtube xml captions ?
[16:19] <ubitux> nope
[16:19] <ubitux> you can write a demuxer though
[16:19] <cbsrobot> yeah I thought so
[16:20] <ubitux> how does it look like?
[16:20] Action: ubitux is afraid of "xml" in the name
[16:20] <cbsrobot> eh
[16:20] <cbsrobot> wait
[16:21] <cbsrobot> http://pastie.org/5479245
[16:21] <cbsrobot> ubitux: ^
[16:22] <ubitux> erm, you'll need a decoder too :)
[16:22] <ubitux> (to unescape the html codes)
[16:22] <ubitux> maybe the text decoder can be changed
[16:23] <ubitux> it looks like realtext a little
[16:23] <ubitux> shouldn't be hard to support
[17:17] <bakers> Using libx264 if I use -preset slow that means I can't use -vb correct?
[17:19] <JEEB> bakers, incorrect
[17:19] <JEEB> although you should now use -b:v to set video bit rate
[17:19] <bakers> -vb = old buster -b:v = new hotness?
[17:20] <JEEB> basically a lot of settings were modified by elenril IIRC to add those suffixes
[17:20] <JEEB> -vb probably still works
[17:20] <JEEB> it's like there's now -c:v/a/s for codec selection too
[17:21] <JEEB> anyways, libx264 presets have /nothing/ to do with rate control
[17:21] <JEEB> you still pick your rate control mode
[17:21] <JEEB> and use it
[17:21] <JEEB> it's true that in many cases the default crf rate control mode is better than bitrate based stuff, because it's closer to "constant quality" than "hit exactly this file size"
[17:21] <JEEB> and most people don't have size limits but that they just want a good quality video for their own eyes
[17:22] <JEEB> (also crf only needs one pass because you basically set the value constant that multipass bitrate based encoding tries to optimize for while hitting your target size)
[17:22] <Nick-S> wow
[17:25] <Nick-S> i want to convert a png sequence to mp4, but i want to be able to connect a few of those mp4's into one big mp4, so i thought using the mpg format which is concatable
[17:25] <Nick-S> is it a good approach or i am losing quality?
[17:25] <JEEB> if you are re-encoding them, yes :P
[17:26] <Nick-S> mpeg-2(?) into mp4
[17:26] <Nick-S> so whats my options
[17:26] <Nick-S> mp4 is not concatable
[17:26] <JEEB> yes, it isn't
[17:26] <Nick-S> mpeg-2 is
[17:26] <JEEB> uhh, bit stream formats generally are methinks >_> so is MPEG-4 Part 10 (AVC)
[17:26] <Nick-S> hmm
[17:26] <JEEB> although I bet you mean H.222 aka MPEG-TS
[17:26] <theos> ah sweet! didnt know this channel existed
[17:27] <Nick-S> so i could concat them all
[17:27] <Nick-S> theos: if you came a minute before you could see my wow when i entered the channel
[17:27] <Nick-S> it was for the same reason
[17:27] <theos> Nick-S yea i can understand that feeling :)
[17:27] <Nick-S> JEEB, whatever ffmpeg takes for .mpg
[17:27] <JEEB> that is a type of H.222, MPEG-PS then
[17:28] <Nick-S> it's like a little paradise
[17:28] <JEEB> (MPEG-TS and MPEG-PS are containers basically)
[17:28] <Nick-S> jeeb: so i should try concating similaryly mp4(h264?)
[17:28] <theos> how can i compress mp4 files? with what i try, the output file is always bigger than the input :D
[17:28] <JEEB> theos, the usual procedure, pastebin your current command line and its output onto a pastebin provider of your choice
[17:28] <JEEB> and link it here
[17:28] <theos> ok
[17:29] <JEEB> Nick-S, you can't concat mp4 (you probably mean the container colloquially called "MP4")
[17:29] <Nick-S> JEEB, so h264 comes in the same contaibner?
[17:29] <JEEB> at least like files
[17:29] <Nick-S> JEEB, is there a h264 compliant container that i could use for concating?
[17:29] <theos> JEEB do you know the general way of compressing mp4 files?
[17:29] <JEEB> theos, just show me what you're doing right now and I'll (most probably) tell you what you're doing wrong and what you could be doing better
[17:30] <JEEB> now do what I said
[17:30] <JEEB> Nick-S, you could of course possibly concat H.264 in MPEG-TS or just concat multiple raw H.264 streams (without container)
[17:30] Action: Nick-S likes JEEB approach
[17:30] <bakers> JEEB: If did something without any options like ffmpeg -i input.avi /tmp/output.mp4
[17:31] <JEEB> ok
[17:31] <bakers> It selects x264 and aac, and just uses CRF?
[17:31] <bakers> at some default setting?
[17:31] <JEEB> ok
[17:31] <JEEB> that's all fine
[17:31] <JEEB> now basically try to find the highest crf that still looks good with -crf
[17:31] <Nick-S> JEEB, how would i produce h264 in mpeg-ts?
[17:31] <JEEB> Nick-S, call the output file something.ts ?
[17:31] <Nick-S> i prefer the latter
[17:32] <theos> JEEB i am trying --> ffmpeg -i input.mp4 -acodec mp2 output.mp4
[17:32] <JEEB> ...
[17:32] Action: JEEB slaps theos 
[17:32] <JEEB> why mp2 audio >_>
[17:32] <theos> :/
[17:32] <theos> because some dude said to use it on his blog :D
[17:32] <JEEB> I told you, start out from -crf 23 and go higher if it still looks good
[17:33] <JEEB> then when it stops looking good, you go one back
[17:33] <JEEB> and you're at the highest crf that still looks good for you
[17:33] <Nick-S> jeeb, does .ts means h264 necesserially?
[17:33] <JEEB> tell me when you've more or less done that
[17:33] <theos> oh ok. let me try. thanks
[17:33] <JEEB> Nick-S, naturally no
[17:33] <JEEB> it just makes ffmpeg assume that you want the MPEG-TS container
[17:34] <JEEB> if your mp4s already are H.264, then you use -c:v copy (and if you have audio -c:a copy too), if you're only encoding you naturally use -c:v libx264
[17:35] <theos> (what does -crf do?)
[17:36] <JEEB> it sets the value for the crt rate control (constant rate factor), it is the closest thing you have in any video encoder available to "constant visual quality"
[17:36] <theos> cant find it in the manual
[17:37] <JEEB> it is a libx264-specific setting
[17:37] <JEEB> since no other video encoder has it :P
[17:37] <theos> oh :D
[17:37] <JEEB> higher -> more compression, might look worse; lower -> less compression, might look better (until a limit of course where the difference from the source becomes too small for your eyes to notice)
[17:40] <theos> ffmpeg: unrecognized option '-c:v'
[17:40] <JEEB> uh-oh
[17:40] <klaxa> compile from source! D:
[17:40] <theos> :(
[17:40] <JEEB> does the version line of your ffmpeg say Libav or FFmpeg at the end?
[17:40] <bakers> Or get a static build, that's what I do.
[17:40] <klaxa> but... but... --omg-optimized !!
[17:40] <JEEB> if it's Libav, switch to using the command avconv
[17:41] <JEEB> if it's ffmpeg, then you'll just have to update in one way or another, because that's a sign of a build from before elenril's updates
[17:41] <theos> FFmpeg version SVN-r0.5.1-4:0.5.1-1ubuntu1.3, Copyright (c) 2000-2009 Fabrice Bellard, et al.
[17:41] <JEEB> ok, that's just old old
[17:42] <theos> hehe
[17:42] <JEEB> theos, ^ have some static linux builds
[17:42] <JEEB> grab the newest and try with that
[17:42] <theos> ok thanks
[17:43] <JEEB> I don't think a 2009 ffmpeg even has x264 or crf by default :P
[17:43] <JEEB> wouldn't be surprised if it did mpeg4 (MPEG-4 Part 2) video with a crappy bit rate
[17:48] <theos> looks like a lot has changed since i last updated :s
[17:48] <JEEB> 2011 saw a lot of changes to how several defaults were set, as well as added quite a few ways of doing things
[17:49] <JEEB> (removed a few of the old ways)
[17:49] <Nick-S> JEEB, then i need to do -vcoded?
[17:49] <JEEB> Nick-S, you sure come straight out of the bush so to speak, which leads me to ask "pardon me?"
[17:50] <Nick-S> JEEB, sorry
[17:51] <theos> ah crap! its for 64bit :/
[17:52] <JEEB> yeah, one of those two was IIRC limited to 64bit
[17:53] <theos> today is not a good day
[17:55] <JEEB> well, in the worst case compiling ffmpeg is not that hard if you just need to encode H.264 and AAC, for example
[17:56] <theos> yea downloading source >.>
[17:56] <bakers> theos: You're still running 32 bit?
[17:56] <JEEB> (you just need to probably compile yasm, x264, fdk-aac and ffmpeg itself
[17:56] <theos> bakers yes. i like 32bit :)
[17:56] <bakers> How come?
[17:56] <theos> its cheap. consumes less power
[17:57] <JEEB> what...
[17:57] <JEEB> I'm sorry but that just made no sense
[17:57] <JEEB> anyways
[17:57] <theos> i know :)
[17:57] <bakers> JEEB: I'm glad you said it :)
[17:57] <theos> its an old pc. thats why its 32bit
[17:57] <JEEB> yasm because IIRC current x264 would prefer a newer than 1.0 yasm, x264 to get an up-to-date H.264 encoder, fdk-aac because it's the best aac encoder atm, and then finally ffmpeg
[17:57] <JEEB> yes, that's a better reasoning :P
[17:58] <theos> :)
[17:58] <JEEB> most old PCs nowadays use a /lot/ more power than old ones
[17:58] <JEEB> I mean, an ivy bridge machine uses a /lot/ less than say Athlon XP or Pentium 4
[17:58] <theos> thats didnt make sense either :P
[17:58] <JEEB> or K8
[17:58] <theos> you mean new pcs?
[17:59] <JEEB> new CPUs
[17:59] <Nick-S> JEEB, how do i put a .ts into a mp4 container?
[17:59] <JEEB> Nick-S, whaaaaat
[17:59] <theos> hmm but they can process more in less time
[17:59] <JEEB> you've completely lost me, sorry
[17:59] <JEEB> theos, yes? And they're still less power-hungry than most of the other CPUs during the 2000s
[18:00] <bakers> Nick-S: .ts files are the "container"
[18:01] <Nick-S> but when i want to provide back a real .mp4 file?
[18:02] <JEEB> ok, I'm going to assume things here then
[18:02] <bakers> Nick-S: It's either a .ts file or a .mp4 file they're different "containers"
[18:02] <JEEB> you encoded H.264 in mpeg-ts, or copied the H.264 streams from the mp4 files you already encoded
[18:02] <JEEB> then you concaterated the mpeg-s files to make one
[18:02] <JEEB> and now you are asking how to put it back into one mp4 file?
[18:02] <JEEB> Is this correct?
[18:03] <Nick-S> wait let me follow this correctly
[18:03] <JEEB> You never told us what you were exactly doing and where you are right now
[18:03] <JEEB> So I am now /assuming/ things
[18:03] <Nick-S> yes
[18:03] <JEEB> I am not an esper, unfortunately
[18:03] <Nick-S> correct
[18:03] <JEEB> ffmpeg -i concaterated_ts_file.ts -c:v copy (if you have audio also -c:a copy) out.mp4
[18:04] <JEEB> it might want you to add some bit stream filter in the middle, and if it does -- please oblige
[18:04] <Nick-S> wait i want to encode image sequence into mpeg-ts h264
[18:05] <JEEB> well that's the point, you just said "correct" to "have you already completed the other steps I earlier mentioned to get yourself mpeg-ts files and then concaterated them"
[18:05] <Nick-S> well, currently i do it with just out.mp4 whuch does not create .ts
[18:06] <JEEB> quite naturally
[18:06] <JEEB> so you are still in the phase of "I want these pictures into H.264 in mpeg-ts"?
[18:06] <JEEB> aka in the very beginning
[18:06] <Nick-S> ya, like -f image2 -i filemask out.ts?
[18:06] <JEEB> after the filemask
[18:07] <Nick-S> after the filemask what?
[18:07] <JEEB> I'm writing that damn it
[18:07] <JEEB> it's not like I'm an automated answering bot for christ's sake
[18:07] <Nick-S> i wasn't suspecting that
[18:07] <JEEB> -c:v libx264 -crf YourPreferredCRFValue -preset YourPreferredLibx264Preset -an out.ts
[18:08] <theos> but you sure type like hell
[18:08] <theos> i thought you had shortcuts for pasting stuff :)
[18:08] <JEEB> I wish I had at times :P
[18:08] <bakers> JEEB is super human
[18:09] <theos> JEEB is the man/woman
[18:09] <bakers> JEEB: What do you do for a living?
[18:09] <Nick-S> ok mr JEEB, one more question, what if i want to concate .webm?
[18:09] <Nick-S> can i put that into .ts too?
[18:10] <Nick-S> wtf is -crf btw?
[18:10] <theos> heh
[18:10] <Nick-S> shit i have to go to rehersal, life is so busy
[18:10] <bakers> Nick-S: http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[18:10] <theos> constant rate factor
[18:11] <bakers> That's a good explanation of CRF
[18:11] <Nick-S> ok, do i need =c:v when encoding the image sequence? normally i don't use it
[18:12] <bakers> Nick-S: probably not, ffmpeg tries to guess based on output file extension
[18:12] <JEEB> that choice can often be not what you want tho
[18:12] <JEEB> so setting the output codec is often a good idea
[18:13] <Nick-S> btw, have anyone tried xuggler? didnt' really work for me, but good part of it is that i can start encoding while rendering, which when i run the executable i can not
[18:13] <bakers> JEEB: true, but it's a usually pretty good...
[18:13] <bakers> assuming you only want tthe common formats: x264/webm/avi
[18:13] <JEEB> Nick-S, you should be able to concaterate webm (which is matroska) with mkvtoolnix's mkvmerge
[18:13] <Nick-S> JEEB, yes i did that, is mkv merge is actually ffmpeg?
[18:13] <JEEB> no
[18:14] <JEEB> but it's one of the best tools for that kind of job :P
[18:14] <Nick-S> but can't ffmpeg do it?
[18:14] <theos> JEEB while compiling, should i enable everything from external lib support?
[18:14] <JEEB> no
[18:14] <JEEB> theos, uhh no
[18:14] <Nick-S> JEEB, will you be here at another time?
[18:14] <JEEB> you should only enable what you need in general
[18:14] <theos> how do i find out what i need?
[18:14] <JEEB> Nick-S, no -- ffmpeg cannot concaterate muxed contents
[18:14] <JEEB> theos, so far it looks like you need libx264 and libfdk_aacenc
[18:14] <bakers> theos: Ya use external libs
[18:15] <theos> hmm
[18:15] <JEEB> because you noted that you need H.264 and a good AAC encoder to put that stuff into MP4
[18:15] <bakers> theos: Unless you need a static build
[18:15] <theos> yes
[18:15] <theos> i just need to compress mp4 right now
[18:15] <JEEB> <bakers> JEEB: What do you do for a living? <- During the summer I coded up a ut video encoder for libavcodec in a Google Summer of Code project, after that I've had no income :P
[18:15] <Nick-S>  ok
[18:16] <Nick-S> i'm gonna use that precious knowledge later on i must leave now, thank you good people
[18:16] <theos> cya
[18:16] <bakers> JEEB: What's a "ut video encoder"
[18:17] <theos> JEEB looks like you had a good mentor
[18:17] <JEEB> theos, yes -- kshishkov <3
[18:17] <theos> :)
[18:17] <bakers> How is participating in the Google summer of code? Is it cool?
[18:17] <JEEB> bakers, Ut Video is a Japanese lossless not-so-hard-by-design huffman-base video format
[18:17] <JEEB> *huffman-based
[18:19] <theos> bakers its a competition where google gets code for cheap from students. if students cant complete on time, their mentors have to complete it :D thats the short version
[18:20] <JEEB> (no they don't have to btw)
[18:20] <theos> (cool)
[18:20] <JEEB> (and it's not a competition, it's just a project that Google gains goodwill with, giving money to students on work for open source projects)
[18:22] <theos> (competition for getting selected by a mentor)
[18:25] <theos> ERROR: libfdk_aac not found :)
[18:26] <theos> i will try tomorrow. thanks for help. night
[19:15] <mykul> So, i'm transcoding audio streams on the fly using an http input stream and a pcm_s16le wav output stream. ffmpeg cannot determine the length of the output to put in the WAV header, so is there a way I can manually set the header?
[19:27] <mykul> using ffmpeg that is
[19:28] <klaxa> you could output raw pcm and remux it into a wav container
[19:28] <klaxa> that'd be a two step approach though
[19:29] <mykul> hmm that might have to work
[19:30] <mykul> i couldn't find any command line options to set the header values
[19:30] <klaxa> i don't think there are any
[19:31] <mykul> klaxa: how do i output raw pcm?
[19:31] <mykul> without a container?
[19:32] <relaxed> -f s16le
[19:32] <klaxa> yeah that
[19:32] <mykul> cool
[19:32] <mykul> thanks folks
[19:32] <relaxed> You can get a list via `ffmpeg -codec` & `ffmpeg -formats`
[19:33] <relaxed> -codecs
[19:35] <mykul> ah so by two step approach you mean an intermediate file would have to be written to disk
[19:36] <mykul> ?
[19:36] <relaxed> yes, the raw pcm
[19:38] <mykul> alas
[19:38] <JEEB> if you only have audio and want to just read that up on the other side
[19:38] <JEEB> you could just stream raw pcm
[19:38] <JEEB> or raw pcm in something like matroska
[19:38] <JEEB> that can be read without a header
[19:38] <JEEB> and read it up as you go on the other side
[19:39] <mykul> well this is for a web interface that's offering a file download in /wav format
[19:39] <mykul> .wav*
[19:39] <JEEB> oh
[19:39] <mykul> the server is downloading from a cdn and transcoding the stream and piping it to the client as a download
[19:40] <klaxa> so you're downloading a file, reencode it and offer it for download again?
[19:41] <klaxa> ah no it's a stream, right?
[19:41] <mykul> yes a stream
[19:41] <mykul> so nothing is written to the webserver
[19:41] <relaxed> if you're not streaming it to the client it makes no difference, correct?
[19:41] <mykul> just hangs out in a buffer
[19:41] <mykul> we are streaming it to the client
[19:43] <mykul> say an ogg vorbis file is uploaded to our cdn. a user can request that file in wav, and the webserver will transcode it on the fly while piping the transcoded stream to the client
[19:43] <relaxed> you could use -c:a pcm_s16le output.flv
[19:44] <mykul> what would that help with?
[19:46] <relaxed> Sorry, I was thinking streaming but that's not what you want.
[19:46] <mykul> ah yeah
[19:46] <mykul> i could have been more clear
[19:46] <mykul> the thing is, it's not a huge deal
[19:47] <mykul> but certain players wont play the .wav with a length of 0
[19:47] <mykul> cough itunes
[19:47] <klaxa> eww
[19:47] <mykul> which i'm guessing a lot of people have set as their default player
[20:14] <bakers> What exactly does -tune animation do?
[20:14] <JEEB> it's a tuning of libx264's settings that is aimed towards something alike to cell animation with not much added grain or noise
[20:15] <JEEB> lowers AQ strength, raises either bframecount or refcount
[21:26] <pjoseph> I have 13820 png files to convert to a video 1752 secs long, but using 'ffmpeg -f image2 -i ./test-%05d.png -r 7.89 -y input.mp4' I only get a video 552 secs long.  What option should I change?  http://pastebin.com/WaSpbLDN
[21:27] <JEEB> try setting a frame rate before -i
[21:27] <JEEB> with -r
[21:27] <pjoseph> JEEB: Ok, thanks.
[21:45] <t4nk357> I have a legal question.  If I charge for software that does not embed or get distributed with ffmpeg, but it issues command line commands to ffmpeg on the end user's machine, does this violate the licensing terms for ffmpeg?
[21:49] <JEEB> I don't think we have lawyers here, but if you are talking about using a GPL-licensed ffmpeg over the command line, then I /think/ the general concensus is that since you are not linking to it or including its code, you aren't making your application GPL with that. Just that if you distribute ffmpeg or any (L)GPL component, you have to give out its source code as well
[21:49] <t4nk357> Thank you Jeeb.
[21:51] <JEEB> IANAL tho
[21:51] <JEEB> as I said
[21:52] <t4nk357> I know...  This is a chat room, I wouldn't seek legal advice here.  Just trying to get a sense of things before I actually do throw money at a lawyer
[22:00] <An_Ony_Moose> I'm recording my screen (ffmpeg -f x11grab -framerate 25 -video_size 1280:1024 -r 25 -i :0.0 -b 20M -bt 5M record.mpg) and I get a very large amount of "packet too large" and "buffer underflow" errors
[22:00] <An_Ony_Moose> What causes this?
[22:29] <relaxed> remove -bt
[22:30] <relaxed> Are you encoding to mpeg1? why?
[22:36] <An_Ony_Moose> relaxed: cause it's the default... What should I be encoding to?
[22:37] <An_Ony_Moose> relaxed: and removing it doesn't fix the errors
[22:39] <An_Ony_Moose> ok, switched to ogv.
[22:46] <relaxed> ffmpeg -f x11grab -framerate 25 -video_size 1280:1024 -i :0.0 -c:v libx264 -crf 20 output.mp4
[22:47] <relaxed> decrease the -crf value for more quality, increase for smaller size
[22:50] <An_Ony_Moose> relaxed: thanks
[23:25] <llogan> An_Ony_Moose: -i :0.0 -c:v libx264 -crf 0 -preset ultrafast output.mkv
[23:50] <norbert_> hi all, can anyone tell me where I can get the latest tar.gz for (lib)avutil?
[23:50] <norbert_> I tried Googling for "libavutil source download" and "avutil download" and so on, but just information about .deb and .rpm packages and such shows up
[23:50] <norbert_> tried checking the git here http://git.videolan.org/ but it doesn't seem to be there either
[23:51] <iksik2> hello
[23:51] <norbert_> iksik2: hi
[23:51] <brx_> hello, could someone please take a glance at my ffmpeg problem here please: http://pastebin.com/rbNd7G4q
[23:51] <cbsrobot> norbert_:  you only need libavutil ?
[23:51] <norbert_> cbsrobot: I think so
[23:52] <cbsrobot> see: http://git.videolan.org/?p=ffmpeg.git;a=tree
[23:52] <iksik2> i'm trying to create a small video file from images. video file IS created, and its length/speed seems to be okey also, but only first image is displayed across the whole movie: ffmpeg -y -r 1 -f image2 -i img%03d.jpg -qscale 4 -vcodec flv out.flv
[23:52] <norbert_> it is a too old version on my Debian stable (+ debian-multimedia) system
[23:52] <iksik2> what am i doing wrong here? :(
[23:54] <llogan> brx_: do not use sameq
[23:54] <brx_> on both commands llogan ?
[23:54] <llogan> ever
[23:54] <llogan> http://superuser.com/a/478550/110524
[23:54] <llogan> also, it no longer exists in recent ffmpeg
[23:55] <norbert_> cbsrobot: so, I used git to get the whole ffmpeg pack and how I can I now just compile/link/install libavutil?
[23:55] <cbsrobot> norbert_: I'm not sure what you  are trying to solve
[23:56] <brx_> llogan, from what you saw is that all i need to do, or was that just the first thing you saw?
[23:56] <cbsrobot> I think you will need to check ./configure --help to see what options you have
[23:57] <llogan> brx_: that was the first thing
[23:57] <norbert_> cbsrobot: I'm trying to compile x264 but its configure told me: "Warning: PIX_FMT_RGB is missing from libavutil, update for swscale support"; because "pkg-config --modversion libavutil" tells me my libavutil is too old (50.43.0)
[23:58] <cbsrobot> norbert_: well I suggest if you update x264 to the newest version, first try to get ffmpeg to the newest version
[23:58] <norbert_> but my goal is to get an ffmpeg with x264 :)
[23:59] <llogan> brx_: replace -sameq with "-q:v 2" for your mpg output
[23:59] <norbert_> I guess I could make ffmpeg without x264, to get an avutil that would allow me to compile x264 to get ffmpeg with x264
[23:59] <norbert_> :)
[23:59] <teratorn> anyone know of an ffmpeg-using project that synthesizes audio and encodes to (any) container format?
[23:59] <teratorn> I need example of setting pts correctly
[23:59] <brx_> llogan, ty i shall try that now
[23:59] <llogan> [mp1 @ 0xb0d420] Header missing
[23:59] <teratorn> since ffmpeg has shit for examples
[23:59] <llogan> doesn't look promising
[00:00] --- Wed Dec  5 2012


More information about the Ffmpeg-devel-irc mailing list