[Ffmpeg-devel-irc] ffmpeg.log.20120529

burek burek021 at gmail.com
Wed May 30 02:05:01 CEST 2012


[01:14] <burek> saivert, use libaacplus
[01:15] <burek> it has superb quality
[01:23] <pasteeater> specifically for <=64kbps bitrates
[01:36] <walker> Hi folks! I have ffmpeg live-transcoding an RTP stream. It works well, save that I'd like it to process video packets as fast as it gets it, rather than sleep for rtp timestamps. Any options for that that I'm overlooking? Thanks!
[01:37] <walker> (in other words: I want it to "catch up" as fast as it can, rather than transcode true to the rtp timestamps)
[01:48] <burek> maybe there is a reason why it waits for those timestamps :)
[01:49] <walker> It's ideal for transcoding to files, but not so much a live stream broadcast.
[03:06] <chymist_chris> what are the flags for ripping an mp3 from a video file with avconv
[03:09] <pasteeater> chymist_chris: for ffmpeg it's: ffmpeg -i input -acodec copy output.mp3
[03:09] <pasteeater> or -c:a copy
[03:09] <pasteeater> assuming your input contains mp3
[03:10] <pasteeater> otherwise you have to re-encode: ffmpeg -i input -acodec libmp3lame -aq 4 output.mp3
[03:11] <chymist_chris> thats what I was using before
[03:11] <chymist_chris> but I've upgraded to ubuntu 12.04
[03:11] <pasteeater> then i guess you can install libavcodec-extra-53 to enable libmp3lame for the repo "ffmpeg".
[03:12] <chymist_chris> and now whenever I run ffmpeg it says this program is not developed anymore and is only provided for compatibility
[03:12] <pasteeater> yeah, that's an annoying phrase.
[03:13] <pasteeater> http://stackoverflow.com/a/9477756/1109017
[03:13] <pasteeater> there's a bug report in launchpad to get it changed, but i don't know what the status of it is
[03:14] <pasteeater> apparently a fix was released, but i don't know if that means it made it in precise or not
[04:48] <taqattack> Hello! I'm trying to read in the wav output from ffmpeg using stdin. Does anyone know what type of byte encoding ffmpeg outputs to console?
[07:58] <qrtt1> hello, I see the libavformat provide the function av_close_input_file. Does it provide av_close_output_file, too ?
[08:08] <grepper> trying to compile latest ffmpeg - I updated libvpx and installed to non-standard location, but can't get ffmpeg's configure to find it.  I set PKG_CONFIG_PATH to the pkgconfig dir of the libvpx install but no luck
[08:08] <ubitux> qrtt1: do you see any av.*output.* function?
[08:14] <grepper> PKG_CONFIG_PATH=/opt/ffmpeg-05292012/lib/pkgconfig/ pkg-config --modversion vpx
[08:14] <grepper> 1.1.0
[08:14] <grepper> dunno what the problem is
[08:37] <qrtt1> ubitux: thanks, I check it now
[08:43] <CruX|> hello all
[08:43] <CruX|> are there any ffmpeg static binaries vor linux64 ?
[08:43] <CruX|> i've found only this site https://sites.google.com/site/linuxencoding/builds
[08:43] <CruX|> and seems to be outdated
[09:03] <ubitux> CruX|: http://ffmpeg.gusari.org/static/ ?
[09:04] <CruX|> thx
[09:07] <qrtt1> ubitux: finally, I use the avio_close to close muxer file
[09:08] <ubitux> yep, there is no av.*output functions in the API, that was why i asked you to look for one :)
[09:09] <ubitux> the avio stuff is most likely what you were looking for
[09:09] <qrtt1> ubitux: thank you :)
[09:09] <ubitux> qrtt1: API examples are available in doc/examples
[09:09] <ubitux> you might want to have a look
[09:15] <qrtt1> ubitux: It's very important information :)
[09:16] <ubitux> http://git.videolan.org/?p=ffmpeg.git;a=tree;f=doc/examples;hb=HEAD
[10:09] <furan> jesus christ what's the deal with deprecation in ffmpeg? some of these are just #define values
[10:10] <furan> like someone renamed CODEC_TYPE_VIDEO because it made them feel good
[10:11] <ubitux> API consistency
[10:11] <ubitux> we use av/AV prefix for public stuff
[10:14] <furan> it's cleaner
[10:14] <furan> just a pain, heh
[10:15] <ubitux> yes, sorry about that :)
[10:19] <furan> that wasn't too bad, all working again
[10:19] <furan> now I can compile it on my arm device :D
[10:19] <ubitux> the CODEC_TYPE stuff is not the more painful thing to replace :)
[10:20] <furan> indeed
[13:44] <undercash> hi
[13:45] <undercash> SSR not implemented. Update your FFmpeg version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.
[13:45] <undercash> but it works fine and the file i m converting was previously converted using ffmpeg too
[13:56] <Keshl> Could anyone tell me why running: ffmpeg.exe -i "C:\ffmpeg\encode\frame_%03d.png" -vcodec libx264 -preset slow -crf 25 -r 60 -threads 0 "frame_x264.avi"   results in a video at 30 FPS, not 60? oÉo? (According to the metadata and actual playback speed)
[14:14] <LexSfX> Keshl: i'm not an expert, so don't consider this an answer; just a question.  does it still make a 30-fps video if you use ".mkv" instead of ".avi"?
[14:17] <Keshl> Yes, oÉo.
[14:17] <LexSfX> :o
[14:18] <LexSfX> what complete insanity
[14:18] <Keshl> Agreed. x.x
[14:18] <Keshl> According to the sites I poe at, -r 60 should do it.
[14:18] <Keshl> *poke
[14:18] <Keshl> And it /does/.... But only if I first convert the PNG's to an AVI, then convert the AVI to raw, then convert the raw /back/ to AVI and loose a crapload of detail.
[14:18] <Keshl> But PNG to .avi, it gets stuck at 30 FPS.
[14:18] <Keshl> maybe 25, can't tell.
[14:18] <LexSfX> let me see what armageddon-recorder does (uses ffmpeg to make 50 fps game video from Worms-Armageddon-output image sequences)
[14:19] <LexSfX> while i check, does it still make it 30 if you do -r 50?  maybe 60 is some magic number it chokes on
[14:20] <Keshl> Still does 30, oÉo.
[14:20] <Keshl> actually it's 25, I think. I have 900 frames and it's coming out to about 32 seconds, not 30.
[14:20] <Keshl> Maybe 25's some weird default somewhere it keeps using?
[14:20] <LexSfX> what about 120?  :P
[14:20] <LexSfX> oh
[14:21] <LexSfX> hehe, ya, maybe
[14:21] <Keshl> 120 would blow my TV up when it tries to play 3D video. <É<
[14:21] Action: Keshl does anyway.
[14:21] <Keshl> Still 25.
[14:21] <LexSfX> try 60/1
[14:21] <LexSfX> instead of 60
[14:22] <Keshl> (For the record, this is just two videos standing on top of each other (1280x1440 instead of 1280x720) -- As far as ffmpeg has to care, this is a normal run of the mill .avi, nothing special. It shows up fine, just plays horribly slow.)
[14:22] <Keshl> Tried that, 25.
[14:22] <LexSfX> hmm
[14:22] <Keshl> ...
[14:22] Action: Keshl does 120/2
[14:22] <Keshl> 25. x.x
[14:22] <Keshl> Tricking it doens't work. x.x
[15:00] <LexSfX> Keshl: armageddon-recorder has this: %06d.png
[15:00] <LexSfX> rather than %03d.png
[15:00] <LexSfX> i'm not sure if that makes a difference
[15:01] <LexSfX> but all i know is that's the only difference between your command and what armageddon-recorder is doing
[15:01] <LexSfX> oh, that number is just the number of digits
[15:01] <LexSfX> in the file name's frame number
[15:01] <LexSfX> hmm
[15:01] <LexSfX> so that couldn't possibly be it
[15:06] <LexSfX> Keshl: what happens if you put the "-r 60" immediately before or after the input path (the "-i [...]" part)?
[15:08] <JEEBsv> it used to be so that before -i it would override the frame rate of the input, after -i there is a frame rate conversion to the set rate (copying/removing frames)
[15:08] <JEEBsv> but now the first one doesn't work like that unless there are no specific timestamps off of the format
[15:08] <JEEBsv> the latter still works like it used to
[15:08] <LexSfX> well, the input has no rate in this case
[15:09] <LexSfX> so should -r 60 work if placed before "-i [...]"?
[15:09] <LexSfX> Keshl: that might be your answer.  try it!
[15:09] <JEEBsv> my answer would be "depends", because I remember getting herpy derpy results at times :|
[15:09] <LexSfX> oh yeah!  armageddon-recorder has the rate before the "-i [...]" part
[15:10] <LexSfX> so that's almost certainly it
[15:11] <LexSfX> i keep forgetting that the order of arguments actually matters for ffmpeg
[15:11] <LexSfX> i don't know any other CLI software for which that's true
[15:12] <LexSfX> well, libav excepted, of course
[15:12] <JEEBsv> the funniest thing is that -r isn't that straightforward any more >_> You can't use it as a blind AssumeFPS() because it will just error out in some cases
[15:12] <LexSfX> :x
[15:13] <LexSfX> sounds very messy
[16:11] <burek> is this a bug in latest ffmpeg or the output is damaged: http://pastebin.com/MitbF4kR
[16:11] <burek> I didn't see this kind of error before..
[16:41] <jo_le_rigolo> Hi there
[16:43] <jo_le_rigolo> small question about 0.11 rel: where can I find CVE information ?
[16:44] <jo_le_rigolo> chanelog only stats CVE numbers and NVD db doesn't have anything, all CVE are candidate without any useful information
[16:54] <eok> i'm hoping someone could help me with a problem. i have a ffmpeg config that outputs to vlc for streaming
[16:55] <eok> i'm using theora and vorbis and encoding in ogg
[16:55] <eok> here is the config and the error message i get
[16:55] <eok> http://codepad.org/HXtyTnwV
[16:56] <eok> if i output to a file it seems to work somewhat, i dont get the overlay, but everything else seems to be as it should
[17:06] <burek> eok
[17:06] <burek> can you please use pastebin.com, to show your command line and its output?
[17:07] <burek> jo_le_rigolo, cve?
[17:08] <eok> burek, sure
[17:08] <jo_le_rigolo> Common Vulnerabilities and Exposure
[17:08] <ubitux> burek: common vuln exposure
[17:08] <burek> why do you need that?
[17:08] <burek> jo_le_rigolo, ?
[17:08] <AnywhereIs> already stop hiliting me and look @ sources
[17:08] <eok> http://pastebin.com/i1jQBuXk
[17:08] <ubitux> to exploit them in libav obviously
[17:09] <ubitux> ;)
[17:09] <burek> eok, uncut output
[17:09] <burek> not bits and peaces
[17:10] <jo_le_rigolo> on the contrary ubitux, i need that for risk assessment ^
[17:10] <burek> well, read the source code then? :)
[17:10] <burek> all I care about is that if security bug was found, it has been fixed :) that's all :)
[17:11] <jo_le_rigolo> let's call that plan B
[17:11] <AnywhereIs> what a long words
[17:11] <AnywhereIs> you know how to use language, right?
[17:11] <AnywhereIs> ROFL
[17:11] <ubitux> jo_le_rigolo: git log -p --grep j00ru
[17:11] <ubitux> most of the recent security fixes come from this
[17:11] <AnywhereIs> kids today...
[17:12] <burek> why would anyone spend time on that without some obvious benefit?
[17:12] <eok> burek, ok here you go: http://pastebin.com/1bCky0AA
[17:12] <eok> dropped the overlay stuff, it's not relevant to the problem
[17:13] <burek> is ogg format "streamable" ?
[17:13] <ubitux> of course
[17:13] <burek> it supports seeking backwards and stuff?
[17:14] <burek> I mean, I didn't see a lot of cases that use ogg when streaming, that's why I ask
[17:14] <ubitux> i use ogg for my personal radio
[17:14] <eok> i sort of have to use ogg, i plan on testing streaming in html5
[17:15] <burek> eok, I think you should google for "[libvorbis @ 0xa6ad300] Que input is backward in time"
[17:15] <burek> what does html5 have to do with ogg?
[17:15] <eok> actually haven't noticed it :)
[17:16] <eok> http://en.wikipedia.org/wiki/HTML5_video
[17:16] <eok> Formats supported by different web browsers
[17:16] <burek> :)
[17:16] <burek> well
[17:16] <burek> what does that have to do with anything at all when you are remuxing it in vlc into mpegts?
[17:17] <eok> ohh lol, did i forget to change it?
[17:17] <burek> :facepalm: :)
[17:17] <eok> then i copy pasted wrong
[17:17] <eok> still gives the same error though
[17:17] <eok> 2 sec
[17:17] <burek> ok :)
[17:17] <burek> try using -f mpegts instead of -f ogg
[17:18] <burek> and in vlc, choose ogg as a final format
[17:18] <eok> http://pastebin.com/u8tXYjhn
[17:18] <burek> still "Que input is backward in time"
[17:19] <burek> your error happens in ffmpeg not vlc
[17:19] <eok> i've tried it before.. ill test again
[17:20] <eok> hmm the "Que input is backward in time"
[17:20] <eok> is still there
[17:20] <eok> but the other errors are gone
[17:20] <eok> the stream wont start though
[17:21] <eok> dead input
[17:21] <eok> cannot pre fill buffer
[17:21] <burek> read more above, there has to be an error that caused cannot prefill buffer
[17:22] <eok> http://pastebin.com/AQwh6mb8
[17:24] <burek> eok, could you use named pipes
[17:24] <burek> mkfifo
[17:24] <burek> and stuff
[17:24] <burek> just to separate the outputs
[17:24] <burek> because I'm not sure who is spawning errors
[17:24] <burek> vlc or ffmpeg
[17:24] <burek> mkfifo /tmp/fifo1
[17:25] <burek> ffmpeg -i ... /tmp/fifo1
[17:25] <burek> vlc /tmp/fifo1 ...
[17:26] <eok> i haven't done that before, sorry for being a noob, but how do i write that in the command?
[17:28] <burek> well, you are actually splitting your pipe into 2 commands
[17:28] <burek> you forward your output from the first command into a special file (named pipe)
[17:28] <burek> and from the 2nd terminal, you take the input from that special file
[17:29] <burek> the effect is the same as if using just 1 command with pipe |
[17:29] <eok> made fifo1, and added -i /tmp/fifo1 to ffmpeg and /tmp/fifo1 to vlc
[17:29] <eok> is that right=
[17:29] <eok> ?
[17:29] <burek> well -i /tmp/fifo1 won't help
[17:29] <burek> because you need to send output from ffmpeg to that fifo
[17:30] <burek> ffmpeg -f alsa -ac 2 -i pulse -f x11grab -s 1680x1050 -r 25 -i :0.0+0,0 -vcodec libtheora -pix_fmt yuv420p -vf yadif,scale=1280:720 -r 20 -acodec libvorbis -ab 160000 -g 30 -threads 0 -f mpegts /tmp/fifo1
[17:32] <eok> hmm
[17:33] <burek> think of it as if you would replace the "|" sign with a file
[17:33] <eok> i dont think that command works.. wouldnt that send the actual data to fifo1?
[17:33] <eok> and not the output
[17:34] <burek> it would send the data to fifo
[17:35] <burek> and vlc would then read that data from the fifo
[17:35] <burek> but the effect would be that you run both commands in their own terminal
[17:35] <burek> having their error logs separated
[17:35] <burek> replace your original commands with 2 cmds
[17:36] <burek> ffmpeg -f alsa -ac 2 -i pulse -f x11grab -s 1680x1050 -r 25 -i :0.0+0,0 -vcodec libtheora -pix_fmt yuv420p -vf yadif,scale=1280:720 -r 20 -acodec libvorbis -ab 160000 -g 30 -threads 0 -f mpegts /tmp/fifo1
[17:36] <burek> and
[17:36] <burek> vlc -I dummy /tmp/fifo1 --sout '#std{access=http,mux=ogg,dst=:8080}'
[17:36] <burek> got it now?
[17:38] <eok> fuuu
[17:38] <eok> now i get these PTS DTS errors spamming
[17:38] <eok> 2 sec
[17:39] <eok> is this how you wanted it?
[17:39] <eok> http://pastebin.com/HCJLYz7j
[17:40] <eok> samn
[17:40] <eok> damn
[17:40] <eok> now the mus=ts is there again, lol
[17:41] <eok> there: http://pastebin.com/hW88KKnN
[17:43] <burek> eok, do you know how to run 2 terminals?
[17:45] <eok> not sure what you mean
[17:45] <burek> the black little window in which you type your commands :)
[17:45] <eok> yea, i have many terminals :D
[17:45] <burek> great :) that's cool :D
[17:45] <eok> hehe
[17:45] <burek> now
[17:45] <eok> what do you want me to do?
[17:45] <burek> run each command in its own terminal
[17:46] <burek> not in 1 cmd
[17:46] <burek> ffmpeg in 1 terminal, vlc in another terminal
[17:46] <burek> ok?
[17:46] <eok> uhm, but how would ffmpeg give anything to vlc if i dont pipe it to vlc?
[17:47] <burek> :)
[17:47] <burek> now, sit back and relax and read about what I've been telling you for a whole time :)
[17:47] <sacarasc> eok: A fifo, or named pipe, is piping it to vlc.
[17:48] <eok> ahh
[17:48] <eok> i think i get it, lol
[17:48] <eok> this is really new to me
[17:48] <eok> i'm like stumbling in the dark!
[17:48] <burek> :)
[17:49] <eok> getting these buggers
[17:49] <eok> DTS 10206755, next:10208114 st:0 invalid dropping
[17:49] <eok> PTS 10206755, next:10208114 invalid dropping st:0
[17:49] <eok> spamming up my terminal
[17:49] <burek> in ffmpeg or vlc?
[17:49] <eok> ffmpeg
[17:50] <burek> now, that's what we need
[17:50] <burek> can you now paste the ffmpeg's only output to pastebin
[17:50] <eok> get those something, but they usually dont affect anything
[17:50] <eok> need to get the output to a file
[17:50] <burek> ffmpeg -f alsa -ac 2 -i pulse -f x11grab -s 1680x1050 -r 25 -i :0.0+0,0 -vcodec libtheora -pix_fmt yuv420p -vf yadif,scale=1280:720 -r 20 -acodec libvorbis -ab 160000 -g 30 -threads 0 -f mpegts /tmp/fifo1 2>file.txt
[17:51] <eok> just tried it :)
[17:52] <eok> hmm need to type yes
[17:52] <eok> lol
[17:52] <burek> add -y
[17:52] <burek> ffmpeg -y -f ...
[17:53] <eok> there we go
[17:53] <eok> hold on i'll get the outputs for you, and thanks for you patience :)
[17:54] <burek> ok :)
[17:55] <eok> ffmpeg: http://pastebin.com/6mQsQEw5
[17:56] <eok> vlc: http://pastebin.com/hvhTv0VM
[18:02] <burek> hmh
[18:02] <burek> try adding -vvv to vlc's options
[18:02] <burek> to get more verbose output
[18:02] <burek> in ffmpeg, only 1 warning is there: Que input...
[18:03] <eok> ok
[18:03] <eok> vlc:http://pastebin.com/mRHrSHLF
[18:04] <eok> ffmpeg: http://pastebin.com/JtEb1KNj
[18:06] <eok> ffs
[18:06] <eok> ohh nvm :P
[18:06] <Know1edge> anyone know if its possible to generate a waveform of an audio file with ffmpeg?
[18:07] <saste> Know1edge: check "showwaves" in the ffmpeg-devel archive
[18:07] <saste> wip
[18:08] <Know1edge> thanks :D
[18:09] Action: ShinyObjects high fives burek
[18:10] <ShinyObjects> I have "beer" on my word watchlist. Over the weekend you mentioned beer and when I got back pidgin was very excited to tell me that.
[18:11] <eok> i thought for a sec i had used mux=ts
[18:12] <eok> because of this: main demux debug: using demux module "ts"
[18:12] <eok> but i have mux=ogg
[18:13] <eok> its demuxing mpegts from ffmpeg off course.. im stupid :(
[19:12] <burek> eok, this might be a problem "VLC media player 1.0.6"
[19:12] <burek> try updating your vlc
[19:19] <burek> what is a proper way to buffer the input a little bit in ffmpeg, for example to avoid network issues when streaming and stuff
[19:20] <burek> abuffer is doubled in html documentation
[19:20] <burek> http://ffmpeg.org/ffmpeg.html
[19:20] <burek> abuffersink too
[19:41] <rschmidt> hi guys, I'm trying to debug an issue with a freshly-compiled version of ffmpeg (built from latest git sources)
[19:42] <rschmidt> I'm on Linux (CentOS 6), trying to read video + audio data from a PVR located at /dev/video0, and output an FLV file
[19:43] <eok> lhttp://pastebin.com/mRHrSHLF
[19:43] <rschmidt> the encoding starts, but encodes a few hundred frames (about 300 - 350) and then hangs
[19:43] <rschmidt> however, doing a copy (no reencode) doesn't hang.
[20:01] <eok> burek: here is something to look at http://paste.ubuntu.com/1013403/
[20:02] <eok> just tried as basic as possible using only x11grab with libtheora and mpegts format
[20:02] <eok> why are there so many duplicates?
[20:03] <eok> vlc output when i try playing the file: http://paste.ubuntu.com/1013411/
[20:22] <eok> burek: figured it out :) i can't use mpegts because libtheora can't be contained by it. when using ogg it gets borked for some reason
[20:22] <eok> burek: tried matroska and that worked!
[22:22] <dericed> I have 24 bit wav files to encode to flac. ffmpeg converts to 16 bit when using 'ffmpeg -i wav.wav flac.flac'. Does ffmpeg support 24 bit flac encoding? How can I preserve the bit depth?
[22:24] <burek> eok, great :)
[22:25] <burek> eok, btw how did you figure out libtheora cannot go together with mpegts?
[22:25] <burek> dericed, did you check audio filters?
[22:26] <dericed> i just see "auto-inserting filter 'auto-inserted resampler 0' between the filter 'src' and the filter 'aformat'
[22:26] <dericed> [aresample @ 0x111c19920] r:44100Hz -> r:44100Hz"
[22:26] <burek> rschmidt,
[22:26] <burek> can you please use pastebin.com, to show your command line and its output?
[22:26] <burek> dericed,
[22:26] <rschmidt> sure thing, burek
[22:27] <burek> can you too use pastebin
[22:31] <dericed> burek, thanks: http://pastebin.com/kGpT7fPj
[22:33] <burek> dericed, this might be related: http://ffmpeg.org/trac/ffmpeg/ticket/210
[22:36] <dericed> burek, probably. I just ran with a ffmpeg compiled on 7/12/11 to see if it was a regression. That version still output a 16 bit flac, but at least gives a warning: "Convertion will not be lossless'"
[22:37] <dericed> burek: the flac utility supports 24 bit encoding. I'll go there for now.
[22:38] <rschmidt> burek: http://pastebin.com/fdynLB9g
[22:44] <burek> dericed, I just asked devels, ffmpeg does not support 24bit encoding
[22:44] <burek> only decoding
[22:44] <burek> :/
[22:44] <JEEB> yes, only libflac / the official flac encoder supports >16bit
[22:45] <burek> rschmidt, I'm afraid you can't use -i /dev/video0 directly, because ffmpeg figures it's an mpegts format
[22:45] <burek> you're maybe better off using video4linux2
[22:45] <burek> try this instead:
[22:46] <burek> ffmpeg -y -f v4l2 -i /dev/video0 -s 360x240 -r 30 -vcodec libx264 -acodec aac -strict experimental -ar 44100 /home/impact/streaming.flv
[22:46] <rschmidt> burek: aha... yeah, I tried v4l2, but the hdpvr drivers don't support the streaming I/O method
[22:46] <dericed> burek  :-/ thanks for checking. I added a message to the ticket to say that the non-lossless warning no longer works, though it did last year
[22:46] <burek> try googling for ffmpeg video4linux2 Hauppauge hdpvr
[22:46] <rschmidt> burek: but thanks for the input... I'll keep poking around
[22:47] <burek> ok :)
[23:11] <ashka> hello
[23:11] <ashka> I'm trying to extract a thumbnail from a video
[23:13] <ashka> but it get stuck at the encoding
[23:13] <ashka> I'm using this command : ffmpeg -i Benjamin_Bayart_25092009_soir.flv -vframes 1 -an -r 1 -s 231x106 -ss 3554 Benjamin_Bayart_25092009_soir.flv.jpg
[23:13] <ashka> and it get stuck at frame=    0 fps=  0 q=0.0 size=      -0kB time=00:00:00.00 bitrate=   0.0kbits/s
[23:19] <vgrimmer> hello, I am trying to compile ffmpeg on Mac with x264 support and I ran into some trouble.
[23:19] <vgrimmer> ./configure tells me that ERROR: libx264 not found
[23:20] <vgrimmer> g++ -Lshared/lib -Wl,-dynamic,-search_paths_first -o /var/folders/h8/4mwpk_k5411818h22qlk650w0000gn/T//ffconf.r4xgjS6W /var/folders/h8/4mwpk_k5411818h22qlk650w0000gn/T//ffconf.iG9PeUzs.o -lx264 -framework CoreFoundation -framework VideoDecodeAcceleration -framework QuartzCore -lbz2
[23:20] <vgrimmer> Undefined symbols for architecture x86_64:
[23:20] <vgrimmer>   "x264_encoder_encode()", referenced from:
[23:20] <vgrimmer>       _main in ffconf.iG9PeUzs.o
[23:20] <vgrimmer> ld: symbol(s) not found for architecture x86_64
[23:20] <vgrimmer> collect2: ld returned 1 exit status
[23:20] <vgrimmer> ERROR: libx264 not found
[23:20] <vgrimmer> does anyone know how to solve this problem?
[23:21] <vgrimmer> libx264.a and libx264.dylib are both in the shared/lib folder
[23:23] <ashka> vgrimmer: maybe you don't have the proper libs for your architecture
[23:23] <ashka> I don't know much about compilation things
[23:24] <vgrimmer> I thought about that too, but the x264 lib is compiled right before I compile ffmpeg
[23:25] <vgrimmer> ffmpeg will complain for a mismatch of the x264 dylib if I change my configure from --cc="g++ -m64" to --cc="g++ -m32"
[23:26] <vgrimmer> so I thought they are both for the x86_64 architecture
[23:26] <ashka> the command I pasted before finally worked, it just took a very long time, but it stretches the image which I don't want
[23:31] <RobertNagy> are the AVPackets return from e.g. avcodec_encode_video2 safe between calls?
[23:31] <RobertNagy> or do I need to call av_dup_packet? the same as for av_read_frame?
[23:32] <ashka> I have another problem, I'd like to apply a filter
[23:33] <ashka> but it says unrecognized option -filter
[23:33] <pasteeater> ashka: you can try moving -ss as an input option. it will be much faster, but potentially less accurate.
[23:33] <ashka> accuracy doesn't really mind
[23:33] <pasteeater> ashka: try -vf or -filter:v
[23:33] <pasteeater> for video filtering
[23:33] <ashka> okay
[23:33] <eok> burek: i know wikipedia shouln't be a source but it says there it can be contained by matroska and ogg
[23:33] <eok> http://en.wikipedia.org/wiki/Theora
[23:33] <ashka> btw, what do you mean by moving it as an input option ?
[23:33] <ashka> I'm quite a noob with ffmpeg
[23:34] <pasteeater> before -i
[23:34] <pasteeater> the -an and -r 1 are not needed
[23:34] <ashka> okay
[23:34] <pasteeater> as for stretching you can "-vf scale=231:-1"
[23:35] <pasteeater> it will preserve aspect ratio
[23:35] <ashka> hmm, it's still flattened
[23:36] <ashka> it tries to fit to 231x106 but it flatten the thumb
[23:37] <pasteeater> did you try another viewer?
[23:37] <ashka> for the image ? yep
[23:37] <pasteeater> and use a pastebin site to show your ffmpeg command and the complete console output
[23:38] <ashka> btw, where do I place the scale filter ?
[23:38] <pasteeater> usually as an output option (after -i input.foo)
[23:38] <pasteeater> can you upload the image somewhere?
[23:38] <eok> burek: i've given up on theora, i'm gonna try vp8 and webm to stream in html5
[23:39] <eok> burek: i suspect that i will give up on html5 all together, but it's worth a try. I suspect i can't get webm with webgl to work on firefox
[23:59] <Keshl> ffmpeg is beating me up. .É.
[23:59] <Keshl> (In other words, I can't read the documentation. At all. Ho do I get cellauto to actually work? <É<)
[23:59] <Keshl> (*How)
[00:00] --- Wed May 30 2012


More information about the Ffmpeg-devel-irc mailing list