[Ffmpeg-devel-irc] ffmpeg.log.20120804

burek burek021 at gmail.com
Sun Aug 5 02:05:01 CEST 2012


[00:22] <melkor> Hello, I am trying to rotate a movie and I get an Error while opening encoder for output stream #0.0 - maybe incorrect parameters such as bit_rate, rate, width or height
[00:25] <llogan> otherwise i can try to guess
[00:25] <melkor> Cool, Ill post it.
[00:28] <melkor> http://pastie.org/4386264
[00:31] <melkor> I got it, I had to set the audio rate. Now lets see if I can get it t stop at 2:30
[00:32] <llogan> melkor: -t 00:02:30
[00:34] <melkor> thanks
[00:40] <llogan> melkor: you're not using ffmpeg from FFmpeg by the way.
[00:55] <melkor> llogan: I realize that, but I just wanted to do this quick.
[00:55] <llogan> melkor: you should copy the audio instead of re-encode
[00:56] <llogan> -c:a copy
[00:56] <llogan> i'm not sure of the syntax for ffake ffmpeg though
[00:57] <melkor> I didn't know it could be done, actually the audio sounds a little better after being reencoded.
[00:57] <llogan> weird
[00:57] <melkor> Maybe that is just on these headphones.
[01:15] <trysten> thanks llogan
[01:19] <VooDooNOFX_> I've been googling for some time now, but how can I turn a pile of (DPX|TIFF) images into a dnxhd?
[01:20] <VooDooNOFX_> "-i /tmp/*.dpx" is creating a .mov with a single frame, not a sequence of frames
[01:23] <llogan> VooDooNOFX_: http://ffmpeg.org/faq.html#How-do-I-encode-single-pictures-into-movies_003f
[01:24] <VooDooNOFX_> llogan, so the answer is to pack them into a mpg first?
[01:24] <llogan> no. that's just what the example uses.
[01:26] <llogan> also see this patch that i never applied: http://pastebin.com/nmRqZeb5
[01:29] <llogan> VooDooNOFX_: does it make sense?
[01:30] <VooDooNOFX_> llogan, interesting. So you're saying there's an option now that's -start_number, or that you never enabled -start_number
[01:30] <llogan> no. the option is there. the patch is just for updating the faq that i linked to
[01:30] <llogan> sorry, that wasn't very clear
[01:31] <VooDooNOFX_> ah, cool! I was having to renumber frames in nuke to 0000, but with this, I can start at the normal 1001
[01:34] <VooDooNOFX_> llogan, in my case, substituting %04d for the input worked instead of the glob pattern *.dpx. However, i've got another question about the compression settings in the dnxhd module. There appears to be a large amount of compression artifacts in the encode
[01:34] <VooDooNOFX_> are there flags to control how that's processed?
[01:36] <llogan> i'm ignorant of dnxhd but here's some old info: http://www.itbroadcastanddigitalcinema.com/ffmpeg_howto.html#Encoding_VC-3
[01:37] <llogan> VooDooNOFX_: why dnxhd?
[01:38] <VooDooNOFX_> it's being requested by my work for vfx previews
[01:42] <VooDooNOFX_> llogan, when was start_number added? I'm using 0.10
[01:44] <llogan> 24 May
[01:44] <VooDooNOFX_> ah, min eis from mar 5.
[01:45] <VooDooNOFX_> np, time to rebuild it anyway.
[01:45] <llogan> what distro?
[01:45] <VooDooNOFX_> centos 5.7, build from source
[01:45] <llogan> https://ffmpeg.org/trac/ffmpeg/wiki/CentosCompilationGuide
[01:45] <llogan> for 6 but should work i guess
[01:46] Action: llogan needs to finish the "local install" version
[01:47] <llogan> you can probably jsut use zlib from the repo. someone else added the zlib compile seciton. seems unnecessary to me.
[01:57] <VooDooNOFX_> llogan, thank you for your help. I'll try building the latest and see if that helps the compression artifacts also.
[05:22] <powerNET> i'm getting some issues trying to stream to a file to rtmp server
[05:22] <powerNET> http://pastebin.com/w2Hsb2Dh
[10:23] <fturco> Hello. Is there a way to apply a "smart" deinterlace filter in ffmpeg? That is, a deinterlace filter that is able to automatically detect if video is interlaced or not, or even if single frames are.
[10:24] <fturco> There is one in Avidemux, for example.
[10:24] <saste> fturco: what about idet?
[10:25] <fturco> Never heard of it. I'll try. Thank you saste
[10:26] <fturco> saste, I even heard there are videos that have partially interlaced frames. Is this true?
[10:26] <JEEBsv> not really
[10:26] <saste> fturco: have no idea
[10:27] <JEEBsv> the "frame" either is interlaced or not, although with interlaced video one deals with fields
[10:27] <JEEBsv> of course, what is within those two fields is a completely separate issue
[10:29] <fturco> The problem is I record from TV, and TV programs use interlaced/progressive material almost at "random"
[10:29] <fturco> I cannot manually check every time
[10:30] <JEEBsv> unelss you have a veryt lulzy TV station that actually switches between progressive and interlaced encoding, you're talking about "true interlacing" vs telecine
[10:31] <JEEBsv> personally I would just do encoding manually and cut off commercials etc. that are unneeded/most possibly truly interlaced
[10:31] <JEEBsv> but if you want to automate
[10:32] <JEEBsv> yadif or handbrake's decomb filter are probably the best you can get on something that is cross-platform
[10:32] <JEEBsv> (not to be mistaken with the Avisynth decomb filter, which is an inverse telecine filter)
[10:33] <fturco> JEEBsv, so you recommend to keep the original TV recording as is (without commercials of course), and not to apply additional filters
[10:33] <JEEBsv> both ways will most probably leave you the extra "frames" in telecined content, but at least you should have no combing (the usual artifact when fields contain different pictures
[10:33] <JEEBsv> fturco: uhh, no
[10:34] <JEEBsv> I meant, that I would just cut off commercials in something frame-exact, and would then deal with the content accordingly to what it happens to be
[10:36] <fturco> I just checked a video, and it seems only a portion of the screen in a certain scene has combing effect, while the rest seems progressive
[10:39] <JEEBsv> that means that the two fields were taken at different times
[10:39] <JEEBsv> anyways, unless you want to do it Correct (TM), I recommend those two filters I mentioned
[10:40] <JEEBsv> also, "doing it Correct (TM)" means using certain software that you need to use through wine on *nix, unfortunately. Since the replacement isn't done yet
[10:41] <fturco> doing it correct (tm) consists of? what win software?
[10:43] <JEEBsv> Avisynth and its filters, that are really the only thing with which you can do it "correctly"
[10:43] <JEEBsv> (Esp. with commercial cutting etc.)
[10:43] <JEEBsv> it has worked on wine since 2003 or so though
[10:45] <saste> JEEBsv: file feature requests for filters you want ported
[10:45] <saste> we have a socis student allocated exactly for that
[10:47] <JEEBsv> k
[10:49] <JEEBsv> will do at some point, although it's mostly about getting frame-exact frame-based cutting and joining, as well as all kinds of other stuff that ffmpeg isn't really usually meant for
[11:34] <Eraindil> chaos, I have a question about recording
[11:34] <Eraindil> I would like to make a screencast and record all audio input output
[11:34] <Eraindil> I use alsa
[11:35] <Eraindil> Have you got an idea?
[11:48] <fturco> JEEBsv, it seems there is a port of avisynth for linux: http://code.google.com/p/avxsynth/
[14:26] <deuxbierre> Bonjour
[14:26] <deuxbierre> http://paste.ubuntu-fr-secours.org/src-108403
[16:11] <deuxbierre>  scale=220:176 -vcodec mpeg4 -b:v 1M  -acodec mp2 * -ac 2 -ar 44100 -ab 64k
[16:11] <deuxbierre> [NULL @ 0x9b0a2c0] Unable to find a suitable output format for 'scale=220:176'
[16:11] <deuxbierre> scale=220:176: Invalid argument
[16:15] <benpro> Trying to compile ffmpeg-git with libx264 support. I got libavcodec/libavcodec.so: undefined reference to `x264_encoder_open_125'. libx264.a is at /usr/local/lib/libx264.a. I don't know why ld fail. Any idea ? Need I set LD_LIBRARY_PATH var ?
[16:16] <benpro> same with LD_LIBRARY_PATH, mhhh
[16:29] <burek> benpro
[16:29] <burek> probably you have some x264 installed with your package manager
[16:29] <burek> which conflicts with your compiled one
[16:30] <burek> try uninstalling your x264 first and then compile it from source
[16:30] <burek> you have some instructions on wiki
[16:32] <benpro> YEah seems to be that
[16:33] <benpro> I would like to not uninstall official repo
[16:33] <benpro> When running make, libavodec.so si linked to ibx264.so.124 => /lib/libx264.so.124 (official repo)
[16:33] <burek> of course, system libs take precedence over user compiled libs
[16:34] <benpro> Why LD choose /lib/libx264.so when I specify PKGCONFIG_PATH to /usr/local/lib
[16:34] <burek> dunno :)
[16:34] <benpro> mhhh
[16:34] <burek> try google :)
[16:34] <benpro> Yeah but not easy to type the good keywords :P
[16:35] <burek> or http://www.linuxquestions.org/questions/slackware-14/installing-two-versions-of-one-package-904595/
[16:35] <burek> :)
[16:36] <benpro> Ok, I'll search system library conflicts, seems to be the case
[16:38] <benpro> Or I will not keep offical ffmpeg ... tired of resolving this case
[16:38] <burek> why do you keep the system installed x264/ffmpeg?
[16:38] <burek> they are in 99% outdated a lot
[16:38] <benpro> Well not to outdated
[16:38] <burek> do you have a specific need for it?
[16:39] <benpro> I use archlinux
[16:39] <benpro> I would like to keep "stable" things, and my compile in /usr/local :)
[16:39] <burek> well, just take a look how many commits there are per day in ffmpeg's git
[16:39] <burek> and you tell me if it's outdated or not :)
[16:39] <benpro> ffmpeg 1:0.11.1-1 for achlinux
[16:39] <burek> what date?
[16:39] <benpro> but not ffmpeg-mt BTW
[16:40] <burek> !dnld
[16:40] <benpro> 2012-07-08
[16:40] <burek> 2012-06-07
[16:40] <burek> 2 months ago
[16:41] <burek> it gets even worse
[16:41] <burek> 0.11.1 was released on 2012-06-07. It is the latest stable FFmpeg release from the 0.11 release branch, which was cut from master on 2012-05-25
[16:41] <benpro> Well just uninstalled it
[16:41] <burek> do as you like, I suggest you always use the latest git version
[16:42] <burek> and when you want to update, just: make uninstall && make clean && git pull && make && make install
[16:42] <benpro> Im recompile it, but I failed in my PKGBUILD and have rm -rf src, redownloading the git sources with my slowpoke connection ...
[16:42] <burek> use --max-depth=1
[16:42] <burek> or --max-level=1
[16:42] <burek> man git
[16:42] <benpro> Yep good idea !
[16:43] <benpro> but sometimes, this fails, dunno why
[16:43] <burek> beer never fails :)
[16:44] <benpro> In fact already use --depth 1
[17:05] <benpro> and, compilation done \o/ Thx for the help and tips burek.
[19:05] <burek> benpro :beer: :)
[19:08] <benpro> :)
[19:20] <ratbert90> hey, would anybody be around that has dealt with grabbing a test picture from /dev/video0?  I am trying to do some hardware verification on a camera and I am not too sure where to go from here.  This is a embedded system.  The camera on the o-scope checks out just fine for voltages and clock readout, so I am thinking it's more of a v4l2 issue
[19:21] <ratbert90> whenever I run ffmpeg -f video4linux2 -i /dev/video0 -pix_fmt yuvj420p -vframes 1 -vsync 2 -bufsize 1843200 test.jpeg It spits back out at me a Frame Dropped! message and that The v4l2 frame is 2073600 bytes, but 1843200 bytes are expected
[19:22] <ratbert90> This is a omnivsion OV2715 camera that uses 10bit bayer
[20:11] <benpro> ratbert90: read frist 10 frames but get only first one ?
[20:17] <ratbert90> I just want to get the first frame from the camera
[20:18] <ratbert90> heck,  I don't care if it's a garbled mess, it has no optics on it right now, it would be good enough for hardware verification
[20:18] <ratbert90> I was digging around in DM365_IPIPE.c (TI dm368 board is what I am using.) and it seems like ipipe may not like bayer.
[20:22] <burek> ratbert90
[20:22] <burek> remove -vsync and -bufsize
[20:22] <ratbert90> good idea, that way I can show you the errors it gives with those removed. :)
[20:22] <burek> use the pastebin please
[20:23] <ratbert90> of course
[20:26] <ratbert90> http://pastebin.com/W0N3mHcT
[20:29] <ratbert90> we have a lot of printk's going for some more verboseness :)
[21:00] <pyoor> hi all.  is there a way to use the ffmpeg command line to parse a file and output the type of codec used?  I tried doing "ffmpeg -i foo.mov | grep Stream" but I was curious if there was a better way
[21:00] <pyoor> I'd like to batch it out so that I can parse of a directory of media files and determine which are h234, jpeg2000, etc.
[21:03] <pyoor> h264*
[21:06] <microchip_> pyoor: try ffprobe
[21:10] <pyoor> great thanks microchip_
[21:55] <anon^_^> Hi, I have a question regarding invalid time/duration being reported by a MPEG-TS file.  Is there a way to correct this with FFmpeg or some other linux utility?
[23:03] <Foxhoundz> does ffmpeg support Window Media Center's WTV format?
[23:04] <Foxhoundz> Well, WTV is the extension, I don't know how it's encoded
[23:04] <Foxhoundz> I recorded a video using Windows Media Center from my TV Tuner and wish to convert it to a more friendly AVI format
[23:05] <Foxhoundz> or perhaps even MP4
[23:05] <sacarasc> Try playing it with ffplay. If it works, it's supported!
[23:07] <Foxhoundz> it plays it
[23:07] <Foxhoundz> still, ffmpeg fails to convert it
[23:07] <Foxhoundz> Also, ffplay seems to show weird scan lines when it's playing it
[23:07] <Foxhoundz> and stutters
[23:16] <spaam> Foxhoundz: what does ffmpeg -i file.wtv say?
[23:22] <Foxhoundz> spaam, http://pastie.org/4391135
[23:22] <Foxhoundz> I had to manually copy the contents because it wasn't writing the outputs to a file
[23:25] <spaam> from the .wtv file to somewhere else?
[23:30] <Foxhoundz> spaam, no I  mean ffmpeg's stdout output
[23:30] <Foxhoundz> Anyway, it gives me an error when attempting to convert
[23:31] <Foxhoundz> This was the error: http://pastie.org/4391161
[23:31] <Foxhoundz> I tried to use the -sameq flag
[23:31] <Foxhoundz> but it still keeps giving me the errors
[23:31] <Foxhoundz> and I usually don't have to supply bitrate arguments to ffmpeg
[23:32] <spaam> you want to use h264?
[23:32] <Foxhoundz> Yes, yes I do.
[23:32] <spaam> ffmoeg -i file.wtv  -vcodec libx264 -crf 22 -preset slow file.mp4
[23:33] <spaam> try that
[23:35] <Foxhoundz> Failed
[23:35] <Foxhoundz> ogg seems to work
[23:35] <spaam> ah you want to add -acodec libvo_aacenc -ab 128k
[23:54] <burek> Foxhoundz
[23:55] <Foxhoundz> burek, yes?
[23:55] <burek> next time instead of just " >file.txt "
[23:55] <burek> use 2>file.txt
[23:55] <Foxhoundz> >>?
[23:55] <Foxhoundz> Hmm
[23:55] <Foxhoundz> Why?
[23:55] <burek> because ffmpeg writes to stderr
[23:55] <burek> not stdout
[23:55] <Foxhoundz> Ahhh
[23:55] <burek> stdout is reserved if you want to use it to output file there (to pipe it somewhere else)
[23:55] <burek> to write the output file*
[23:56] <burek> for ex. ffmpeg -i input -acodec .. -vcodec .. - | ffplay -
[23:56] <burek> anyway, just use something like ffmpeg -i ... 2>log.txt
[23:57] <burek> ok, your ffmpeg is up-to-date
[23:58] <burek> US:Serbia wolleyball :) nice :)
[23:59] <burek> oh, your audio is a problem
[23:59] <burek> 48k is probably not supported, so try forcing 44.1k with -ar 44100
[00:00] --- Sun Aug  5 2012


More information about the Ffmpeg-devel-irc mailing list