[Ffmpeg-devel-irc] ffmpeg.log.20130102

burek burek021 at gmail.com
Thu Jan 3 02:05:01 CET 2013


[00:00] <JEEB> I don't know random people's random shit
[00:00] <misterno> apparently there is option CRF enable or disable
[00:00] <JEEB> in that guy's random GUI shit
[00:00] <JEEB> with x264 you pick a rate control mode
[00:00] <JEEB> it's not an on/off switch
[00:01] <JEEB> it's either constant quant (for development only), constant rate factor (crf) or average bit rate (bitrate-based)
[00:01] <misterno> interesting comparing filesize
[00:01] <JEEB> well you have NO FUCKING IDEA WHAT HE IS COMPARING AGAINST
[00:01] <JEEB> that thing makes my head hurt so much so I just fucking ask you to forget about it
[00:02] <misterno> alright don't get mad :)
[00:02] <klaxa> since he has "[X] CRF (overrides video bitrate settings)" in the GUI i'd assume the "fallback" is bitrate based encoding
[00:02] <klaxa> but yeah still retarded not using radio buttons or some shit
[00:02] <JEEB> with the same settings, 1pass crf and 2pass average bit rate should give the same visual quality level, except for the fact that bit rate distribution can end up different
[00:02] <JEEB> (with the same end bit rate)
[00:03] <JEEB> the latter part gets really important with really, really short samples
[00:03] <misterno> is this guide good http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[00:03] <JEEB> because 2pass generally tends to do worse there
[00:04] <JEEB> in general that guide goes over whatever I was saying as well
[00:04] <JEEB> 1) deal with your source (deinterlacing/IVTC, cropping, possible resize)
[00:04] <JEEB> 2) use -ss to seek to some part of the clip where you know it should be like it is generally in the source, and set -t to like 10-15 seconds
[00:05] <JEEB> 3) encode with some crf, starting from 23 which is the default
[00:05] <misterno> wait wait, possible resize?
[00:05] <misterno> so normally you dont even resize?
[00:05] Action: JEEB facepalms
[00:05] <JEEB> jesus freaking christ...
[00:05] <misterno> why you facepalm on me
[00:05] <misterno> i came up forward saying i dont know this stuff
[00:05] <misterno> why do that?
[00:05] <JEEB> let me scroll up and copy what I already IIRC said about that
[00:05] <JEEB> and paste it here
[00:06] <misterno> that is mean
[00:06] <misterno> that only makes me not want to ask anymore questions
[00:06] <JEEB> 1) you crop the source [overcropping is better than undercropping] 2) you resize the video as you see fit [features that might be reasons to downscale: bad quality of source, lack of detail and so forth and so forth]
[00:06] <JEEB> sorry, I just get derpy when I have to repeat myself
[00:06] <JEEB> don't take it too personally
[00:06] <JEEB> 2) includes "do not resize"
[00:07] <JEEB> every source I deal with is different, and depending on what kind of source I am dealing with I shall either resize or not resize
[00:08] <misterno> ok but i ask why you wouldn't resize, resizing affects quality too much?
[00:08] <misterno> because in my mind is lower the res = smaller the file size
[00:08] <misterno> and i want to keep some files small
[00:09] <misterno> and i have seen some excellent quality videos with reduces res
[00:09] <misterno> meh nvm , i grilled you too much for today, happy new year!
[00:09] <JEEB> sure was a quick leave
[01:54] <espr3ss0> is this for Development or general help, I might be @ the wrong place
[01:54] <espr3ss0> ?
[01:56] <sacarasc> This is general help.
[01:59] <espr3ss0> thanks sacarasc :-)
[01:59] <espr3ss0> can i ask at the moment ?
[01:59] <sacarasc> You can ask whenever you want.
[02:00] <espr3ss0> i'm not sure the easiest way to achieve what I'm wanting
[02:01] <espr3ss0> specifically at the moment, I have an ".avi" file & the video track is "mpeg-4", and without transcoding make the container ".mp4"
[02:03] <espr3ss0> running "OS X Snow Leopard"
[02:04] <espr3ss0> i've tried using "ffmpeg -i infile -vcodec copy -an -f mp4 outfile.mp4"
[02:05] <espr3ss0> but playback is jumpy and only shows the first several minutes
[02:10] <espr3ss0> oh btw, ...
[02:11] <espr3ss0> do i need to re make / install ffmpeg if i alter "./configure" ?
[02:12] <sacarasc> Yes.
[02:23] <espr3ss0> checked x264 and an error came up, "Warning: PIX_FMT_RGB is missing from libavutil", is there a good place for library downloads ?
[02:23] <espr3ss0> i'm not exactly running the latest n greatest ;-p
[03:09] <defaultro> hey folks, why am I getting this error? I also have the latest x264 compiled/installed. http://pastebin.com/GYca5JqT
[03:11] <sacarasc> Do you have multiple instances of x264 installed?
[03:12] <defaultro> looks like
[03:12] <defaultro> but not a package for sure
[03:12] <defaultro> i have an install back in 2010
[03:12] <defaultro> what's the ideal way of cleaning it? I also installed it from source. I still have the old dir. Will make uninstall work?
[03:14] <sacarasc> Should work.
[03:14] <defaultro> k
[03:14] <sacarasc> Then in the ffmpeg directory, you'll have to make clean && ./configure again.
[03:14] <defaultro> k
[03:15] <defaultro> ran it. It deleted x264.h, libx264.a, x264 binary and x264.pc
[03:16] <defaultro> should I use --enable-static or shared
[03:16] <defaultro> i used static earlier
[03:22] <defaultro> make is running in ffmpeg
[03:24] <defaultro> still failed with the same error
[03:26] <defaultro> i am not sure if it's needed but I don't see libx264.a
[03:28] <defaultro> i'm going to regit x264
[03:41] <defaultro> i see I have x264-lib and x264-devel as package installed
[03:41] <defaultro> i uninstalled it. Hopefully it works this tie
[03:41] <defaultro> time
[03:41] <defaultro> yes, it worked!
[04:04] <pinkette> what is the difference between 5.1  and dolby surround
[10:04] <gxk> i am started to use ffmpeg to stream h264 over mpeg2ts/udp. it works well. i need to low latency, so the cache was disabled in vlc. this is causes to frame lost. What can be done in ffmpeg? my command line http://pastebin.com/u51jRUZq
[10:11] <klaxa> hmm not really a lot i think
[10:12] <gxk> just seconds before found something like '-tune zerolatency' is it helps?
[10:19] <klaxa> i think you would have to re-encode the video stream then
[10:29] <gxk> is there a way to measure latency inside ffmpeg?
[11:00] <yellabs-r2> hello all
[11:01] <yellabs-r2> can i use ffmpeg to turn an colour mp4 to black and white movie ?
[11:10] <ubitux> yellabs-r2: does -vf format=gray works?
[11:10] <ubitux> you might want to play with pix format encoding etc
[11:12] <yellabs-r2> i had this one , but it did not work
[11:12] <yellabs-r2> ffmpeg -i input.mp4 -strict -2 -pix_fmt yuvj420p -vcodec libx264 -acodec aac output.mp4
[11:13] <ubitux> ?
[11:13] <ubitux> and where is the convert to grayscale supposed to be done with your cmd line?
[11:14] <yellabs-r2> its wrong , i guess
[11:14] <ubitux> yellabs-r2: -pix_fmt gray may help
[11:14] <yellabs-r2> [format @ 0x9a0ba60] Unknown pixel format: grey
[11:14] <yellabs-r2> Error initializing filter 'format' with args 'grey'
[11:14] <yellabs-r2> Error opening filters!
[11:14] <ubitux> read again
[11:15] <yellabs-r2> i see
[11:15] <ubitux> note: it's stupid to use -vf format=gray -pix_fmt yuvj420p
[11:15] <ubitux> use -pix_fmt gray direct
[11:15] <ubitux> directly
[11:15] <yellabs-r2> trying
[11:16] <ubitux> since you're not touching audio, -c:a copy instead of -strict -2 -acodec aac would be wise too
[11:17] <yellabs-r2> hmm,  i am to much of an newbee ..
[11:17] <yellabs-r2> ffmpeg -i '/home/yellabs-r2/Bureaublad/ewo-mp4/snt1.mp4' -pix_fmt gray output.mp4
[11:17] <yellabs-r2> tried this, still colour
[11:18] <ubitux> Incompatible pixel format 'gray' for codec 'libx264', auto-selecting format 'yuvj420p'
[11:18] <ubitux> hum.
[11:19] <ubitux> well, then just keep the -vf format=gray
[11:20] <yellabs-r2> testing
[11:20] <ubitux> and drop the -pix_fmt from your cmd line
[11:20] <yellabs-r2> it works
[11:20] <yellabs-r2> :)
[11:21] <yellabs-r2> thanks for the help
[11:21] <ubitux> of course it works
[11:21] <yellabs-r2> :)
[11:21] <ubitux> np
[11:32] <Max__> How can I cut yuv video files with ffmpeg? None of the obvious google hits turned up anything useful.
[11:52] <buhman> Max__: man ffmpeg
[11:52] <buhman> Max__: line 1471,1478
[12:11] <Max__> buhman: thanks but I can't find the relevant lines. I looked at three separate versions of the man pages and I can't seem to find it.
[12:12] <buhman> 04:52 < buhman> Max__: line 1471,1478
[12:12] <buhman> how can you not find them when I told you *exactly* which lines are of interest
[12:12] <Max__> Because line 1471 in my man file is talking about ac3 audio codec.
[12:13] <buhman> you must have some ancient ffmpeg then
[12:14] <Max__> My mistake, perl metadata.
[12:14] <Max__> version 0.10.6
[12:14] <Max__> ^ffmpeg
[12:17] <ubitux> we are in 1.0+
[12:19] <buhman> Max__: http://ffmpeg.org/ffmpeg.html#Main-options paragraphs 6+8
[12:19] <Max__> 0.10.6 built Nov 13 2012, pretty recent. Nothing newer available for my distro. :\
[12:19] <buhman> no that is not recent
[12:19] Action: Max__ submits package request
[12:19] <klaxa> compile from source or get a static build
[12:20] <Max__> Will do, thanks.
[12:21] <buhman> Max__: for latest manpages: git clone git://git.videolan.org/ffmpeg.git && cd ffmpeg/doc && make doc/ffmpeg
[12:21] <buhman> takes all of 30 seconds
[12:22] <buhman> hrm I guess you need to configure first :P
[12:22] <buhman> and not go into doc/
[12:23] <ubitux> Max__: built doesn't mean the sources aren't ancient
[12:24] <ubitux> Max__: https://www.ffmpeg.org/ffmpeg-formats.html#segment_002c-stream_005fsegment_002c-ssegment
[12:24] <Max__> ubitux: I know, I just couldn't find the right line in the man page.
[12:25] <ubitux> all the doc is available only, but is for git/master/HEAD
[12:25] <ubitux> s/only/online/
[12:26] <buhman> well, in his case it *should* be applicable :P
[12:26] <ubitux> segment format has evolved quite a bit recently
[12:26] <ubitux> i'm not even sure it was in 0.10
[12:26] <Max__> 0.10 is only a month older than 1.0...
[12:26] <ubitux> 12:23:12 <@ubitux> Max__: built doesn't mean the sources aren't ancient
[12:27] <ubitux> 0.10 is about 1 year old
[12:27] <buhman> Max__: ^ was released 27-Jan-2012 00:00
[12:27] <Max__> I just did -version and it said 0.10.6 Nov 13 2012.
[12:27] <ubitux> January, 27, 2012, FFmpeg 0.10
[12:27] <ubitux> We have made a new major release (0.10)
[12:27] <buhman> Max__: that's the *BUILD* date
[12:27] <Max__> Oh.
[12:28] <buhman> Max__: as you've been told multiple times now
[12:28] <ubitux> we release 0.11 in the meantime
[12:28] <ubitux> and 0.1
[12:28] <ubitux> erm
[12:28] <ubitux> and 1.0*
[12:29] <buhman> Max__: but really there's no reason to not use git://git.videolan.org/ffmpeg.git
[12:29] <ubitux> and we might release another soon
[12:29] <buhman> as far as I'm concerned, anything prior to 8bc74221f8ef6fe97801025866943013b94f5e2f might as well have been released in 40,000 B.C.
[12:30] <Max__> buhman: there is a very good reason. It's my wife's laptop and I can't always pull and build the newest versions of everything she needs for her research.
[12:30] <buhman> I have no idea what that means, but ok
[12:30] <ubitux> buhman: haha
[12:32] <buhman> also, why does https://github.com/FFmpeg/FFmpeg exist? it's ~8 hours behind at the moment
[12:33] <Max__> Anyway, the segmenter seems to be what she needs. Thanks.
[12:33] <ubitux> buhman: a relic of the past certainly
[12:34] <ubitux> a legendary mirror
[12:34] <buhman> heh
[12:40] <buhman> wow the segmenter is really cool
[12:41] <ubitux> :)
[12:41] <buhman> much easier than -ss and -t :D
[12:45] <buhman> ubitux: why is https://ffmpeg.org/ffmpeg.html much shorter than doc/ffmpeg.1 ?
[12:47] <ubitux> buhman: it's not
[12:47] <ubitux> maybe you're still living in the past
[12:47] <buhman> O.o
[12:47] <ubitux> it's been split recently
[12:47] <ubitux> see latest news
[12:47] <buhman> ffmpeg(1) is 8502 lines
[12:48] <ubitux> 1685 ./doc/ffmpeg.1
[12:48] <ubitux> [~/src/ffmpeg]- ls doc/*.1
[12:48] <ubitux> doc/ffmpeg.1                    doc/ffmpeg-codecs.1   doc/ffmpeg-filters.1  doc/ffmpeg-protocols.1  doc/ffmpeg-scaler.1  doc/ffplay.1   doc/ffserver.1
[12:48] <ubitux> doc/ffmpeg-bitstream-filters.1  doc/ffmpeg-devices.1  doc/ffmpeg-formats.1  doc/ffmpeg-resampler.1  doc/ffmpeg-utils.1   doc/ffprobe.1
[12:49] <buhman> wat
[12:49] <buhman> what commit was that O.o
[12:50] <ubitux> there are several
[12:50] <ubitux> 12:47:37 <@ubitux> see latest news
[12:51] <buhman> somewher between November 14 and December 30 I guess
[12:51] <ubitux> b448677e6ea734b9f7a57970560f600a2a893a9b was the final commit
[12:52] <buhman> so I guess you do some magic to convert texi to that html format
[12:53] <ubitux> that was already the case
[12:53] <ubitux> Stefano improved a bit the script
[12:55] <NuxRo> hello guys, I've got a dd copy of a dvd, basically a iso9660 4GB img file i guess, how could I extract the video from it into some more usable format?
[12:59] <buhman> NuxRo: mount -o loop iso9660-4gb.iso /mountpoint && dvdunauthor --magic-arguments && ffplay -i whatever-combined.vob --more-magic-arguments output.desired-container-and-so-on
[12:59] <buhman> s/ffplay/ffmpeg/
[13:00] <NuxRo> aha, need to look up dvdunauthor, cheers
[13:00] <buhman> http://dvdauthor.sourceforge.net/
[13:27] <gxk> if ffmpeg converts h264 to mpeg2video while streaming mpeg2ts? http://pastebin.com/Gw1RVSLd
[13:28] <burek> gxk, what is your question exactly
[13:31] <gxk> how to run ffmpeg without transcoding from h264 to mpeg2video?
[13:31] <burek> -c:v copy
[13:33] <gxk> hmm, got error "Failed to open bitstream filter h264_mp4toannexb for stream 0 with codec copy: Invalid argument"
[13:34] <gxk> got error - http://pastebin.com/UpAh9TkV
[13:36] <burek> gxk, try removing -bsf h264_mp4toannexb
[13:39] <burek> if that doesn't help, can you try updating your ffmpeg to latest version?
[13:39] <gxk> 1 sec :)
[13:50] <gxk> ffmpeg fresh from git, same problem
[13:56] <gxk> http://pastebin.com/qjBu2BeQ
[13:58] <burek> try adding back -bsf
[14:00] <gxk> http://pastebin.com/2JJbsHD0
[14:01] <burek> you don't need -b:v 4M
[14:01] <burek> since you are doing -c:v copy
[14:01] <burek> but that shouldn't matter in this error
[14:03] <gxk> same problem - http://pastebin.com/yiFiR0SU
[14:06] <burek> it might be a bug, i guess
[14:06] <burek> could you report it on our bug tracker?
[14:07] <burek> you could provide both these logs in your report: http://pastebin.com/qjBu2BeQ  and  http://pastebin.com/yiFiR0SU
[14:11] <ubitux> -fflags +genpts?
[14:11] <ubitux> or specify a rate or something?
[14:11] <burek> with -c:v copy?
[14:11] <ubitux> that's the codec
[14:11] <ubitux> timing is handled at container level
[14:12] <burek> couldn't it be figured out from input
[14:12] <ubitux> input looks like raw h264
[14:12] <burek> so fps is missing?
[14:12] <ubitux> i don't know if it has any usable timing info
[14:12] <ubitux> i'd try to -r 25 or something as input option
[14:12] <burek> gxk, you've heard the man :)
[14:13] <gxk> burek: full attention
[14:16] <gxk> tried all of below - http://pastebin.com/MAMwaaWe
[14:18] <burek> try ffmpeg -re -r 25 -i ./transformers/trans3-720p.h264
[14:18] <burek> ...
[14:20] <gxk> burek, no. do you need another pastebin?
[14:21] <burek> how did you generate that .h264 file
[14:22] <gxk> downloaded mkv of Transformers3, extrackted video es by mkvtools
[14:26] <burek> could you try with the original mkv
[14:26] <burek> or
[14:26] <burek> when you extract the video
[14:26] <burek> try extracting it to mkv container
[14:26] <burek> to preserve all the needed info
[14:26] <AL13N> i'm using ffmpeg in cli and i notice that whatever -r option i give, be it 25 or 12 or 10 or 6; i have only about 80% of the framerate... since i'm using it with output file '-' (inside a php passthru) i would like some kind of way to encoding into webm/ogg that would be faster than it's output rate, so that the browser can continue
[14:27] <burek> this way you've got a raw .h264 video, without some needed info, which causes problems
[14:27] <AL13N> can anyone help me pass options to make it's encoding go faster?
[14:28] <burek> btw, you don't encode things in php page man..
[14:28] <AL13N> http://pastebin.com/dMNr1PqV
[14:29] <AL13N> this gives about fps=10; if i specify -r 25, i have fps=20 if i specify -r 6; i have fps=5
[14:29] <burek> also, that pastebin is not what was asked from you
[14:29] <AL13N> i tried lowering bitrate, but it doesn't help
[14:29] <burek> read more carefully
[14:29] <AL13N> ah
[14:33] <AL13N> http://pastebin.com/7SsbddfX
[14:33] <AL13N> burek: ^^
[14:33] <AL13N> now, i know that this is an older PC, but i'd like someway of speeding things up
[14:34] <burek> you might update your ffmpeg
[14:35] <burek> also
[14:35] <burek> your fps depends on video encoding parameters mostly
[14:35] <burek> which you haven't specified except for the -g
[14:35] <burek> try using -b:v
[14:35] <AL13N> k
[14:36] <burek> http://ffmpeg.org/ffmpeg-codecs.html#libtheora
[14:38] <AL13N> burek: i tried alot of settings: http://pastebin.com/9eYHQprh
[14:38] <AL13N> but nothing helps
[14:38] <burek> no need to focus on audio encoding options
[14:38] <burek> audio encoding is usually fast
[14:38] <burek> focus on video
[14:38] <AL13N> oic
[14:38] <AL13N> any other options i can use?
[14:39] <burek> try something like -b:v 1M
[14:39] <burek> to make it encode video faster
[14:40] <gxk> burek: created new ticket 2098. welcome to call me for testing the issue. BTW, it is my 1st bug with ffmpeg, so hope it created well.
[14:40] <burek> gxk :) :beer: thanks
[14:41] <AL13N> burek: i also tried this: http://pastebin.com/Cvz4tCSV which didn't help either
[14:41] <AL13N> burek: 1M also has 10 fps btw
[14:41] <AL13N> burek: this is what is annoying
[14:42] <burek> maybe your cpu is just not fast enough
[14:42] <burek> can you run screen
[14:42] <burek> and in one screen run ffmpeg
[14:42] <burek> in another run top
[14:42] <burek> to see the cpu usage
[14:42] <AL13N> k
[14:42] <burek> or run ffmpeg in the background
[14:42] <burek> and type top
[14:42] <AL13N> maybe the stats are slowing down the output?
[14:43] <burek> (put -t 20) to end encoding after 20 seconds
[14:43] <burek> while you are testing
[14:43] <AL13N> 99.3% ffmpeg
[14:43] <AL13N> all the time
[14:43] <AL13N> (single core machine)
[14:43] <burek> arm?
[14:43] <AL13N> no
[14:43] <AL13N> Athlon 64
[14:44] <burek> well, 99% tells you the reason
[14:44] <AL13N> 2.4Ghz
[14:44] <AL13N> burek: right, so, is the problem the input quality?
[14:44] <AL13N> that's what my last thought was
[14:44] <AL13N> considering whatever -r:v i gave, i got about 80% of it
[14:45] <burek> try again this example http://pastebin.com/Cvz4tCSV
[14:45] <burek> without -b:v 16k
[14:45] <AL13N> k
[14:45] <burek> or put -b:v 2M
[14:45] <AL13N> ok
[14:47] <AL13N> burek: http://pastebin.com/gMrgKNsM
[14:47] <AL13N> strange
[14:48] <burek> remove -r 6
[14:48] <AL13N> do i leave the scale to 80 ?
[14:48] <burek> yes
[14:50] <AL13N> http://pastebin.com/LELQ8tgR
[14:50] <burek> fps= 50
[14:50] <AL13N> since input rate is 60, now fps=50
[14:50] Action: AL13N sighs
[14:50] <burek> did you check your cpu
[14:50] <AL13N> it's quite strange
[14:51] <AL13N> same
[14:51] <AL13N> 99.3%
[14:51] <burek> now, try with real file output
[14:51] <burek> and after that type ffmpeg -i outputfile.webm
[14:51] <burek> to see what's the real frame rate in the file
[14:52] <burek> and if you want to do live re-encoding, for viewers to watch the video while you are encoding
[14:52] <burek> then use ffmpeg -re -i /media...
[14:53] <AL13N> http://pastebin.com/dFFE1exb
[14:54] <AL13N> the output file has 60 as well, it's just produced at 50fps ... sigh
[14:54] <AL13N> i'll try -re
[14:54] <burek> this output is good
[14:54] <burek> 59.94 fps
[14:54] <AL13N> yes
[14:54] <burek> -re will read the input in "real-time"
[14:55] <AL13N> i'll give it a shot
[14:55] <burek> i.e. ffmpeg will not try to do everything as fast as it can
[14:56] <AL13N> http://pastebin.com/R0961EJH
[14:56] <AL13N> 99.3% cpu again
[14:56] <AL13N> it seems to be a small bit better
[14:56] <AL13N> but not 60 fps
[14:56] <burek> well, I'm not sure, but everything is telling me that either something is not well optimized for your hardware
[14:56] <AL13N> should i try to lower bitrate?
[14:56] <burek> or that cpu is slow
[14:56] <burek> try to lower the output frame rate
[14:57] <burek> add -r 30 before -f ogg
[14:57] <AL13N> ok
[14:57] <AL13N> fps seems to be 26 now before it's finished
[14:58] <burek> that fps which is shown during the encoding is the approx. curr. fps of the encoding
[14:58] <AL13N> 27fps
[14:58] <burek> also, why do you use -g 0
[14:58] <AL13N> http://pastebin.com/7XcnzS8e
[14:59] <AL13N> burek: some docs told me that -g 0 would disable me to make it faster
[14:59] <burek> -g 0 means "make every frame a key frame"
[14:59] <burek> why would you want that
[15:00] <AL13N> ah, ok, losing that
[15:00] <burek> you are wasting (cpu) time on every frame, to encode all the info from the frame, instead to encode only differences from the next frame
[15:01] <AL13N> http://pastebin.com/pi0PevjT
[15:01] <AL13N> seems to be 27fps still
[15:01] <burek> doesn't have to be
[15:01] <burek> try file output
[15:01] <burek> and ffmpeg -i output
[15:02] <AL13N> k
[15:03] <AL13N> http://pastebin.com/vpTvPFsU
[15:04] <AL13N> still 27
[15:04] Action: AL13N sighs
[15:05] <burek> Input #0, ogg, from 'file.ogg':
[15:05] <burek> Stream #0:0: Video: theora, yuv420p, 80x45 [SAR 1:1 DAR 16:9], 30 fps, 30 tbr, 30 tbn, 30 tbc
[15:06] <burek> where did you see 27?
[15:06] <AL13N> at the end of encoding, the speed at which it encoded
[15:06] <burek> that's not accurate
[15:06] <burek> i told you already
[15:06] <burek> it's approx
[15:06] <AL13N> yes
[15:06] <AL13N> but even though the output is correct at 30fps
[15:07] <AL13N> i would like that the speed would be 27 fps as well, so that i have no delay when people play it in a <video></video> tag
[15:07] <AL13N> err, i mean 30fps
[15:07] <burek> it IS 30
[15:07] <AL13N> the output rate is 30
[15:08] <AL13N> but the speed at which the frames are produced by ffmpeg is approx 27fps
[15:08] <burek> loose -re
[15:08] <AL13N> which means the output of this in a pipe will be coming in at around 27fps, thus playing a bit and then pausing to get more data and then playing
[15:08] <burek> no, it doesn't mean that
[15:09] <AL13N> ok, loosing -re
[15:09] <burek> it just means ffmpeg took some time to start up
[15:09] <burek> initialize encoders, pipes and filters
[15:09] <burek> which, when calculated in current fps, makes it 27
[15:09] <gxk> burek: original mkv file works good. you was right!
[15:09] <burek> because of that (idle) time when it didn't encode
[15:09] <burek> gxk, well, ubitux told you, your h264 file is just raw video frames
[15:10] <burek> without info about frame rate or timestamps
[15:10] <AL13N> burek: right, but is there a way for me to get the output rate lower than the approx fps it produces the frames?
[15:10] <burek> which creates problems eventually :)
[15:10] <burek> mkv keeps all that info, that's why it works
[15:10] <burek> AL13N just try a live test
[15:10] <burek> and you'll see what im telling you
[15:11] <burek> it is working well, it's just the start of the encoding that makes it appear as less than 30 fps
[15:11] <burek> which it actually isn't
[15:11] <burek> and with live encoding/viewing, you can't send more data to viewer
[15:11] <burek> than it is already available
[15:12] <gxk> burek: how can ffmpeg extract video track with needed info from mkv?
[15:12] <burek> that would seriously undermine einstein's relativity theory :)
[15:12] <AL13N> burek: ok
[15:12] <burek> gxk, just extract it like: ffmpeg -i original.mkv -an -c:v copy justvideo.mkv
[15:13] <burek> -an = no audio
[15:13] <burek> so, you are just copying the video streams inside original mkv file into a new mkv file, which will contain just video stream(s)
[15:14] <burek> AL13N, you might check if the viewers/players have any caching options
[15:14] <burek> which gives you what you are after
[15:14] <AL13N> k
[15:17] <AL13N> burek: well, i'm using html5 video tags, so i donno if there is something... however, it seems the speed is not really enough... i would have thought it to be possible to encode something with bad quality faster than it actually takes to play (like double)
[15:18] <burek> you'll have to test more
[15:18] <burek> im not sure if it's the cpu
[15:18] <burek> if you were using libx264 instead of libtheora
[15:18] <burek> there would be more options to fine-tune your output
[15:18] <burek> and cpu usage
[15:18] <burek> with libtheora im not that much familiar :/
[15:20] <AL13N> it seems MP4 is not streamable, if i have output file '-' than -t mp4 gives error like that
[15:24] <sacarasc> You need to do special things with MP4 to make it streamable.
[15:31] <AL13N> burek: i retried on another machine, and that one can have approx fps much higher than the output rate
[15:31] <AL13N> sacarasc: any idea what?
[15:33] <sacarasc> Moving the moov atom to the start of the file using qtfaststart for non-live streaming. For live streaming, I think you have to do that and break the files up into multiple little ones.
[15:37] <ambro718> Hi. Has someone implemented using OpenMAX in ffmpeg to offer transparend hardware acceleration?
[15:37] <gxk> which option removes metadata builtin stream like http://pastebin.com/pUUDT6jD
[15:48] <Agiofws> hello
[15:48] <Agiofws>  i have this  kind of  media file which a dvd player of mine will not play http://www.pasteall.org/38473 can anyone advise to  re encoded it with what options  thank you  ?
[15:49] <burek> sacarasc: http://ffmpeg.org/ffmpeg-formats.html#MOV_002fMP4_002fISMV (especially -movflags faststart)
[15:50] <burek> AL13N, then your cpu is too old..
[15:50] <burek> or your ffmpeg was not compiled for your architecture
[15:50] <burek> try compiling the latest git version
[15:50] <Agiofws> anyone ?
[15:50] <burek> and latest libtheora of corse
[15:50] <burek> course*
[15:51] <AL13N> my cpu is old-ish, but still...
[15:51] <burek> <burek> try compiling the latest git version <burek> and latest libtheora of course
[15:51] <AL13N> yes, but this is a production server, so it's not that easy
[15:51] <burek> or try static builds
[15:51] <AL13N> but thanks for the help, i guess i know where to look
[15:52] <AL13N> i was hoping there was something that could drop frames when reading or something
[15:52] <AL13N> so that at least that would be faster read
[15:52] <burek> Agiofws, try typing: ffmpeg -i yourfile
[15:52] <burek> and copy the output to a pastebin-like site (like www.pastebin.com)
[15:53] <burek> AL13N, you can forse the input frame rate
[15:53] <burek> force*
[15:54] <AL13N> i tried with -r before -i , but i got: invalid "frame_rate" property
[15:54] <burek> but still, the frames sent to the player/viewer will be with the same timestamps
[15:54] <AL13N> in this MP4
[15:54] <burek> you won't be able to send "more data" in advance
[15:54] <AL13N> i'm not particularly worried about that
[15:54] <burek> (with live video of course)
[15:54] <burek> with pre-recorded video, well, just offer a .flv file or .mp4
[15:54] <AL13N> burek: i'm not using a streaming protocol
[15:55] <burek> and it will work, depending on the bandwidth the viewer has got
[15:55] <AL13N> if there was a way to produce less cpu when reading the file, it might go better
[15:55] <AL13N> so that's what i'm thinking
[15:55] <burek> AL13N, what is your overall (logical) goal?
[15:56] <burek> gxk, did you read about metadata in ffmpeg docs?
[15:56] <AL13N> atm, i'm thinking that this server is to slow to read the input file at 60 fps; so i'm trying to get it to use less cpu when reading the file, which is likely the bottleneck atm
[15:57] <burek> gxk, maybe something like -map_metadata -0
[15:57] <burek> AL13N, if that's the case
[15:57] <burek> you'll need to re-encode your input video
[15:57] <burek> to lower the frame rate
[15:57] <gxk> burek: checking...
[15:58] <Agiofws> burek,  http://pastebin.com/8RhLJ7JV
[15:58] <Agiofws> thnak you
[15:58] <burek> oh man
[15:58] <burek> i think you have the first version of ffmpeg
[15:58] <burek> ever compiled
[15:58] <burek> :D
[15:59] <burek> you really need to update your ffmpeg Agiofws :)
[15:59] <burek> however, Agiofws, do you have a sample media file that your dvd CAN play?
[16:00] <AL13N> burek: myes, i had figured that, but trying to avoid
[16:00] <Agiofws> burek,  no i don't sorry ...
[16:00] <AL13N> burek: if all else fails, i'll just hold off outputting until done (buffering in server)
[16:00] <burek> Agiofws, well there's no way we can predict what your dvd can play :(
[16:01] <burek> AL13N, what is your overall (logical) goal?
[16:01] <burek> what are you trying to accomplish?
[16:03] <Agiofws> but generally it does play avi's i think it just does not play movies which are in a blue ray ? format .. if you can give me some pointers ? maybe its the not normal resolution ?
[16:04] <burek> Agiofws, avi is a container, like a box, containing media streams (audio/video/subtitles)
[16:04] <burek> so your dvd doesn't play avi's, but what's inside of it
[16:05] <Agiofws> burek,  i assume its the AAC sound codec ?
[16:06] <burek> that's why i asked if you have any other media that you can play in your dvd
[16:06] <burek> to run an info on it and see what it is
[16:06] <burek> brb
[16:08] <Agiofws> agiofws at ixus:/media/usb0/chris/clouldAtlas$ file 0001-247812.avi
[16:08] <Agiofws> 0001-247812.avi: RIFF (little-endian) data, AVI, 720 x 304, 23.98 fps, video: FFMpeg MPEG-4, audio: MPEG-1 Layer 3 (stereo, 44100 Hz)
[16:26] <wieshka> guys help out with -vf scale - I need to scale 16:9 video (actual video area is 4:3, with black margins) to 4:3 with cuted out black area
[16:26] <wieshka> input 1920x1080, taget ouput: 800x600
[16:27] <Mavrik> you'll have to use crop filter to cut away the black bands before scaling
[16:28] <Mavrik> so something like
[16:28] <wieshka> -filter:v "crop=800:600:560:0" ?
[16:28] <Mavrik> depending on the order of filters
[16:29] <Mavrik> if you want to crop before scaling you do something like
[16:29] <Mavrik> crop=1440:1080:240:0,scale=800:600
[16:29] <Mavrik> that will crop the frame first to 4:3 and then scale it down to 800x600
[16:31] <wieshka> Mavrik, thx, figured out
[16:31] <wieshka> only i got: [swscaler @ 0x27f3de0] Warning: data is not aligned! This can lead to a speedloss
[16:31] <wieshka> as warning
[16:32] <wieshka> setdar=4:3 ?
[16:32] <Mavrik> nah, that's something else
[16:32] <Mavrik> (image frame data is not aligned to 64-bit boundaries)
[16:32] <Mavrik> it has nothing to do with images :)
[16:32] <Mavrik> probably a bug that has to be addressed
[17:02] <canaux> Hi all
[17:16] <canaux> does ffmpeg permits to manage the aac encoder delay in another way than assuming a fixed arbitrary delay ? I need to encode files with ffmpeg that will be well "understood" by Apple tools.
[17:16] <canaux> Here are more information about encoder delay: http://developer.apple.com/library/mac/#documentation/QuickTime/qtff/QTFFAppenG/QTFFAppenG.html
[18:49] <Diogo> hi anyone know a good repository of video in different formats to test encodings?
[18:54] <saste> Diogo, the ffmpeg FATE samples suite
[19:03] <Diogo> to encode video with audio aac and Bit rate mode: Variable
[19:13] <Diogo> saste: but this samples i view a webm , mp4 file, a wmv ..examples in different formats
[19:13] <Diogo> ?
[19:21] <Diogo> this commando put me audio bitrate a constant why ? : /servers/ffmpeg/bin/ffmpeg  -i '.$file.'  -r 25 -s 580x326 -c:v libx264  -vprofile baseline  -b:v 700k  -keyint_min 50 -pass 2 -acodec libfaac -ar 44100 -ab 128k -ac 2  -f mp4 '.$to
[19:33] <llogan> Diogo: because you used "-ab 128k"
[19:33] <llogan> try with "-q:a 100" when using libfaac for VBR
[19:34] <llogan> it is similar to "faac -q 100"
[19:35] <llogan> but VBR audio doesn't really make sense for a two-pass video encode
[20:00] <tmkt> Afternoon
[20:00] <tmkt> any idea how to fix "stream 0, error opening alias: path='/Users/',"
[20:00] <tmkt> on a .mov file i received from a user?
[20:01] <llogan> probably one of those mov files don't actually contain video but point to it
[20:02] <tmkt> http://pastebin.com/QdaLp3F3
[20:12] <tmkt> guessing it was created/exported incorrectly from that info
[20:37] <llogan> tmkt: ask them to send you an actual video file instead of a referenced mov
[20:38] <tmkt> they aren't smart enough to know what they did
[20:38] <tmkt> they told me they renmaed the file
[20:38] <tmkt> and re-uploaded it
[20:38] <tmkt> and that version worked
[20:38] <tmkt> the first file was 5k..the 2nd one was 1.8M
[20:39] <llogan> i love clients like that
[20:43] <llogan> tmkt: tell them to check the "make movie self-contained" box
[20:47] <tmkt> k...thanks
[20:53] <Sashmo_> for the life of me, I cant figure out how to get ffmpeg to play a range of files.  Is it possible with out having to script it?  Is there a way to play all files in a directory?
[20:59] <klaxa> ffplay * ?
[20:59] <klaxa> mmh...
[20:59] <klaxa> doesn't look like it
[21:01] <klaxa> you can always do:
[21:01] <klaxa> for i in *; do ffplay "$i"; done
[21:01] <klaxa> only if your shell supports that of course
[21:02] <klaxa> or use a player for playback, i think ffplay isn't supposed to be used for anything else than testing files
[21:04] <llogan> consider adding -autoexit to klaxa's example
[21:09] <espr3ss0> hi again ! ;-p
[21:10] <espr3ss0> do i need to specify "-vpre" for ffmpeg with "libx264" every time ?
[21:11] <Sashmo_> klaxa: thats what I want to avoid....
[21:11] <klaxa> avoid what? using a for-loop or using a mediaplayer?
[21:11] <klaxa> just get mplayer or mplayer2
[21:11] <klaxa> then you can run mplayer *
[21:16] <llogan> espr3ss0: it depends on your ffmpeg version
[21:16] <llogan> recent ffmpeg will use -preset instead of -vpre, and default is "-preset medium" if you do not declare a preset.
[21:19] <celesteh> i've got a commandline argument string that will record audio via jack, which is fantastic. i have to connect the inputs via the JACK Control GUI. Is there a way to specify them in the command line?
[21:21] <klaxa> celesteh: maybe some JACK cli?
[21:23] <llogan> try #jack
[21:32] <maujhsn209_> Hi I'm new to kdenlive and want to know which version of ffmpeg is best compatible with kdenlive...I've used this site as a guide: https://launchpad.net/~sunab/+archive/kdenlive-svn
[21:34] <Sashmo_> klaxa: yeah, I could.... I'll give it a try later....
[21:37] <espr3ss0> llogan: "FFmpeg SVN-r26402" on Mac's "Snow Leopard"
[21:37] <ubitux> wat
[21:38] <ubitux> svn, really?
[21:40] <llogan> espr3ss0: that's incredibly old
[21:41] <espr3ss0> llogan: i know ;-p
[21:41] <espr3ss0> thanks !
[21:42] <llogan> maujhsn209_: are you familiar with ffmpeg?
[21:43] <maujhsn209_> llogan Yes!
[21:45] <maujhsn209_> llogan I am concerned with compatibility issue w/ kdenlive!
[21:46] <maujhsn209_> llogan See this: https://launchpad.net/~sunab/+archive/kdenlive-svn
[21:50] <Sashmo_> anyone have experience converting PAL to NTSC ?
[21:51] <llogan> maujhsn209_: then i recommend you look at the kdenlive presets and see if their syntax will work with recent ffmpeg...or just ask at #kdenlive
[21:53] <Sashmo_> in my conversion I am getting some bad looking text in the news scrolls at the bottom of the screen, anyone know if there is some special filters that I should add?
[21:54] <maujhsn209_> llogan...So it okay to just dowload ffmpeg like I would do normally & worry about the preset stuff later...is this correct?
[22:02] <llogan> maujhsn209_: i suppose. you can always make your own presets too.
[22:05] <maujhsn209_> Okay thanks!
[22:09] <Sashmo_> http://pastebin.com/bwUgSbH2
[22:09] <Sashmo_> thanks guys
[22:12] <llogan> Sashmo_: do you get the bad looking text if you omit "-r 29.97"?
[22:12] <llogan> why -vsync 1?
[22:12] <Sashmo_> llogan: yes
[22:13] <Sashmo_> llogan: I found that the vsync helped
[22:13] <llogan> then it's probably due to interlacing
[22:13] <Sashmo_> llogan: a bit
[22:13] <Sashmo_> I tried yadif filter
[22:13] <Sashmo_> it helped a bit, but the text was still fuzzy and choppy
[22:14] <llogan> and you should use -r ntsc or -r 30000/1001 instead of 29.97
[22:14] <llogan> and you didn't include the complete output
[22:14] <Sashmo_> let me try
[22:15] <llogan> it won't make a difference
[22:15] <Sashmo_> llogan: complete output http://pastebin.com/YK4YULkj
[22:29] <Sashmo_> llogan: what do you think?
[22:40] <celesteh> when somebody starts ffmpeg at the command line with "ffmpeg -f jack -ac 2 -i ffmpeg -f x11grab ... etc" can somebody explain what's going on with the syntax?  Is -f x11grab a couple of arguments to ffmpeg or is x11grab the name of a separate program that ffmpeg invokes, such that the user would be able to specify additional arguments to x11grab or to jack?
[22:54] <Sashmo_> so here is what I have been able to do, its better, but the quality is still fuzzy on the text, and the scrolls are choppy not smooth, any ideas guys? http://pastebin.com/8SBjBhgH
[22:56] <llogan> celesteh: x11grab is an input device that can accept various options
[23:22] <maujhsn209> I have encountered a build error I have included the ffmpeg config.log...pastebin.com/baZMfDVj need help thanks!
[23:25] <maujhsn209> I have encountered a build error I have included the ffmpeg config.log...http://www.pastebin.com/baZMfDVj need help thanks!
[23:26] <klaxa> wha-
[23:26] <klaxa> can you... paste the file raw without the nano editor stuff?
[23:27] <maujhsn209> klaxa I'll get back to you in just a few moments...thanks!
[23:32] <maujhsn209> klaxa Here it is: http://www.patebin.com/n4hgte82
[23:34] <klaxa> install libvpx-dev
[23:34] <ubitux> patebin :)
[23:34] <klaxa> that too :P
[23:35] <maujhsn209> klaxa http://www.pastebin.com/n4hgte82 :)
[23:36] <klaxa> maujhsn209: install libvpx-dev
[23:44] <maujhsn209> klaxa sudo apt-get -y remove the "previous"  libvpx-dev....cd ~/libvpx make clean git pull...Yes or No?
[23:44] <klaxa> wait... you already have libvpx-dev?
[23:44] <maujhsn209> It was one of the steps
[23:45] <maujhsn209> In the build process!
[23:45] <Sashmo_> the current yadif filer in ffmpeg, is that interpolating or blending?
[23:45] <klaxa> hmm...
[23:46] <maujhsn209> klaxa I am not making this stuff up!
[23:46] <klaxa> yeah well it would probably be best then to compile from source
[23:47] <maujhsn209> No thanks!
[23:48] <maujhsn209> Bye!
[23:49] <klaxa> what i said was a yes to: "sudo apt-get -y remove the "previous"  libvpx-dev....cd ~/libvpx make clean git pull...Yes or No?"
[00:00] --- Thu Jan  3 2013


More information about the Ffmpeg-devel-irc mailing list