[Ffmpeg-devel-irc] ffmpeg.log.20120502

burek burek021 at gmail.com
Thu May 3 02:05:02 CEST 2012


[00:00] <Shimmy> burek, thing is, amv
[00:01] <Shimmy> sorry, amv's underlying audio, is adpcm_ima_amv, which leads us to this issue i posted a while ago: https://ffmpeg.org/trac/ffmpeg/ticket/747
[00:01] <Shimmy> https://ffmpeg.org/trac/ffmpeg/ticket/747
[00:02] <burek> Shimmy, did you click on "specifications" ? :)
[00:03] <Shimmy> burek, where
[00:04] <Shimmy> yeah, that's what im saying
[00:04] <Shimmy> perhaps i should look into converting to amv (which is pain for itself)
[00:04] <burek> its not
[00:05] <Shimmy> anyway i have that native.avi file i mentioned that does work on my device
[00:05] <burek> ffmpeg -i native.avi output.amv
[00:06] <Shimmy> burek, BIG RED ERROR: [NULL @ 0000000001f04950] Unable to find a suitable output format for 'output.amv' output.amv: Invalid argument
[00:06] <burek> can you please use pastebin.com, to show your command line and its output?
[00:06] <Shimmy> sec
[00:07] <Shimmy> http://pastebin.com/ASzr2j72
[00:07] <Shimmy> burek here u go http://pastebin.com/ASzr2j72
[00:07] <Shimmy> gonna make a coffee now excuse me
[00:07] <burek> can you type
[00:07] <burek> ffmpeg -codecs
[00:07] <burek> and see if there is AMV
[00:07] <Shimmy> amv is pain in the @$$
[00:08] <Shimmy> it appers there
[00:08] <burek> does it have DEV letters left to it
[00:08] <Shimmy> yes
[00:08] <burek> try ffmpeg -i bla.avi -f amv out.amv
[00:08] <burek> wait
[00:08] <burek> my bad
[00:08] <burek> its encoder, not format
[00:08] <burek> just a sec
[00:09] <burek> oh
[00:09] <Shimmy> same error as before [NULL @ 0000000001e25550] Requested output format 'amv' is not a suitable output format output.amv: Invalid argument
[00:09] <burek> try ffmpeg -i bla.avi -vcodec amv out.avi
[00:09] <burek> amv is encoder in ffmpeg, not a format
[00:09] <Shimmy> sec
[00:09] <burek> wikipedia says it's a variant of AVI format, so this would be the reasonable command
[00:09] <burek> ok
[00:09] <Shimmy> right, checkin
[00:10] <burek> of
[00:10] <burek> ffmpeg -i bla.avi -vcodec amv -f avm2 out.amv
[00:10] <burek> that would be my second guess
[00:11] <burek> just ignore my last line.. amv <> avm...
[00:12] <Shimmy> i almost forgot about my coffee, ill have to reboil the water for the 3rd time :P
[00:12] <Shimmy> meanwhile copying files to device
[00:12] <burek> :) ok
[00:15] <Shimmy> ok it didn't work. lemme see if i can find an amv file in my archive. i once had one that's supposed to be workin
[00:16] <burek> btw, did you try plain old ffmpeg -i bla.avi -s ... output.avi
[00:21] <Shimmy> what do you mean -s?
[00:22] <Shimmy> oh
[00:22] <Shimmy> just setting size
[00:22] <burek> yes :)
[00:22] <Shimmy> i dont believe that's gonna work, i can give it a shot tho
[00:22] <burek> also, use -t 30
[00:22] <burek> to shorten time
[00:22] <burek> for encoding/copying/testing
[00:22] <Shimmy> it's only 30 secs anyway...
[00:23] <burek> ok
[00:23] <burek> btw, if it's mp4 player, why didn't you try -vcodec libx264 -f mp4 out.mp4
[00:25] <Shimmy> it didn't work
[00:25] <Shimmy> and nope, it's not an mp4 player :p
[00:25] <Shimmy> it's a nasty obsolete device
[00:25] <Shimmy> but i have to make it work
[00:25] <Shimmy> otherwise im fired
[00:27] <Shimmy> btw, how long do u think it can take for ffmpeg gurus to solve these issues? maybe i should hire someone, what are the avg rates anyway?
[00:27] <Shimmy> i live in a far country...
[00:27] <burek> you can check here http://ffmpeg.org/consulting.html
[00:28] <burek> since when the Israel is a far country :)
[00:35] <burek> Shimmy, I think this was what you were looking for all the way: http://ffmpeg.org/faq.html#How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f
[00:35] <burek> shortly, use -vcodec mpeg4 and -vtag XVID
[00:38] <Shimmy> ok, that's what im using all the way long
[00:38] <burek> no, you were using -vcodec libxvid
[00:38] <burek> which is not the same as -vcodec mpeg4
[00:38] <Shimmy> oh oh, lemme retry
[00:47] <elkng> whe doing 2 pass decoding, what happens on first pass ? there is file created with strings like: "in:144574 out:144574 type:2 q:236 itex:0 ptex:0 mv:0 misc:767 fcode:1 bcode:1 mc-var:0 var:1394 icount:0 skipcount:697 hbits:65;", what are those
[00:47] <elkng>  strings ?
[00:47] <burek> decoding?
[00:48] <burek> you mean encoding? :)
[00:48] <cbreak> elkng: 2 pass encoding's only purpose is to encode a movie so that the quality of the whole thing is similar all across the timeline
[00:49] <cbreak> and to do that and be able to reach a given filesize/average rate goal, you need to know ahead in time which parts of the move that are complex and which are not
[00:49] <burek> 2-pass encoding is very helpful because in the 1st pass the input video is analyzed and statistics are made, so that 2nd pass can better encode it (smaller size, higher quality)
[00:49] <cbreak> that's what the log file is for
[00:49] <cbreak> if you don't need to reach a filesize goal, you can reach a similar or maybe better quality with crf encoding with x264
[02:25] <elkng> I'm trying to encode video using: "ffmpeg -i 1.avi -vcodec mpeg4 -s 320x240 -aspect 5:4 -r 15 -acodec libmp3lame -ar 44100 -ac 2 -ab 128k -y -f 3gp 2.3gp", but it says: "[3gp @ 0x8099640] track 1: could not find tag, codec not currently supported in
[02:25] <elkng>  container Could not write header for output file #0 (incorrect codec parameters ?)", does that mean "3gp" cant contain "libmp3lame" audio ?
[02:30] <burek> can you please use pastebin.com, to show your command line and its output?
[02:30] <burek> and don't use -s and -aspect together, there's no need for both
[02:30] <burek> either use -s or -aspect
[02:37] <elkng> burek: the command I use: http://sprunge.us/DgPF the output: http://sprunge.us/cXLW
[02:40] <burek> elkng, try changing mp3 to aac
[02:40] <burek> -acodec aac -strict experimental -ar .. -ac .. -ab ...
[02:42] <elkng> "don't use -s and -aspect together", which of them has precedence if using together ?
[02:44] <burek> as a matter of fact
[02:44] <burek> I'm reading of PAR/DAR/SAR
[02:44] <burek> and I might be wrong about that
[02:44] <burek> if -aspect defines DAR param, then you need to use it
[02:44] <burek> but from the documentation I don't see if it does that or not
[02:45] <elkng> "-acodec libfaac" works without "-strict experimental", is there difference between "aac" and "libfaac"
[02:49] <burek> yes
[02:50] <elkng> what is better to use "aac" or "libfaac" ?
[02:51] <burek> I really don't know
[02:51] <burek> try and see
[02:51] <elkng> "aac" is experimental and unstable ?
[02:51] <burek> that's internal ffmpeg's encoder
[02:51] <burek> libfaac is external library
[02:53] <elkng> is there difference in power consumption playing video on smartphone, one encoded with "bitrate: 500 kb/s" and same video encoded with "bitrate: 2000 kb/s" ? what video can drawn less power ? with "bitrate: 500 kb/s" ?
[02:54] <burek> 2 mbps makes 4 times more data than 500 kbps
[02:54] <burek> you decide yourself
[03:50] <mancha> hi there folks. any dev around? i've identified a bugeroo
[04:37] <Guest6074> Can anyone tell me how to compile ffmpeg with h264 on open suse?  Im having troubles.
[04:38] <Guest6074> Do I even need to do it any more?  I'm trying to use x11grab so I can screencast
[06:10] <taqattack> FFmpeg seems to get throttled when streaming to RTMP in Windows. I streamed the same file for same file with same parameters using FFmpeg build from same source and I get different results. http://i.imgur.com/u1sWy.png
[07:37] <redcheckers> hi! i had a quick question as a newbie to using ffmpeg,   i was following an online guide/blog that used the   "   -b " flag with ffmpeg, and when i try t use it, my ffmpeg complains that it is ambiguous "Please use -b:a or -b:v, -b is ambiguous"   is that referring to a=audio or v=video?
[07:39] <redcheckers> (and is that the same as using the " -av " flag?
[07:40] <aphid> yep, the guide is probably dated
[07:41] <redcheckers> ok thanks! i'm gonna go try encoding video
[09:56] <juanmabc> how could i use ffmpeg as a metadata editor without converting ? based on the idea of: ffmpeg -i in.avi -metadata title="my title" out.flv
[09:56] <ubitux> add -c copy
[09:56] <juanmabc> still further?
[09:57] <ubitux> as output option, so before out.flv
[09:57] <ubitux> but after in.avi
[09:57] <juanmabc> as in in-place, so not even copying frames
[09:57] <ubitux> no i think there is no solution except remuxing
[09:58] <juanmabc> damn! :P
[09:58] <ubitux> (and btw you will need to remux in your example since you're switching format)
[09:58] <juanmabc> yeah, that's from man ffmpeg, it's just the -metadata idea
[09:58] <juanmabc> would be cool to support that, next release!
[09:59] <juanmabc> or next next! :P
[09:59] <ubitux> it's kind of complicated
[09:59] <ubitux> you can't do that with all formats without remuxing
[10:00] <juanmabc> yeah, that could be true
[10:02] <ubitux> and given the current design of ffmpeg, i'd guess it's not a simple task
[10:02] <ubitux> but feel free to give it a try yourself :)
[10:50] <burek> juanmabc, google for flvmeta
[10:50] <juanmabc> i wanted ffmpeg for its multicodec support
[10:52] <burek> well, as ubitux has said, not all formats support injecting metadata without remuxing
[10:53] <juanmabc> at least, i got -c copy
[10:54] <ubitux> in case of multiple stream only one of each type will be selected
[10:54] <ubitux> you might want to consider the -map option to keep them all while remuxing
[10:55] <burek> is there something like -c:all
[10:59] <burek> well "-c:v copy -c:a copy -c:s copy -c:d copy -c:t copy" will copy everything
[10:59] <burek> accourding to this http://ffmpeg.org/ffmpeg.html#Stream-specifiers-1
[11:00] <juanmabc> cool
[11:01] <ubitux> burek: no, -c copy already does this
[11:01] <ubitux> the thing is, if you have multiple audio streams for instance
[11:01] <ubitux> only the "best" one will be selected
[11:02] <burek> well, that's not logical
[11:02] <ubitux> i think you need sth like -map -1 (i don't remember the exact syntax)
[11:02] <ubitux> why?
[11:02] <burek> why would ffmpeg choose anything if you specified "all"
[11:02] <ubitux> where did you specified all?
[11:02] <burek> -c
[11:03] <ubitux> -c copy means "codec will be the same"
[11:03] <ubitux> not "all the streams will be copied"
[11:03] <ubitux> in case of kept streams, then the codec will be copied
[11:03] <burek> well, that's what's not logical to me
[11:03] <ubitux> -map is here to polish the selection
[11:03] <burek> if you didn't specify -c:something
[11:04] <burek> then you obviously meant -c:all
[11:04] <ubitux> if you didn't transcoding is done
[11:04] <ubitux> still, on the selected streams
[11:04] <burek> frankly, when people do: ffmpeg -i input -c copy output
[11:04] <burek> isn't it logical to assume they want to copy everything
[11:04] <burek> and just remux
[11:05] <juanmabc> well, ffmpeg picks some defaults according to output format
[11:05] <burek> yes, I realize that, I'm just wondering why does it pick anything
[11:05] <burek> why doesn't it try to copy all
[11:05] <burek> and if not possible to hit a warning
[11:05] <burek> and pick something default
[11:06] <burek> I mean, I don't say this way is wrong
[11:06] <burek> it's just not obvious for beginners to understand why doesn't -c copy just copy everything
[11:07] <burek> Here for example http://ffmpeg.org/ffmpeg.html#Stream-specifiers-1
[11:07] <burek> "E.g. the stream specifier in -b:a 128k matches all audio streams."
[11:07] <juanmabc> it was even shocking to me that ffmpeg -i file.avi file.mpg remuxed even codecs to me
[11:07] <burek> that's logical
[11:07] <juanmabc> yeah, now
[11:08] <burek> oh, so you would expect ffmpeg for that command to just remux the input into output?
[11:08] <ubitux> burek: -c isn't related at all to selection
[11:08] <ubitux> it's related to how the transcode/transmux will be done
[11:08] <ubitux> to *stream* selection
[11:09] <ubitux> -map is related to stream selection
[11:09] <ubitux> For example, to map ALL streams from the first input file to output
[11:09] <ubitux> ffmpeg -i INPUT -map 0 output
[11:10] <ubitux> so: ffmpeg -i INPUT -map 0 -c copy OUTPUT
[11:12] <burek> ubitux, I believe it's not, it's just, if there is something like -c[:selection]
[11:12] <burek> and if selection is omitted, I would expect it to mean "all"
[11:12] <burek> and not "pick something by default"
[11:12] <burek> that's what I'm saying
[11:12] <ubitux> it's more like -c[:filter_appliance]
[11:13] <ubitux> -c  option will affect all kind of codec types
[11:13] <ubitux> -c:a  option will affect audio codecs
[11:13] <burek> ok, name it as you like, but if there is no filter, wouldn't it be logical to assume "all"
[11:13] <ubitux> well
[11:13] <ubitux> it's all :)
[11:13] <burek> well it's not :(
[11:13] <ubitux> it is
[11:13] <burek> if ffmpeg picks something
[11:13] <ubitux> this is stream selection
[11:13] <burek> and leaves the other
[11:13] <burek> then it's not quite :)
[11:13] <juanmabc> problem is inputs (map) and codecs of inputs
[11:14] <burek> yes I understand that
[11:14] <ubitux> burek: -c == all the selected streams will be copied
[11:14] <burek> in this case :selection can be used instead of -map
[11:14] <burek> for example
[11:14] <burek> -c:a:0
[11:14] <juanmabc> picks defaults inputs (maps) and says -c copy that inputs codecs, so the pick default inputs (maps) is where got stuck
[11:14] <burek> is the same as -map a:0
[11:14] <ubitux> burek: -c copy ` all the streams will be copied
[11:14] <ubitux> burek: -c copy = all the selected streams will be copied
[11:14] <ubitux> as simple as that.
[11:15] <burek> maybe simple to a developer
[11:15] <burek> but still isn't logical to user
[11:15] <burek> that's my position
[11:15] <juanmabc> what would be cool, is a -map all
[11:15] <ubitux> juanmabc: it's called -map 0
[11:15] <juanmabc> ah
[11:15] <burek> it would be cool to set it as default
[11:15] <juanmabc> then just that with -c copy
[11:15] <burek> and that would solve this
[11:16] <ubitux> -c copy must not affect the selection
[11:16] <burek> btw, why isn't it default already? why is there a need to pick the best stream?
[11:16] <ubitux> but yes, -map 0 could be the default
[11:16] <ubitux> but that's another debate :)
[11:16] <juanmabc> well, i have to agree that picking a default input like the first, instead of all is kinda how it works, but now ideal
[11:17] <ubitux> burek: maybe because a lot of formats don't support multiple streams
[11:17] <juanmabc> so all votes up :D
[11:17] <juanmabc> i think so
[11:17] <burek> I see, makes sense
[11:17] <ubitux> and thus selection 1 video and 1 audio is the most common choice
[11:17] <burek> is there a place where things can be proposed to a vote? :)
[11:18] <burek> I'd like to suggest -map 0 as a default :)
[11:18] <ubitux> ask ffmpeg-devel
[11:18] <burek> oh :) they don't like to be disturbed :)
[11:18] <ubitux> you will have technical explanations on why it's not the case
[11:18] <ubitux> it's an interesting question
[11:18] <ubitux> also, i don't think it will be changed because it will likely break a lot of scripts
[11:19] <burek> when did ffmpeg care about that :D
[11:19] <ubitux> well, we try hard to keep backward compat you know :)
[11:20] <ubitux> ffmpeg kept the old option name when libav dropped them
[11:20] <ubitux> names*
[11:20] <burek> yes, I read about that libav fight
[11:20] <burek> so stupid..
[11:20] <burek> people just got mad at each others like children
[11:20] <burek> and wasted nothing but the time
[11:21] <burek> anyway, it would be a good idea to think about how to make some changes and break the compatibility with previous versions (maybe a new branch or something) because it would be reasonable to redesign some things
[11:22] <burek> and not to keep legacy more then 10-20 years old
[11:22] <ubitux> i don't think the map selection is that bad
[11:23] <burek> it's not bad, it's just not logical when somebody types: ffmpeg -i input -c copy output
[11:23] <burek> a user would think "ffmpeg doesn't work properly"
[11:23] <burek> or "it didn't recognize all my streams"
[11:23] <burek> or something
[11:24] <ubitux> (the -c copy has nothing to do with stream selection! :p)
[11:24] <ubitux> but ok for ffmpeg -i input output
[11:24] <dalkor> need help with vp6f video codec, getting all sorts of errors when trying to encode
[11:24] <burek> dalkor, there is vp6 encoder in ffmpeg?
[11:24] <dalkor> no
[11:25] <burek> vp6f is a decoder, right?
[11:25] <dalkor> but it does support decoding of vp6
[11:25] <burek> can you please use pastebin.com, to show your command line and its output?
[11:29] <dalkor> I closed cmd, it didnt start kicking out errors till like 20min in. "ffmpeg -i *.flv -vcodec libx264 -acodec libfaac *.mp4". The error was: "DTS ####, next:####### st:1 invalid droping"
[11:29] <burek> well, either your input is damaged
[11:29] <burek> or there is a bug in ffmpeg :)
[11:41] <elkng> if I use option "-s 320x240" some videos became overscaled in height, how to get exact "320x240" but with black lines at the top and the bottom of video ?
[11:43] <burek> what is your original video size
[11:44] <elkng> 672x272
[11:45] <elkng> it should be 320x129 plus 2x55 pixels black lines at the top and the bottom
[11:50] <elkng> can ffmpeg or mencoder add black lines in the video as effect ?
[11:56] <burek> yes
[11:56] <burek> use video filter for that
[11:57] <burek> http://ffmpeg.org/ffmpeg.html#toc-Video-Filters
[11:58] <burek> you'll most probably want scale and pad
[11:59] <elkng> will it "-croptop 55 -cropbottom 55" help ?
[11:59] <burek> do you know the difference between crop and scale
[12:01] <elkng> I can do transcode with option "-s 320x129" then again with option "-croptop 55 -cropbottom 55" it will add 110 pixels and I got 320x240 ?
[12:01] <burek> ok
[12:02] <burek> take a look at the link I gave you and read
[12:03] <elkng> no "-padtop 55 -padbottom 55", it will add 110 pixels
[12:04] <burek> well, do you want to crop or to pad?
[12:04] <elkng> is http://ffmpeg.org/ffmpeg.html the same as "man ffmpeg" ?
[12:04] <burek> it's not
[12:05] <elkng> rescale to get width at some value and pad to add to height some black space
[12:05] <burek> that's what I thought, but why are you mentioning crop then?
[12:07] <elkng> mixed it with pad
[12:09] <burek> try: ffmpeg -i input.avi -vf 'pad='320:240:0:55' output.avi
[12:15] <burek> or ffmpeg -i input.avi -vf 'scale=320:130,pad='320:240:0:55' output.avi
[12:19] <elkng> do I need "-s 320x240" ?
[12:19] <elkng> no
[12:20] <elkng> 'pad=320:240:0:55' or pad='320:240:0:55' ?
[12:22] <burek> play around with it
[12:22] <burek> and find what's the one you need
[12:36] <elkng> is http://ffmpeg.org/ffmpeg.html the same as /usr/doc/ffmpeg-0.10.2/ffmpeg.html ?
[12:36] <burek> you can use diff to see
[14:25] <nick_opencv> hello, I am having a problem compiling a c++ program with use of opencv library. I issue the command: sudo g++ Program.cpp -I/usr/include/opencv -L/usr/local/lib -lopencv_core -lopencv_imgproc -lopencv_highgui -lm
[14:26] <nick_opencv> and what I get is: /usr/bin/ld: warning: libavcodec.so.52, needed by /usr/local/lib/libopencv_highgui.so, not found (try using -rpath or -rpath-link)
[14:26] <nick_opencv> /usr/bin/ld: warning: libavformat.so.52, needed by /usr/local/lib/libopencv_highgui.so, not found (try using -rpath or -rpath-link)
[14:26] <nick_opencv> /usr/bin/ld: warning: libavutil.so.50, needed by /usr/local/lib/libopencv_highgui.so, not found (try using -rpath or -rpath-link)
[14:26] <nick_opencv> I have lubuntu 12.04 installed
[14:26] <nick_opencv> I have these libraries installed, but the 53 version
[14:26] <burek> you need libav shared lib
[14:27] <burek> try reading prerequisites for opencv
[14:27] <burek> in their documentation
[14:27] <burek> and install what they say it's needed
[14:27] <mancha> any devs around?
[14:27] <nick_opencv> ok, I'll try that
[14:27] <nick_opencv> thanks
[14:27] <burek> :beer: :)
[14:28] <nick_opencv> and if they say I need the 52 version, should I uninstall what I already have?
[14:28] <burek> well, most probably, you'll have to install some -dev package
[14:28] <burek> or some prepackaged ffmpeg
[14:28] <burek> which will install that version of libav*
[14:29] <burek> but, you need to read the docs of opencv
[14:29] <nick_opencv> I need this for a 1000 dollar project, so I will definitely grant you a beer if this works :-)
[14:29] <burek> to see which version of ffmpeg it is
[14:29] <burek> :)
[14:29] <nick_opencv> thanks ;-)
[14:29] <burek> np :)
[14:47] <nop> Hi. Is there a way to extract video from flv container without re-encoding? Would -copy audio and video to a .mp4 do this?
[14:49] <burek> yes
[14:49] <burek> ffmpeg -i input.flv -vcodec copy video.h264
[14:53] <nop> Why .h264? Will it put it in a .mp4 container whith video.mp4? Will it copy the audio too or do I need to precise -acodec copy too?
[14:54] <Tjoppen> the above produces raw h.264. as you say, use -acodec copy video.mp4  if you want the audio too
[14:54] <nop> All right, thanks a lot for your answers!
[14:57] <mancha> do devs come around these prts?
[14:58] <burek> mancha, what exactly do you need
[14:58] <mancha> to report a bugeroo
[14:59] <burek> do you know how to use trac?
[14:59] <burek> +google ffmpeg bug report
[14:59] <mancha> nope
[14:59] <burek> oh well
[14:59] <burek> do you know how to post on a forum
[14:59] <mancha> that's a helpful answer. do you have an answer to my actual question now?  :)
[15:00] <burek> well, I do, but you won't like it :)
[15:00] <mancha> mucho simple, do any devs come to this channel? (yes/no) and if yes, who?
[15:00] <burek> yes they do :)
[15:00] <mancha> anya nicks i should look out for for when they awaken from the encoding slumber?
[15:01] <burek> now, that's another question :)
[15:01] <mancha> yes it is
[15:01] <mancha> which is why i asked it in a separate line
[15:02] <burek> try #ffmpeg-devel
[15:02] <mancha> thank you kindly
[15:02] <burek> yw
[15:38] <technologov_> hi ! Is there any experimental H.265 encoder flying on the net ?
[15:41] <JEEB> there's the reference implementation
[15:41] <JEEB> HM
[15:41] <technologov_> where?
[15:41] <JEEB> http://hevc.kw.bbc.co.uk/
[15:41] <JEEB> you'll want the git subversion mirror
[15:41] <JEEB> and http://hevc.kw.bbc.co.uk/git/w/jctvc-hm.git there
[15:42] <technologov_> ok, thx
[16:28] <Diogo> hi one question to loop a filename in a input this is possible
[16:28] <Diogo> Command: /servers/ffmpeg/bin/ffmpeg -re -i FILENAME.mp4 -flags +loop  -isync  -vcodec copy  -acodec libfaac -ab 128000 -ar 32000 -f flv "rtmp://SERVER/live/STREAM_NAME"
[16:29] <[FAIL]> How can i say per 001.jpg to 002.jpeg should have 4 seconds delay? this is very fast: $ ffmpeg -r 10 -b 1800 -i %03d.jpg test1800.mp4
[16:29] <burek> remove -flags +loop
[16:29] <burek> and add -loop 1
[16:29] <burek> and put it before -i
[16:29] <Diogo> thanks burek
[16:29] <burek> servers/ffmpeg/bin/ffmpeg -loop 1 -i FILENAME.mp4 -isync  -vcodec copy  -acodec libfaac -ab 128000 -ar 32000 -f flv "rtmp://SERVER/live/STREAM_NAME"
[16:30] <burek> [FAIL], use -r 1/4
[16:30] <[FAIL]> burek, WOW - thanks Guru.
[16:30] <burek> :beer: :)
[16:31] <burek> Diogo, -ab 128000 is == -ab 128k
[16:35] <Diogo> burek: Option loop not found.
[16:43] <[FAIL]> burek, $ ffmpeg -sameq -r 1/4 -b 528000 -i %03d.jpg 4000.avi   the quality is not 100% accurate of my jpg files while doing that.
[16:47] <[FAIL]> When i apply that for this picture, its not making 100% quality to .avi file. http://81.164.41.138:7007/tmp/001.jpg
[17:02] <[FAIL]> Will this help? $ ffmpeg -r 1/4 -i %03d.jpg -b 5000 -vcodec mjpeg -qscale 1 4000.avi
[17:02] <[FAIL]> zero quality loss i mean.
[17:05] <sacarasc> Try removing -b blah and chang it to -vcodec copy
[17:05] <sacarasc> *changing
[17:15] <[FAIL]> YES - much better $ mplayer http://81.164.41.138:7007/tmp/jpeg.avi
[18:07] <rburkat> If I'm using 2 pass encoding with ffmpeg to 3 different quality/screen sizes, can I share the first pass log data to speed up the process?
[18:09] <Mavrik> no you can't :\
[19:01] <thang> so i encoded a video from h264 to vp8, and can play it back in chrome but in firefox the first 5 seconds are unplayable
[19:02] <thang> the command: http://pastebin.com/VUuGkgRi and the actual webm file: http://tinyurl.com/7af7bwk
[20:37] <poropeked> As in Linux to make the time stamp in the ffserver?
[20:39] <poropeked> pliz help ...
[20:54] <KeitaroNL> good day
[20:54] <KeitaroNL> I'm running into a little trouble trying to install ffmpeg on centos, could someone please help me out?
[20:55] <KeitaroNL> to expand on my question, I tried installing through subversion, but I keep running into "ERROR: libx264 version must be >= 0.99."
[20:55] <KeitaroNL> so, I googled it and it stated that I had to git it instead.. so I did
[20:55] <sacarasc> Subversion hasn't been used for ffmpeg since the beginning of last year.
[20:56] <KeitaroNL> but now it says ERROR: libx264 version must be >= 0.118.
[20:56] <KeitaroNL> yes, that is what the google message said as well
[20:56] <sacarasc> You'll need to get a recent x264, probably compile it from git.
[20:57] <KeitaroNL> hmm okay
[20:57] <KeitaroNL> I'll try that then
[20:57] <KeitaroNL> thank you
[21:03] <KeitaroNL> thanks sacarasc!
[21:03] <KeitaroNL> it works now
[21:03] <KeitaroNL> I'm gonna setup ffmpeg-php now, should I also get that from github or something?
[21:04] <sacarasc> ffmpeg-php won't work with a new ffmpeg.
[21:06] <burek> why do people use ffmpeg-php ?
[21:06] <burek> I mean what's its purpose
[21:06] <sacarasc> Put it on a webserver and try to be YouTube.
[21:07] <KeitaroNL> not exactly :P
[21:07] <KeitaroNL> my website streams video's amongst others and makes snapshots
[21:07] <Tjoppen> why not just system()?
[21:07] <KeitaroNL> for those snapshots, I need ffmpeg-php
[21:07] <KeitaroNL> the script is configured that way
[21:08] <sacarasc> ffmpeg-php hasn't been updated in so long, you'll probably have to downgrade ffmpeg a lot.
[21:08] <KeitaroNL> ah damn..
[21:08] <KeitaroNL> hmm thanks, I'll give it a shot.. and otherwise.. try wgetting and untarring
[21:09] <burek> let me clear up my question.. why use ffmpeg-php when you have ffmpeg cmd tool?
[21:09] <burek> what's the advantage of ffmpeg-php
[21:09] <KeitaroNL> the script I use for my system makes use of ffmpeg-php
[21:10] <burek> that's the only reason?
[21:10] <KeitaroNL> I'm not that tech savy or anything, this is my first time installing ffmpeg manually (instead of a shell script or something)
[21:10] <KeitaroNL> yes, I believe so
[21:11] <Tjoppen> it seems to be doing what you could be accomplished with a small shell script
[21:11] <Tjoppen> -you
[21:11] <KeitaroNL> I'm afraid I don't know how to set something like that up :(
[21:12] <burek> well, that will leave you behind everyone else
[21:13] <burek> considering ffmpeg-php is not being updated
[21:13] <Tjoppen> WIDTH=`ffprobe $1 blahblah | grep -m1 ^width | sed yaddayadda`
[21:13] <Tjoppen> etc.
[21:13] <KeitaroNL> hmm
[21:14] <KeitaroNL> so basically, you're telling me that even if the script is configured to make use of ffmpeg-php
[21:14] <KeitaroNL> I could find a way around it
[21:14] <KeitaroNL> and not by modifying the script itself
[21:14] <KeitaroNL> but by adding something else?
[21:14] <burek> your scripts use "the script", which uses ffmpeg-php, which uses ffmpeg
[21:15] <burek> why not making your script to use ffmpeg directly?
[21:15] <Tjoppen> we need to go deeper
[21:15] <KeitaroNL> because it is not my script
[21:16] <KeitaroNL> and it is written in a language I know nothing about
[21:16] <KeitaroNL> I cannot modify it
[21:16] <KeitaroNL> unless I hire a perl expert or something like that do it for me
[21:16] <KeitaroNL> that
[21:16] <KeitaroNL> and a part of it is encoded to protect the script against piracy and such
[21:16] <burek> then "feed the pigs and don't touch anything" :)
[21:16] <poropeked> As in Linux to make the time stamp in the ffserver?
[21:17] <KeitaroNL> @burek: haha, yeah I guess so
[21:17] <KeitaroNL> I wouldn't be using it, if it wasn't for the case that it was the best script on the market for its purpose
[21:17] <KeitaroNL> either that or have something custom developed
[21:17] <burek> poropeked?
[21:18] <poropeked> that
[21:18] <poropeked> what?
[21:18] <burek> poropeked, can you rephrase your question please :)
[21:19] <poropeked> should be date and time stamp to put on a video
[21:21] <KeitaroNL> okay
[21:21] <KeitaroNL> I'm reinstalling FFMPEG
[21:21] <KeitaroNL> this tut should probably work
[21:21] <KeitaroNL> http://www.phpfox.com/kb/article/337/install-ffmpeg-mplayer-mencoder-ffmpeg-php-on-centos-5-x-automatically/
[21:21] <KeitaroNL> as for the script I make use of
[21:21] <KeitaroNL> if anyone is interested
[21:21] <KeitaroNL> http://sibsoft.net/xfilesharing.html
[21:23] <KeitaroNL> I  make use of the Video Mod which comes with other instructions and says that one'd need ffmpeg-php
[21:23] <burek> poropeked, you need to apply video filters in ffserver?
[21:23] <burek> KeitaroNL, watch about ffmpeg version that your script uses
[21:24] <burek> newer versions of ffmpeg might have changed API and render your script useless
[21:24] <poropeked> yes
[21:25] <KeitaroNL> hmm
[21:25] <KeitaroNL> after installing, I'll upload a mp4 file and test it out
[21:25] <burek> poropeked, use -vf when running ffmpeg not ffserver
[21:25] <KeitaroNL> a question though
[21:26] <KeitaroNL> being the noob that I am
[21:26] <burek> ffmpeg -i ... -vf ... http://localhost:8090/feed1.ffm
[21:26] <KeitaroNL> why do all of the video based scripts out there (or at least, the majority of them)
[21:26] <KeitaroNL> make use of ffmpeg-php?
[21:26] <KeitaroNL> if there are better/other alternatives?
[21:26] <KeitaroNL> easier to set up or something?
[21:26] <burek> KeitaroNL, I don't know, I don't use ffmpeg-php at all
[21:26] <burek> all my web scripts use pure ffmpeg
[21:26] <burek> without any problem
[21:27] <KeitaroNL> are your scripts custom made?
[21:27] <KeitaroNL> or do you make use of commercial/free scripts?
[21:28] <burek> I make all my scripts :)
[21:28] <KeitaroNL> :P
[21:28] <KeitaroNL> nice
[21:28] <burek> I'm to scared to rely on other people's bugs and errors
[21:28] <burek> :)
[21:29] <KeitaroNL> haha, true
[21:38] <poropeked> ffmpeg: missing argument for option '-vf'
[21:40] <burek> poropeked, http://ffmpeg.org/ffmpeg.html#Video-Filters
[21:41] <burek> -vf 'FILTER_NAME=FILTER_PARAMETERS,ANOTHER_FILTER=ANOTHER_PARAMETERS,...'
[21:42] <burek> you most probably want -vf 'drawtext=...'
[21:53] <poropeked> ffmpeg -i /var/www/webcam_yesterday.flv -vf "fieldorder=bff"
[21:53] <poropeked> ffmpeg: unrecognized option '-vf'
[21:53] <burek> can you please use pastebin.com, to show your command line and its output?
[21:54] <poropeked> http://pastebin.com/hj4t8YST
[22:04] <poropeked> oh
[22:06] <burek> poropeked: FFmpeg version SVN-r0.5.6-4:0.5.6-3, Copyright (c) 2000-2009
[22:06] <burek> 2009.. now it's 2012, you know :)
[22:06] <burek> it might be time to update :)
[22:13] <poropeked> TCP connection to 192.168.1.39:8090 failed: Connection refused
[22:13] <burek> can you please use pastebin.com, to show your command line and its output?
[22:15] <poropeked> http://pastebin.com/ZBXfN83P
[22:15] <burek> did you start ffserver
[22:17] <mbradshaw> I have a question regarding frame PTS values for audio streams and the stream's sample rate
[22:18] <mbradshaw> For example, I have an aac stream, and ffmpeg gives this stream a timebase of 1/44100 (which is what I would expect, because that's its sample rate as well)
[22:19] <mbradshaw> The first frame contains 1024 samples and has a PTS of 0, but for some reason the second frame has a PTS of 1014... what
[22:19] <mbradshaw> what's up with the 10 overlapping samples between the first and second frames?
[22:20] <Mavrik> hmm
[22:20] <Mavrik> mbradshaw, do they contain values?
[22:21] <mbradshaw> what do you mean by "they"? the frames' PTS?
[22:21] <Mavrik> sorry, the overlapping values
[22:22] <Mavrik> meaning that the decoder didn't emit samples for the last 10
[22:22] <Mavrik> in the first frame
[22:22] <juanmabc> does not work that way
[22:22] <mbradshaw> How can you check if there aren't values in the samples?
[22:22] <j-b> mbradshaw: hello
[22:22] <mbradshaw> Hi j-b
[22:22] Action: j-b very interested in your jp2k patch.
[22:23] <j-b> mbradshaw: question: does it support the XYZ color space or something similar that is used in DCPs ?
[22:24] <juanmabc> mbradshaw: variable bit rate, timebase is the minimal unit, and there is codec timebase and stream timebase, stream timebase should be the good
[22:24] <sgfgdf> hello, guys! i have a mp4 file with Video: h264... and Audio: aac (mp4a / 0x6134706D), 44100 Hz, stereo, s16, 151 kb/s. i want to extract only the audio as mp3 or m4a. what is the better format if i want to keep the audio quality? how should i grab the audio without loosing much quality?
[22:25] <mbradshaw> juanmabc: the stream timebase is 1/44100, which is the one I'm referencing
[22:25] <burek> sgfgdf, -acodec copy
[22:25] <mbradshaw> Mavrik: I'll check if the overlapping values are the same...
[22:26] <mbradshaw> j-b: libopenjpeg only has options for RGB, YUV, and gray (though it does have an option for "unknown")
[22:26] <burek> sgfgdf, ffmpeg -i input.mp4 -acodec copy -vn output_audio.mp4
[22:26] <sgfgdf> burek, what should be the result .m4a file or .aac?
[22:26] <sgfgdf> ops
[22:26] <burek> sgfgdf, you can use which ever output format you want
[22:27] <burek> sgfgdf, ffmpeg -i input.mp4 -acodec copy -vn output_audio.aac
[22:27] <mbradshaw> j-b: decoders would have to guess the color space since it would be set to "unknown," and I doubt they would be able to properly guess it
[22:27] <sgfgdf> burek, normaly only audio file which is .m4a is encoded with aac or..?
[22:27] <burek> then output.m4a
[22:28] <mbradshaw> j-b: do you have a XYZ JP2K video?
[22:28] <j-b> mbradshaw: I could, very easily.
[22:28] <sgfgdf> burek, is it reasonable to use mp3 or m4a is better?
[22:28] <j-b> mbradshaw: so openjpeg does the XYZ -> YUV conversion?
[22:28] <burek> sgfgdf, if you want mp3 you'll have to re-encode audio
[22:28] <Mavrik> sgfgdf, for mp3 you'd have to reencode audio
[22:28] <Mavrik> sgfgdf, for m4a you just extract it without changing it
[22:29] <Mavrik> (since it's already aac)
[22:29] <burek> sgfgdf, if you want to keep the original audio, use -acodec copy and save in whichever format you like (mp4, aac, m4a, ...)
[22:30] <mbradshaw> j-b: if you see http://www.openjpeg.org/libdoc/openjpeg_8h.html#a99039de6017a1f64312480f3f4ea448c
[22:30] <sgfgdf> burek, yes, okay. my question was more oriented about those two formats. do people prefer nowadays to store their music in m4a files or they prefer mp3? is m4a any better (worth)?
[22:30] <mbradshaw> j-b: you will see it doesn't support XYZ
[22:31] <j-b> mbradshaw: that is what I was afraid of :)
[22:31] <j-b> mbradshaw: thanks a lot.
[22:31] <mbradshaw> j-b: no problem. hopefully they'll add support for more color spaces
[22:31] <burek> sgfgdf, you should really read more about aac+
[22:31] <burek> it's an apex of streaming audio today
[22:32] <j-b> mbradshaw: well, the issue is more about who is doing the conversion and how to make it faster :)
[22:33] <mbradshaw> j-b: are you converting from xyz to yuv?
[22:33] <j-b> mbradshaw: I am not doing anything! :)
[22:33] <j-b> mbradshaw: my guess is that openjpeg does it :)
[22:33] <cbsrobot_> j-b: nah don't think so
[22:34] <j-b> my issue now is that ffmpeg outputs RGB16, which I do not like.
[22:34] <j-b> cbsrobot_: oh?
[22:34] <mbradshaw> j-b: so you'd prefer if it output xyz?
[22:34] <sgfgdf> burek, so since it is better for streaming does it means it is pointless to store local music files encoded with aac and use mp3 or not?
[22:34] <j-b> mbradshaw: I need fast :)
[22:34] <mbradshaw> j-b: ah, so the encoding is slow?
[22:35] <burek> sgfgdf, I'll just say that I converted all of my mp3s into aac+
[22:35] <burek> it's that good!
[22:35] <j-b> decoding is slow
[22:35] <j-b> openjpeg is slow
[22:35] <burek> and 10x smaller then hi-q mp3
[22:35] <sacarasc> And double the loss!
[22:36] <mbradshaw> j-b: that's very true. I think the decoder has been multithreaded, so that might help (you might already be using multiple threads though)
[22:36] <mbradshaw> j-b: part of the issue though is that jpeg2000 just isn't a great format for decoding/encoding fast
[22:36] <j-b> well, I receive a lot of requests for correct mxf/jp2k decoding and many colourspace issues are present
[22:37] <sgfgdf> burek, btw do you loose quality while doing the conversion?
[22:37] <j-b> amd that openjpeg has little ASM
[22:37] <burek> I didn't notice
[22:37] <sacarasc> sgfgdf: Going to a lossy format, you always lose quality.
[22:37] <burek> sgfgdf, perceived quality is way better then mp3
[22:37] <burek> try yourself, download some flac from the opera
[22:37] <mbradshaw> j-b: openjpeg could definitely be sped up, I'm not arguing against that :)
[22:37] <burek> and convert it to 320kbps mp3 and 32kbps aac+
[22:37] <burek> and listen for the differences
[22:38] <j-b> mbradshaw: I can find money for bounties for that.
[22:38] <burek> people confuse 48 kbps aac+ with uncompressed audio!
[22:39] <mbradshaw> j-b: the openjpeg team may be interested in bounties...
[22:39] <sgfgdf> burek, did you make your aac music from flac or you converted from mp3 -> aac?
[22:39] <burek> mp3->aac
[22:39] <j-b> well, if we have a YUV output first, it will help
[22:40] <mbradshaw> j-b: do you mean just decoding yuv jp2k video?
[22:40] <sgfgdf> burek, hm, so in your case you shouldn't get any better quality while listening the new aac files, right?
[22:41] <burek> sgfgdf, you can't get better quality without some interpolations/remastering or such
[22:41] <sgfgdf> burek, only if you get them from flac may be will notice better aac compared to mp3
[22:41] <j-b> mbradshaw: now, many mxf output from libavcodec in RGB16/channel that VLC does not support and therefore is green
[22:41] <burek> simply converting to another format, you can only loose quality
[22:41] <burek> sgfgdf, if you want to compare qualities, then use flac to produce mp3 and aac+ and compare it
[22:41] <burek> sgfgdf, and if you want just to save your hdd space, convert mp3 directly to aac+
[22:42] <burek> that won't loose any perceived quality
[22:42] <sgfgdf> burek, did you do some things or just want to conert from mp3 to another format?
[22:42] <sgfgdf> burek, ah okay
[22:42] <mbradshaw> j-b: So what exactly is it you'd like to do with ffmpeg/libavcodec?
[22:43] <j-b> anything improved on the openjpeg  integration makes less complaints, makes me happy
[22:43] <sgfgdf> burek, can you show a command which you use to convert yours mp3 files?
[22:44] <sgfgdf> burek, if it doesn't bothers you.
[22:44] <burek> sgfgdf, ffmpeg -i a.mp3 -acodec libaacplus -ar 44100 -ac 2 -ab 32k b.aac
[22:45] <mbradshaw> j-b: I see. Well, I just finished with school for the summer so I finally have more time to work on the openjpeg encoder/decoder. Now that you bring it up though, I may also try to work with the openjpeg team to try and optimize things
[22:45] <burek> sgfgdf, http://ffmpeg.gusari.org/viewtopic.php?f=25&t=38
[22:46] <j-b> mbradshaw: well, if needed, I can send money on that general direction.
[22:46] <j-b> mbradshaw: aim is to be able to decode MXF 2k on a QuadCore machine with a LOT of ram with VLC>
[22:48] <mbradshaw_> j-b: what size of video? 1080p? smaller?
[22:48] <j-b> 2k
[22:48] <j-b> the ones from the DCP
[22:49] <mbradshaw_> j-b: that's going to need to be hella fast ha
[22:49] <sgfgdf> burek, thank you a lot for your help!
[22:49] <j-b> mbradshaw_: well, yeah :)
[22:51] <mbradshaw_> j-b: all of the DCP videos? that is, you need 2k 60fps real time playback?
[22:52] <j-b> 2k 30fps
[22:56] <mbradshaw_> j-b: alright... I guess I can start investigating this
[22:57] <j-b> mbradshaw_: I can provide samples
[22:57] <j-b> mbradshaw_: and I can gather money on that
[22:57] <mbradshaw_> j-b: I'd say that's a good starting point
[22:57] <burek> sgfgdf, :beer: :)
[23:03] <mbradshaw_> j-b: if you want to, you can reach me by email and we can discuss this further
[23:03] Action: sgfgdf sends :beer: to burek (only virtual so far)
[23:03] <j-b> mbradshaw_: I will
[23:04] <sgfgdf> burek, what is the official website of libaacplus, because it seems that i only have faac? is it really this address -- http://217.20.164.161/~tipok/aacplus/ ?
[23:05] <burek> sgfgdf, yes
[23:12] <burek> one of ffmpeg-php requirement is ffmpeg-0.4.9_pre1 or higher :)
[23:13] <sgfgdf> burek, is aac+ any better then the other codecs like faac or other? i saw the last version is from 2010.
[23:14] <burek> faac can't produce aac+
[23:14] <burek> only AAC-LC
[23:15] <sgfgdf> burek, and aac+ is better? is there any other library that produce aac+ except libaacplus?
[23:16] <burek> NeroAacEnc
[23:16] <sgfgdf> burek, is that better than aac+ -- HE-AAC?
[23:17] <burek> take a look at the image on the right
[23:21] <mbradshaw_> Alright, I'm back with my question about packet PTS values and audio... so my audio stream has a sample rate and a time base of 1/44100. The first frame's packet's PTS value is 0 and the frame has 1024 samples. The second frame's packet's PTS value is 1014, and it too has 1024 samples. But the PTS of the second frame looks 10 too early, as it's overlapping the first frame... how can I handle this properly in my decoding library?
[23:21] <sgfgdf> burek, ah so it is a name for aac+.
[23:21] <burek> yes
[23:22] <burek> aac+ = he-aac
[23:22] <sgfgdf> burek, thanks again!
[23:22] <burek> SBR = v1, SBR+PS = v2
[23:22] <burek> :beer: :)
[23:22] Action: sgfgdf sends :beer: (2) to burek
[23:22] <burek> ^^
[23:23] <sgfgdf> you probably will get drunk if i ask you a few more questions so i should stop :)
[23:25] <mbradshaw_> anyone have any ideas?
[23:26] <burek> :)
[23:41] <Orphis> Is there any container / codec that can handle multiple soundtracks for the same video efficiently ? If it's mostly the same music and just a few seconds of dialog that change, there shouldn't be ALL the audio for every language for example
[23:42] <Orphis> If there was a way to do the same for video too, it would be great (same video everywhere but a few texts that change)
[23:44] <burek> Orphis, maybe you could implement such a codec? :)
[23:44] <Orphis> Or maybe not :P
[23:44] <burek> or that :)
[23:44] <Orphis> I just wanted to know if something similar existed
[00:00] --- Thu May  3 2012


More information about the Ffmpeg-devel-irc mailing list