[Ffmpeg-devel-irc] ffmpeg.log.20120206
burek
burek021 at gmail.com
Tue Feb 7 02:05:02 CET 2012
[02:18] <Anthony> I would like to make a high quality mp4 from an mkv file
[02:18] <Anthony> what kinda command should i use
[02:18] <Anthony> ?
[02:21] <Anthony> bcoudurier »
[02:21] <Anthony> supercurio »
[02:21] <Anthony> `md »
[02:21] <Anthony> iive »
[02:21] <Anthony> anyone
[02:23] <iive> Anthony: don't highlight people when you don't get answer
[02:23] <Anthony> ok
[02:23] <iive> especially when you ask question that google can also answer.
[02:24] <Anthony> i should be more specific sorry
[02:24] <Anthony> what should I put for -vpre
[02:25] <Anthony> i just want the max quality
[02:27] <Anthony> i came up with a command but not sure if it's the best
[02:27] <Anthony> ffmpeg -i input.mkv -s 1280x692 -r 24000/1001 -b 2M -bt 5M -vcodec libx264 -vpre max -acodec libfaac -ac 2 -ar 48000 -ab 192k output.mp4
[04:32] <pills> I have 4 videos I am trying to combine into a single video. This is easy enough to do using the video filter options. My problem is the input videos don't seem to decode properly (first frame is dropped). This seems to be due to how they were encoded wrt timecodes. I can fix this problem for a single video by adding the -vsync 0 option however when combined with the video filters only the first video seems to inherit the vsync option. Is
[04:32] <pills> there anyway around this (reencoding is fine and if so what would be the "right" way)? Thanks.
[05:11] <XiL> Was wondering if Debian GNU/Linux 3.1 supports Turbo Boost. cat /proc/cpuinfo is just a label afaik.
[05:11] <XiL> Got quite a lot of rendering to do
[05:14] <XiL> using a i7 2700k
[05:19] <drv> turbo boost doesn't really matter if you are doing multi-threaded encoding anyway
[05:19] <drv> since it only takes effect when only a single core is loaded
[05:22] <XiL> drv true but some parts of my script (grabs the source ffmpeg VLC and kdenlive with all the bells and whistles and compiles them to play nice) would be faster.
[05:23] <drv> i think it is automatically done by the CPU, but i don't know for sure
[05:24] <XiL> drv a lot of ppl say that is the case and others swear it not. At this point idk.
[05:24] <drv> well, you could always try it out and measure the cpu clock speed :)
[05:25] <XiL> time ffmpeg -i thing yeah true.
[06:44] <Perdignus> Hello - I'm converting some flac files to mp3 and it seems like the tag mapping is not working right
[06:44] <Perdignus> for instance, the TITLE tag from the flac file becomes the (260) --> ffmpeg -i 01\ -\ Here\ I\ Am\ \(Come\ And\ Take\ Me\).flac -y -ab 128k -ac 2 -ar 48000 /var/work/KidsMP3s/test.mp3 ffmpeg version N-37576-ge81e0b9 Copyright (c) 2000-2012 the FFmpeg developers built on Feb 5 2012 22:48:40 with gcc 4.5.3
[06:44] <Perdignus> configuration: --enable-ffmpeg --mandir=/usr/share/man --enable-pthreads --prefix=/usr --libdir=/usr/lib64 --shlibdir=/usr/lib64 --enable-shared --cc=x86_64-pc-linux-gnu-gcc --enable-libfaac --enable-zlib --disable-static --enable-gpl --enable-version3 --enable-postproc --enable-avfilter --disable-stripping --disable-debug --disable-doc --disable-network --disable-vaapi --enable-libmp3lame --enable-libvo-aacenc --enable-l
[06:44] <Perdignus> libavutil 51. 37.100 / 51. 37.100
[06:44] <Perdignus> libavcodec 54. 1.100 / 54. 1.100
[06:44] <Perdignus> libavformat 54. 0.100 / 54. 0.100
[06:44] <Perdignus> libavdevice 53. 4.100 / 53. 4.100
[06:44] <Perdignus> libavfilter 2. 61.100 / 2. 61.100
[06:44] <Perdignus> libswscale 2. 1.100 / 2. 1.100
[06:44] <Perdignus> libswresample 0. 6.100 / 0. 6.100
[06:44] <Perdignus> libpostproc 52. 0.100 / 52. 0.100
[06:44] <Perdignus> [flac @ 0x1a7a320] max_analyze_duration 5000000 reached at 5015510
[06:44] <Perdignus> Input #0, flac, from '01 - Here I Am (Come And Take Me).flac': Metadata:
[06:45] <Perdignus> ALBUM : Labour Of Love II
[06:45] <Perdignus> ARTIST : UB40
[06:45] <Perdignus> GENRE : Reggae
[06:45] <Perdignus> TITLE : Here I Am (Come And Take Me)
[06:45] <Perdignus> track : 1
[06:45] <Perdignus> Duration: 00:04:00.18, bitrate: 823 kb/s
[06:45] <Perdignus> Stream #0:0: Audio: flac, 44100 Hz, stereo, s16
[06:45] <Perdignus> Output #0, mp3, to '/var/work/KidsMP3s/test.mp3':
[06:45] <Perdignus> Metadata:
[06:45] <Perdignus> TALB : Labour Of Love II
[06:45] <Perdignus> TPE1 : UB40
[06:45] <Perdignus> TCON : Reggae
[06:45] <Perdignus> TIT2 : Here I Am (Come And Take Me)
[06:45] <Perdignus> TRCK : 1
[06:45] <Perdignus> TSSE : Lavf54.0.100
[06:45] <Perdignus> Stream #0:0: Audio: mp3, 48000 Hz, 2 channels, s16, 128 kb/s
[06:45] <Perdignus> Stream mapping:
[06:45] <Perdignus> Stream #0:0 -> #0:0 (flac -> libmp3lame)
[06:45] <Perdignus> Press [q] to stop, [?] for help
[06:45] <Perdignus> size= 1397kB time=00:01:29.37 bitrate= 128.0kbits/s
[06:45] <Perdignus> video:0kB audio:1396kB global headers:0kB muxing overhead 0.037342%
[06:45] <Perdignus> Received signal 2: terminating.
[06:45] <Perdignus> crap
[06:45] <Perdignus> sorry
[06:46] <Perdignus> as I was saying, again I'm very sorry about that, the TITLE tag becomes the TALB tag and so forth
[09:52] <vlad77> hi2all
[09:53] <vlad77> have anyone tried ismv support in v0.10?
[10:26] <ubitux> what's the most common way of encoding aac nowadays?
[10:26] <ubitux> libfaac is deprecated afaik, libaacplus? libvoaac?
[10:28] <JEEB> I don't think libfaac is deprecated, it just shares status with libaacplus
[10:29] <JEEB> as in, "you can't give the binaries out"
[10:29] <JEEB> libvoaac seems to be... something
[10:29] <JEEB> generally I'd say that on linux you might as well stick to neroaacenc
[10:30] <ubitux> < JEEB> libvoaac seems to be... something // what's that supposed to mean? :)
[10:30] <JEEB> on mac/win you have a selection of either neroaacenc or one of those QT-based command line encoders
[10:30] <Mavrik> ubitux, libvo_aacenc is the replacement library
[10:30] <ubitux> replacement of what?
[10:30] <JEEB> it really isn't a full replacement I hear quality-wise + it can only do 2ch
[10:30] <Mavrik> of libfaac - libfaac has licensing problems
[10:30] <JEEB> but it's the only non-ffaac encoder that is (L)GPL compatible
[10:30] <Mavrik> JEEB, it's pretty much same reference design, but you're right about 2-channel
[10:31] <JEEB> Mavrik, I think they used separate reference encoders
[10:31] <JEEB> but yes, reference code both
[10:31] <JEEB> just that one was licensed and relicensed to apache by someone actually capable of it
[10:31] <Mavrik> mhm
[10:31] <JEEB> also, I _think_ libvo-aacenc can only do CBR
[10:31] <JEEB> not sure
[10:32] <Mavrik> sadly, neroaacenc can't be linked with ffmpeg :\
[10:32] <JEEB> yes, and as it would be closed source it'd be undistributable
[10:32] <JEEB> just like QT's aac encoder would be on windows
[10:32] <JEEB> on OS X QT is a part of the OS so it might fly
[10:33] <JEEB> of course, neroaacenc doesn't have a library to begin with
[10:33] <JEEB> only executables
[10:33] <JEEB> QT's actually has an API
[10:33] <JEEB> neroaacenc doesn't need QT to be installed of course, and there is a 32bit linux binary
[10:33] <JEEB> so yeah
[10:34] <Mavrik> so yeah, getting quality AAC audio in automated video encoding is problematic as hell
[10:36] <JEEB> Well, if that's all you're looking after you either use ffaac or make a ffmpeg patch for the QT encoder .-. But that of course means that you have to build it on every PC and can't distribute it.
[10:36] <Mavrik> JEEB, my use case is mostly automated transcoding on servers, so non-redistributable license doesn't bother me all that much
[10:37] <Mavrik> right now I'm using libfaac yeah, getting AAC+ on higher bitrates would probably prove beneficial (most content is voice - lectures)
[11:20] <rbrt> hello!
[11:20] <rbrt> is there a way to use audio filters with ffplay ?
[11:22] <rbrt> there is an option "-vf" for video filters, but "-af" for audio does not work
[11:23] <ubitux> yes, no -af option yet
[11:23] <ubitux> but you can use ffplay -f lavfi 'amovie=...'
[11:25] <rbrt> but this only plays the audio track.
[11:26] <ubitux> you can append filter
[11:26] <ubitux> rbrt: what are you trying to do?
[11:26] <ubitux> ffplay -f lavfi 'amovie=a.wav,aresample=8000,aconvert=...,pan=...'
[11:27] <rbrt> i have a .avi file with "5.1" ac3, but i have only stereo output, so i want to use the pan filter
[11:27] <ubitux> -ac 2 should downmix it properly
[11:28] <ubitux> maybe you want custom levels?
[11:29] <ubitux> ffmpeg -i in.avi -f lavfi 'amovie=in.avi,pan=...' out.avi should do the trick otherwise
[11:29] <ubitux> ffmpeg -i in.avi -f lavfi -i 'amovie=in.avi,pan=...' out.avi
[11:29] <ubitux> sorry
[11:32] <rbrt> well, yes, i tried it, but that does not work with ffplay. maybe i have to convert it first, and play the converted?
[11:33] <ubitux> mmh maybe ffplay -f lavfi -i 'movie=in.avi[out0];amovie=in.avi,pan=...[out1]' or something?
[11:36] <rbrt> there is an "audio filters" section in ffplay man page, but there is no option to actually use them with ffplay. is this right?
[11:37] <ubitux> isn't the example above making use of the audio filters?
[11:39] <ubitux> ./ffplay -f lavfi 'movie=a.mov [out0]; amovie=a.mov,aresample=8000 [out1]'
[11:39] <rbrt> ffplay -f lavfi "amovie=foo.avi[out0];movie=foo.avi[out1];pan='stereo: c0=FL : c1=FR'" says its expecting channel name, but found 'FL'
[11:39] <ubitux> i just tested that and it seems to work
[11:39] <ubitux> mmh
[11:40] <ubitux> iirc the pan filter has an issue with channel names
[11:40] <ubitux> try c0=c1 : c1=c0
[11:41] <ubitux> rbrt: btw, aren't you just trying to do -ac 2 ?
[11:41] <rbrt> that says : "Not enough inputs specified for the "pan" filter"
[11:41] <ubitux> you didn't write the filter properly
[11:41] <rbrt> i tryed "-ac 2", but its the same result without any filter option
[11:42] <ubitux> "movie=foo.avi[out0]; amovie=foo.avi,pan='stereo: c0=FL : c1=FR'[out1]"
[11:42] <ubitux> try this
[11:42] <ubitux> or just "movie=foo.avi[out0]; amovie=foo.avi,pan='stereo: c0=c0 : c1=c1'[out1]"
[11:42] <ubitux> this will filter out the other channels btw
[11:45] <rbrt> it seems that the 5.1 isthe last one played only audio, no video. thank you for your time, but i will convert that movie with ffmpeg. it seems to be lot easier.
[11:45] <rbrt> the last one played only audio, no video. thank you for your time, but i will convert that movie with ffmpeg. it seems to be lot easier.
[11:45] <rbrt> sorry
[15:17] <bgmarete> Hi. ffmpeg gives me an error when I try to give it the `-profile' option, as in `-profile main' when using libx264. What may I be doing wrong?
[15:18] <wl2776> Hi all. Is it possible to remux raw h264 video to matroska? I'm getting error "Can't write packet with unknown timestamp". Tried -fflags +genpts -vf "setpts" - no effect. Here is paste: http://pastebin.com/q3htv5wF
[15:18] <Mavrik> bgmarete, there's a parameter collision with libfaac
[15:18] <bgmarete> Oh!
[15:18] <Mavrik> bgmarete, use "-profile:v main" or if you have older ffmpeg "-vprofile main"
[15:18] <bgmarete> Mavrik: Thank you very much! Many thanks.
[16:30] <loincloth> hallo
[17:48] <SIFTU> anyone got any tips to keep audio in sync when merging multiple mpg2's for dvd output?
[18:19] <elihubogan> Does anyone have tips on exporting a ProRes422HQ file to something I can use in Cinelerra?
[18:19] <elihubogan> I keep getting files that are <20% of my original, and I think it's got to be lossy
[18:21] <elihubogan> currently trying "ffmpeg -i <infile.mov> -sameq -vcodec mpeg4 <outfile.mov>"
[19:26] <boghog> hi there, is there some known problem with reading video files with mp4/mov containers that indicate absurdly high framerates? (like 1200fps) I have a problem with a file that seems to take ages to transcode, and I notice that un the status line it prints that it has processed over 91000 (ish) frames, even though the output framerate is set to 30 automatically and duration shows 5 minutes, the 'dup' count seems to have a very high number
[19:26] <boghog> too (I guess that might make sense if dup stands for duplicate)
[19:27] <boghog> it shows something like: frame=121047 fps= 85 q=29.0 size= 54506kB time=00:06:43.34 bitrate=1107.0kbits/s dup=108957 drop=0
[19:29] <boghog> http://pastebin.com/PESGP0Tn is the full ffmpeg output
[19:30] <boghog> I guess it has something to do with the mismatch between container and video stream framerate
[19:33] <boghog> it also occurs to me that in the status line, (frame-dup)/29.98 equals roughly 6:43, so it seems to be just processing duplicate frames or something? :s
[20:40] <boghog> managed to sort of fix my problem by passing -vsync 0
[20:40] <boghog> well not sort of, it fixed it entirely
[20:42] <boghog> I'm only a bit worried on what potential sideeffects -vsync 0 might produce in videos that don't need it
[20:42] <boghog> since I'm using ffmpeg for automated conversion of video files
[21:30] <slackyman> hi
[21:31] <slackyman> I found a little buggy behaviour in ffmpeg and ffplay
[21:32] <slackyman> if you compile ffmpeg and ffplay without librtmp the protocol rtmp is still included but if you try to play via rtmp the player freeze
[21:32] <slackyman> so ffmpeg
[21:32] <slackyman> without any notice or alert
[21:32] <slackyman> it simply hangs
[21:38] <slackyman> I was about to report this bug in the forum but I have to wait my account approvation
[21:39] <slackyman> Ok, I read: "The new official forum, that will be located at forum.ffmpeg.org, is being created, and because of that, registrations of new users on this forum has been disabled"
[21:39] <slackyman> but the forum is not yet ready
[21:40] <slackyman> any1 here? :D
[22:14] <slackyman> helloooo
[22:14] <boghog> hi!
[22:14] <slackyman> hi
[22:15] <slackyman> boghog, have U read what I wrote?
[22:15] <boghog> yeah, but not much I can help with
[22:16] <boghog> in fact I only came in here to get help with a problem of my own :D
[22:16] <slackyman> I have another problem
[22:16] <slackyman> maybe we can hellp one each other :D
[22:17] <slackyman> I cannot compile librtmp
[22:17] <slackyman> in linux and mingw32
[22:17] <slackyman> either
[22:17] <slackyman> no matter
[22:17] <boghog> what error?
[22:17] <slackyman> i want to compile librtmp statically linked to polarssl
[22:18] <slackyman> uhm
[22:18] <slackyman> several errors :D
[22:18] <slackyman> can I compile only librtmp with something such as
[22:18] <slackyman> make librtmp?
[22:19] <slackyman> dh.h:75:3: warning: pointer targets in passing argument 3 of 'dhm_calc_secret' differ in signedness [-Wpointer-sign]
[22:19] <slackyman> this is the firts error I can notice
[22:20] <slackyman> I modified tha makefile
[22:20] <slackyman> CRYPTO=POLARSSL
[22:20] <slackyman> SYS=mingw
[22:22] <Gus> Hi. Is it a known bug of recent ffmpeg versions that the features --enable-runtime-cpudetect and --enable-mmx, --enable-mmx2, --enable-sse etc. have become mutualy exclusive?
[22:23] <Gus> By enabling both CPU detect and asm features I get "runtime cpu detection yes" but "MMX enabled no
[22:23] <Gus> MMX2 enabled no
[22:23] <Gus> 3DNow! enabled no" etc.
[22:25] <slackyman> Gus, I'm not an expert in C and C++ programming but I suppose that runtime-cpu-detect excludes any other cpu optimisation is useless
[22:25] <slackyman> or maybe...
[22:25] <slackyman> lemme check a thing
[22:25] <Gus> slackmagic: I think this behaviour of mutual exclusion is new.
[22:26] <slackyman> I don't have this issue
[22:26] <Gus> Typically, enabling MMX, SSE etc. means that the code gets built and is usable within the binary. CPU runtime detection should check upon program start whether the optimized code *may* be used (whether the CPU supports it).
[22:26] <slackyman> yasm yes
[22:26] <slackyman> MMX enabled yes
[22:26] <slackyman> MMX2 enabled yes
[22:26] <slackyman> 3DNow! enabled yes
[22:26] <slackyman> 3DNow! extended enabled yes
[22:26] <slackyman> SSE enabled yes
[22:26] <slackyman> SSSE3 enabled yes
[22:26] <slackyman> AVX enabled yes
[22:27] <slackyman> runtime cpu detection yes
[22:27] <Gus> slackmagic: including specifying --enable-runtime-cpudetect?
[22:27] <Gus> Hmmm.
[22:27] <slackyman> I've re-tryed my ./configure
[22:27] <slackyman> maybe there's something more
[22:27] <slackyman> i.e. I notice other strange behaviours
[22:28] <Gus> I just noticed that this new build of mine is significantly slower on some aspects than earlier versions.
[22:28] <slackyman> if i use --enable-libxavs I have to modify CFLAGS and --extra-libs to get SDL to be recognised
[22:28] <Gus> Which is due to the fact that no hardware optimizations, not even MMX, are used for scaling any more here.
[22:28] <slackyman> else ffplay cannot be compiled
[22:29] <slackyman> yes
[22:29] <slackyman> can I see your configure switches?
[22:29] <pasteeater> Gus: did you --disable-mmx?
[22:29] <slackyman> use pastebin or someone get pissed :D
[22:29] <Gus> pasteeater: not at all.
[22:29] <Gus> pasteeater: as mentioned above, I explicitly enabled all CPU features.
[22:30] <pasteeater> i just entered the channel, so i have little context.
[22:30] <Gus> Version 0.9.1 from January 2012 here, btw.
[22:30] <slackyman> I didn't and they've enabled
[22:30] <slackyman> Gus, try only the --enable-cpu-runtime
[22:30] <Gus> pasteeater: this is my original point: Hi. Is it a known bug of recent ffmpeg versions that the features --enable-runtime-cpudetect and --enable-mmx, --enable-mmx2, --enable-sse etc. have become mutualy exclusive?
[22:30] <slackyman> and what's the result?
[22:31] <Gus> Gus> By enabling both CPU detect and asm features I get "runtime cpu detection yes" but "MMX enabled no
[22:31] <Gus> <Gus> MMX2 enabled no
[22:31] <Gus> <Gus> 3DNow! enabled no" etc.
[22:31] <slackyman> I don't think so
[22:31] <slackyman> please, try ONLY --enable-runtime-cpudetect and let's see the output
[22:31] <Gus> I don't want to go without runtime CPU detection since the build is to be used on different machines, including some older ones which do not have SSE.
[22:31] <Gus> Okay.
[22:32] <slackyman> I'm still finding a way to compile librtmp with polarssl in mingw32... -.-
[22:32] <slackyman> f***ing windoz
[22:33] <Gus> slackmagic: however, something must have changed inm the configure script. Because I use the same options set for more than two years.
[22:34] <Gus> I mean: slackyman ;-)
[22:34] <slackyman> this can be really true, yes
[22:34] <slackyman> my nick means slcakware user even if I'm on a Fedora and a Windowz machine :D
[22:35] <Gus> Only CPU runtime detection given, no other --enable-swicthes set. Result: CPU detection enabled, all CPU features disabled, still.
[22:35] <slackyman> I'm really so tired to recomplile every prog
[22:35] <slackyman> it's the latest snapshot?
[22:36] <slackyman> or you get it by git?
[22:36] <slackyman> i got mine by gittin
[22:36] <Gus> slackyman: snapshot.
[22:36] <slackyman> and I have none of this problem
[22:36] <Gus> From january.
[22:36] <slackyman> ok, try the git one
[22:37] <slackyman> just to make a proof
[22:37] <Gus> Haven't Git available atm.
[22:37] <slackyman> a little test
[22:37] <slackyman> which distro R U using?
[22:37] <Gus> slackyman: LFS.
[22:37] <slackyman> omg >.<
[22:38] <Gus> That's why I use configure and not apt-get, emerge or yast or something like that ;-)
[22:38] <slackyman> cannot simply get git?
[22:38] <Gus> Not that simply.
[22:38] <slackyman> and that's why I'm using Fedora and not Slackware
[22:38] <slackyman> :D
[22:38] <Gus> Besides, I doubt that a fortnight in versions could make this big difference,.
[22:39] <Gus> A-ha! I got it!
[22:39] <Gus> The culprit was another switch!
[22:39] <slackyman> I remember... yes, I used a january snapshot without inconvenience a pair of day ago
[22:39] <slackyman> yeah?
[22:39] <slackyman> which one?
[22:40] <Gus> --cpu=i586
[22:40] <slackyman> -.-
[22:40] <Gus> That one seems to exclude all other CPU features.
[22:40] <Gus> However, this is new behaviour.
[22:40] <slackyman> I'm tempted to let you go to HELL >.<
[22:40] <Gus> Formerly the --enable-feature switches would override the --cpu= restriction.
[22:40] <slackyman> no, really, I think it's ok
[22:41] <slackyman> "formerly", in my opinion, means nothing :D
[22:41] <Gus> As said, I use the same options for more than two years. Just this new build became significantly slower than older ones.
[22:41] <Gus> However, it's nearly a year since my previous build :-S
[22:41] <Gus> So I don't know when this behaviour was changed.
[22:42] <slackyman> I don't use CPU if I'm building a portable one
[22:42] <slackyman> all the job is done by the runtime detection
[22:42] <slackyman> it costs only a little bigger binary
[22:43] <Gus> slackyman: in case of ffmpeg: apparently true. In case of most other software: there isn't CPU runtime detection. That's why I use to specify it explicitly at build time.
[22:43] <slackyman> can I ask U a question?
[22:43] <Gus> slackyman: sure.
[22:43] <slackyman> a PAIR of question
[22:43] <slackyman> :D
[22:43] <slackyman> 1 - Have you ever succeded in compiling librtmp?
[22:44] <slackyman> (with polarssl)
[22:44] <Gus> 1. not recently but I think some years ago I probably did. But most likely not with polarssl, since I don't know what that is.
[22:45] <slackyman> I forgot the second question :P
[22:45] <Gus> Well, I didn't knwo what that is :-)
[22:45] <Gus> I use OpenSSL and GnuTLS instead.
[22:45] <Gus> I think I managed to compile something capable of RTMPS, if that's your goal.
[22:46] <slackyman> it seems that for windows I can use polarssl since openssl lacks some bugs in win
[22:46] <Gus> Windows? Uhm...
[22:46] <Gus> I don't use Windows at all.
[22:46] <slackyman> I'm into a portable project
[22:46] <Freakshow> anyone know how to explicitly force rtsp over tcp with ffmpeg?
[22:46] <slackyman> and I need to compile ffmpeg for Linux and Win
[22:47] <Gus> ffmpeg can be compiled under Windows?
[22:47] <Gus> Every day I learn something new about that OS :-)
[22:47] <slackyman> gus, I compile ffmpeg under Linux FOR windows
[22:47] <slackyman> with mingw32
[22:48] <slackyman> and then test it in bot
[22:48] <Gus> Mingw32 on Linux??
[22:48] <slackyman> I'm writing a wrapper in Lazarus - fpc
[22:48] <slackyman> yes
[22:48] <slackyman> since I think that WinFF it's not so simple to use I re-write my lazarus app from scratch
[22:49] <slackyman> for windoz users
[22:49] <slackyman> i wanna give the app almost all the ffmpeg features
[22:49] <slackyman> with a good GUI
[22:49] <Gus> slackyman: brave intention.
[22:49] <slackyman> and under gpl3
[22:49] <Gus> Very brave intention :-)
[22:50] <slackyman> I've just written a console wrapper
[22:50] <Gus> Freakshow: using ffserver?
[22:50] <slackyman> do you know SUPER?
[22:50] <slackyman> I think it violate gpl
[22:50] <slackyman> but I'm not sure
[22:51] <Freakshow> Gus: I'm sorry, as in the target for the stream?
[22:51] <meta_> Hi all folks! Can someone help me, at least to name what is my problem, as i can't really google for it... http://www.youtube.com/watch?v=OVEo3TRsUAM#t=5m30
[22:51] <meta_> It isn't moves too fast, but horizontally flickers
[22:51] <meta_> I've made this in blender, and it uses ffmpeg
[22:51] <Freakshow> if so, then the short answer is no
[22:52] <Gus> Freakshow: target for a stream sounds like a job for ffserver to me.
[22:52] <Freakshow> shooting to wowza
[22:52] <meta_> So what's the name of this syndrome? And maybe any idea about how can i fix it?
[22:52] <Gus> Freakshow: or do you mean reading from an rtsp source and encoding, possibly transcoding the data?
[22:52] <Freakshow> Gus: nope, shooting to wowza and fanning out from there
[22:53] <slackyman> meta_
[22:53] <slackyman> are you talking about the video on youtube?
[22:53] <meta_> yes
[22:53] <slackyman> I can see it good, without flickkering
[22:53] <meta_> it flickers the same on the original also, i know it's converted
[22:53] <meta_> oh...
[22:53] <Gus> Freakshow: sorry, I'm not familiar with Wowza Media Server.
[22:53] <slackyman> could it be your player?
[22:54] <slackyman> have you tried to see it on another machine?
[22:54] <meta_> i've tried vlc, mplayer, on my host os, on an older version ubuntu in vbox, and in xp in vbox
[22:54] <Freakshow> Gus: I doubt wowza really has anything to do with it in this instance, is there a way to tell ffmpeg to publish rtsp over tcp?
[22:54] <slackyman> meta_ try it on another hardware
[22:54] <meta_> hmm...
[22:54] <slackyman> i can see NO FLICKER
[22:55] <Gus> Freakshow: I don't think so. ffmpeg isn't the publication tool. It's the read and store tool. ffserver is the read and publish tool.
[22:55] <slackyman> the video goes well and fluid
[22:55] <meta_> It's not really flicker, i just can't name it
[22:55] <meta_> if you see the edges it isn't fluid... yeah... for me it seem, i'm going to check elsewhere
[22:55] <Gus> Freakshow: however, ffserver seems somehow limited regarding protocols and format. I don't know whether it supports rtsp.
[22:55] <meta_> thanks!
[22:56] <overflow_0f8b> how do you call it when a line has common part with a rectangle ? line_has_common_part_with_rect() sounds long
[22:56] <slackyman> i can see the video fluid and flawless
[22:56] <Gus> Btw, does anyone know whether there's been any progress in adapting ffserver (or the ffmpeg suite as a whole) to v4l2 lately?
[22:56] <Freakshow> hmm... gotta be a way
[22:56] <Freakshow> { "tcp", "TCP", 0, AV_OPT_TYPE_CONST, {1 << RTSP_LOWER_TRANSPORT_TCP}, 0, 0, DEC|ENC, "rtsp_transport" }, \
[22:56] <Freakshow> in const AVOption ff_rtsp_options
[22:56] <slackyman> anyone here ever compiled rtmp for windoz?!?
[22:57] <slackyman> I'm getting REALLY mad
[22:57] <Freakshow> calm down slackyman
[22:57] <Freakshow> breathe
[22:57] <slackyman> Freakshow: I'm not mad with this chan :D
[22:58] <slackyman> I'm mad with librtmp
[22:58] <slackyman> :D
[22:58] <Freakshow> oh, I'm in the same boat with you there
[22:58] <Freakshow> :D
[22:58] <Gus> I'd get mad just if I had to build anything ambitious for Windoze at all ;-)
[22:58] <slackyman> I compiled EVERY SINGLE lib for win to use them with ffmpeg
[22:58] <slackyman> with only a little patch
[22:58] <meta_> wow
[22:59] <slackyman> but librtmp is a damned hell
[22:59] <Freakshow> nice Gus
[22:59] <Gus> slackyman: what sorts of errors?
[22:59] <Freakshow> I had a hard enough time building it for bsd, I can't imagine doing it for windows
[22:59] <slackyman> every sort
[22:59] <slackyman> :)
[22:59] <Gus> slackyman: same as on Linux or Windows-specifc ones?
[22:59] <Freakshow> HA!
[22:59] <slackyman> rtmp cannot find polarssl
[23:00] <Gus> slackyman: so primarily link errors and not compile errors?
[23:00] <Ginks> would anyone be able to show me how I can merge these into 1 command?
[23:00] <Ginks> -vf "scale=-1:720" & -vf movie=/srv/www/vhosts/testing.com/htdocs/data/watermark.png [logo]; [in][logo] overlay=10:10 [out]
[23:00] <slackyman> gus, the fact is that I tested different versions and had different issues
[23:01] <Gus> slackyman: did you test any version that had none of the issues on Linux?
[23:01] <Freakshow> agreed... compiling librtmp is not very clear on any platform
[23:01] <Gus> So are you sure the issues are fully Windows-specific?
[23:02] <slackyman> on linux i can compile rtmp with openssl but have issues with polarssl
[23:02] <Gus> slackyman: what are the issues with OpenSSL on Windows?
[23:02] <slackyman> gus, I'm cross-compiling
[23:02] <slackyman> but here in Italy it's late and I'm tired
[23:02] <slackyman> :D
[23:03] <Mavrik> Ginks, movie=...,scale=...
[23:03] <Gus> Have you tried building it natively on a Windows machine?
[23:03] <Mavrik> Ginks, add comma ;)
[23:03] <slackyman> librtmp seems to NOT work with openssl in win
[23:03] <Gus> slackyman: link errors?
[23:03] <Gus> Or does it link but not run?
[23:03] <slackyman> no no
[23:03] <slackyman> doesn't link
[23:04] <Gus> At least link errors are often easier to solve than runtime errors I think :-S
[23:04] <slackyman> cannot find polarssl/net.h
[23:04] <Gus> That's a compiler error.
[23:04] <slackyman> even if i #include <polarssl/net.h>
[23:04] <Gus> Should be even easier to solve than a link error.
[23:04] <slackyman> or something I cannot remember
[23:04] <Gus> In this case I usually locate the missing header file manually, copy it to somewhere where it can be included and retry :-)
[23:05] <slackyman> but, really, it's too late and I'm on the phone with my girlfriend so... goobye :D
[23:05] <Ginks> Mavrik, oh
[23:05] <Ginks> lol
[23:05] <Gus> Once I got it working I then may check why the error occurred and how to solve it in a way that could be easier applied by others.
[23:05] <Gus> slackyman: bye.
[23:05] <slackyman> just tried
[23:05] <slackyman> i put polarssl build EVERYWHERE
[23:05] <slackyman> :D
[23:06] <slackyman> in /usr/local
[23:06] <slackyman> into the librtmp folder
[23:06] <slackyman> in rtmpdump folder
[23:06] <Gus> slackyman: just deal with the missing .h file.
[23:06] <Gus> Copy it into /usr/include for the time being if necessary.
[23:06] <slackyman> yeah, that was my goal
[23:06] <slackyman> :D
[23:07] <slackyman> I copied in /usr/local/include and /usr/include
[23:07] <slackyman> i tried even with
[23:07] <slackyman> #include "polarssl/net.h"
[23:07] <Gus> Well, if the compiler cannot find it there, then there seems to be a serious problem with your build environment.
[23:07] <slackyman> and
[23:07] <slackyman> #include <polarssl/net.h>
[23:07] <slackyman> but none
[23:08] <slackyman> gus, I doubt
[23:08] <slackyman> with my env I compiled everything :D
[23:08] <slackyman> openssl, divxlib, freetype
[23:08] <slackyman> sdl
[23:08] <slackyman> everything
[23:08] <slackyman> and they needed only feww changes
[23:08] <Gus> Then you can't tell me that if you have a file x.h in /usr/include and have #include <x.h> in your sources it doesn't find it?
[23:09] <slackyman> gus, it's not so simple with mingw32
[23:09] <slackyman> it create another toolchain env
[23:09] <Gus> Put it into your standard include directory of Mingw32.
[23:09] <Gus> In the same one that provides stdlib.h for instance.
[23:09] <slackyman> ok, I'll give it a try, but now I've to go
[23:09] <Gus> Bye.
[23:09] <slackyman> thanks anyway
[23:10] <slackyman> bye
[23:10] <Ginks> Mavrik,
[23:10] <Ginks> ffmpeg -i hot.flv -an -vf "scale=-1:720,movie=/srv/www/vhosts/testing.com/htdocs/data/watermark.png [logo]; [in][logo] overlay=10:10 [out]" -f mp4 scaled.mp4
[23:10] <Ginks> I tried it a few ways with commas and got different errors
[23:11] <Ginks> ffmpeg -i hot.flv -an -vf "movie=/srv/www/vhosts/testing.com/htdocs/data/watermark.png [logo]; [in][logo] overlay=10:10 [out]", scale=-1:720 -f mp4 scaled.mp4
[23:12] <Ginks> I need to have it scale and then overlay
[23:12] <Ginks> so that the watermark doesn't become distorted
[23:27] <Ginks> ffmpeg -i hot.flv -an -vf "movie=/srv/www/vhosts/testing.com/htdocs/data/watermark.png, scale=-1:720 [watermark]; [in][watermark] overlay=10:10 [out]" -qscale 1 -f mp4 scaled.mp4
[23:27] <Ginks> [overlay @ 0x192ecda0] Overlay area (10,10)<->(2224,730) not within the main area (0,0)<->(320,240) or zero-sized
[23:27] <Mavrik> obviously your overlay parameters are wrong.
[23:28] <Ginks> if I split the command into 2 and run the scale first and then watermark after it works
[23:29] <Ginks> trying to put them together is where apparently the overlay parameters are incorrect
[23:30] <Ginks> if I swap scale to before movie then I get Too many inputs specified for the "movie" filter.
[00:00] --- Tue Feb 7 2012
More information about the Ffmpeg-devel-irc
mailing list