[Ffmpeg-devel-irc] ffmpeg.log.20120403

[burek]

[00:38] <joyindescribable> hi, i'm trying to find out if ffmpeg supports hardware acceleration under debian squeeze?
[00:39] <joyindescribable> i've got a ton of transcoding to be done, and my CPU is absolutely not cutting it.
[00:40] <joyindescribable> all the googling i've done has just led to mailing lists where the topic is debated, but never addressed. even if it can't hardware accelerate the whole process, any degree of acceleration would help a lot.
[01:00] <felix_da_catz> So what is the smallest length of video that I could use to sample to figure out what settings work best when converting the video?  I realize this has as many answers as there are videos out there, but I just want to have a general idea
[01:12] <joyindescribable> can anyone tell me if ffmpeg supports using a GPU to encode/decode/transcode video in debian (squeeze)?
[01:21] <ubitux> no
[01:21] <ubitux> a smart way of using the gpu is yet to find afaik
[02:14] <joyindescribable> well, damn.
[06:16] <mjevans> I'm trying to deinterlace some content that mplayer sees as 24p when playing the 60i (30p in file) mpeg2 stream using ffmpeg.  I'm using -vf 'yadif=1,mp=mcdeint:0:0:16,mp=softskip,mp=harddup,hqdn3d=2' -r 24000/1001 but the 'dropped' frames doesn't quite reach the ratio I expect and it's not clear how to enforce the output rate I want to keep.
[06:38] <relaxed> mjevans: I would try trimming the mp:bullshit as much as you can.
[06:39] <relaxed> It's hasn't been tested very well and never should have been imported in the first place.
[06:42] <mjevans> relaxed: Actually I only used it for two reasons.  1) Primarily because at the moment mencoder seems to hate mcdeint on all my systems (either segfaults or says it doesn't exist).  2) It should be possible to do the same operation entirely in ffmpeg.  The actual output frames do look nice, but there's an occasional stutter in the output and the file acts a bit weird; leading me to think that while it might be /marked/ for playback at 24fps, it's contents are n
[06:42] <mjevans> That can't have fit... I forgot this didn't have a cut-script.
[06:42] <mjevans> leading me to think that while it might be /marked/ for playback at 24fps, it's contents are no longer synced to that.
[06:43] <relaxed> what does ffmpeg probe as the source framerate?
[06:44] <relaxed> fyi, mencoder is broken and abandoned
[06:47] <relaxed> -r 24000/1001 should have the same effect as mp=harddup
[06:48] <relaxed> mjevans: look at yadif's options in the man page and try with all the mp options omitted.
[06:52] <mjevans> yadif just doubles the framerate...
[06:52] <mjevans> Well in this case, it makes each field half it's own frame.
[06:55] <relaxed> yadif can output one frame for each frame or field
[06:56] <mjevans> Interestingly it also reads the source as 23.90 fps, 59.94 tbr, 90k tbn, 59.94 tbc; however even if it's redundant this time I'd still like to have higher quality de-interlacing than just throwing out fields.
[06:56] <mjevans> This example might not need it, but other discs in my collection do...
[07:12] <mjevans> Hum... Arch Linux seems to lack a build of ffmpeg that has avconv, or calls it something different...
[07:29] <mjevans> Glancing at the ffmpeg manual (avconv doesn't yet exist on the distributions I use), I can't seem to find anything that does what I want logically; decoding the input file without trying to natively de-interlace so that I can instead do that with a filter chain using a process I control.
[07:43] <relaxed> mjevans: avconv is from the libav fork
[07:44] <relaxed> it's what I use now
[07:45] <mjevans> Installing a lib package to get a binary?  I'll look in to it.
[07:45] <relaxed> Debian and Ubutntu have moved to it as well. I assume the rest will follow.
[07:46] <relaxed> No. libav forked from ffmpeg.
[07:46] <relaxed> In ffmpeg you have ./ffmpeg and with libav you have ./avconv
[07:47] <mjevans> What I mean is, I ask for the package aur/libav-git in order to get a /binary/ (a program).  This breaks years of (general) convention.
[07:47] <relaxed> ffmpeg funnles back most of the changes from libav so the two programs operate pretty much the same.
[07:48] <relaxed> how so?
[07:48] <mjevans> libs contain library files.  Binaries that use library files should be seperate.
[07:49] <relaxed> You mean the name?
[07:49] <relaxed> starting with "lib"
[07:49] <relaxed> ?
[07:50] <relaxed> In their case the libs are more inportant than the encoding program. Thus the name, I guess.
[07:52] <relaxed> They couldn't use ffmpeg but everyone knew the libs as libav*
[07:54] <mjevans> It's confusing; I guess for the binaries most dists will probably break it out like that and have them by their own name or maybe avtools...
[07:56] <relaxed> Indeed. The fork is unfortunate and confusing but that's what happens when people can't play nice together.
[07:58] <relaxed> It did stir up some competition between the two pojects, which can only be good.
[08:01] <mjevans> Yes, I hope it has some options so I can handle the input data at a slightly lower level without resorting to pipes and something else decoding the source data.
[08:15] <mrec> hi, is there any replacement for av_register_protocol2?
[08:20] <mrec> ah found it
[09:08] <TACPILOT> is there a way to manually change the individual values for such ....   60 fps, 59.94 tbr, 60 tbn, 120 tbc
[09:11] <TACPILOT> I can use the -r option but want to set tbc to a diff value
[10:40] <funyun> hi. how do i force a 16:9 AR with ffmpeg?
[10:40] <Tjoppen> -vf satdar=16:9
[11:01] <funyun> Tjoppen: thanks
[13:08] <mystica555> wow..how long ago did ffmpeg start supporting rendering/playback of ANSI and ASCII artwork (and animations) ?
[13:09] <mystica555> i was just looking for a linux ANSI viewer a couple days ago
[13:15] <mystica555> ok, 2010 is the source copyright
[13:15] <mystica555> im never failing to be impressed by ffmpeg
[13:15] <mystica555> srsly, keep up the amazing work :D
[13:16] <Tjoppen> you can just run git log on the decoder :)
[13:17] <Tjoppen> since july 18th 2010 it seems
[13:19] <Boon> what caused this
[13:19] <Boon> http://pastebin.com/dY8tgC8j
[13:20] <Boon> on gentoo os
[13:21] <ubitux> broken shell?
[13:21] <Boon> any idea?
[13:21] <Boon> i able to install other tool
[13:22] <Boon> like svn, ruby, theora xvid
[13:22] <ubitux> sentence better with verbs
[13:22] <ubitux> :)
[13:22] <ubitux> you didn't share your ./configure line either
[13:22] <ubitux> what's the shell?
[13:22] <Boon> how to check?
[13:23] <ville> Howdy. I am looking to convert iphone generated .mov which has QT 5.1 in it into a more Windows friendly format. I can simply do ffmpeg -i file.mov file.avi but that compresses the hell out of it. It goes to ~10% of the original filesize and pixelates. What sort of options should I pass to try to conserve image quality?
[13:23] <Boon> this is my configure parameter
[13:23] <Boon> ./configure --enable-shared --enable-gpl --enable-version3 --enable-nonfree --enable-pthreads --enable-x11grab --enable-libopencore-amrnb --enable-libopencore-amrwb  --enable-libfaac --enable-libmp3lame --enable-libopenjpeg --enable-libtheora --enable-libvorbis --enable-libx264 --enable-libxvid --enable-zlib
[13:23] <ville> Filesize can remain normal original
[13:23] <ubitux> echo $SHELL maybe
[13:23] <ubitux> you may try "bash ./configure ..."
[13:23] <Boon> echo $SHELL /bin/bash
[13:24] <ubitux> gnu make ?
[13:24] <ubitux> aren't you using a bsd or sth?
[13:24] <Boon> is gentoo
[13:25] <Boon> make --version GNU Make 3.80 Copyright (C) 2002  Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
[13:25] <ubitux> mmh.
[13:25] <ubitux> i don't have any foreach at line 28 in common.mak
[13:25] <ubitux> are you up-to-date?
[13:26] <Boon> svn checkout svn://svn.mplayerhq.hu/ffmpeg/trunk ffmpeg
[13:26] <Boon> i used that
[13:26] <ubitux> we don't use svn since a long while&
[13:26] <ubitux> http://ffmpeg.org/download.html
[13:27] <ubitux> git clone git://source.ffmpeg.org/ffmpeg.git ffmpeg
[13:27] <Boon> ok
[13:27] <Boon> let me try again
[13:28] <Boon> same
[13:28] <Boon> License: nonfree and unredistributable Creating config.mak and config.h... ks392277 ffmpeg # make && make install common.mak:17: *** appel à la fonction « foreach » non terminé : « ) » manquant. Arrêt. ks392277 ffmpeg #
[13:29] <Boon> i get this
[13:29] <Boon> common.mak:17: *** appel à la fonction « foreach » non terminé : « ) » manquant. Arrêt.
[13:29] <Boon> its a OVH server
[13:30] <Boon> any idea?
[13:30] <ubitux> no but sth looks broken in your install
[13:31] <ubitux> this stuff is run all the time and no one noticed this
[13:31] <Boon> ffmpeg is work fine with centos
[13:31] <Boon> i never tested on gentoo
[13:31] <ubitux> i'd say your gnu make is missing a use flag allowing the foreach or sth :)
[13:32] <Boon> update gmake?
[13:32] <ubitux> dunno i don't use gentoo
[13:33] <ubitux> maybe try to join the lines 17 and 18
[13:33] <ubitux> (removing the \)
[13:35] <mrec> ffurl_register_protocol should this be the replacement for av_register_protocol?
[13:36] <ville> when I try to do `ffmepg -i file.mov file.wmv` I get: "[wmav2 @ 0x1bb3ca0] output buffer size is too small" what can I do to fix this?
[13:39] <archman> hello
[13:39] <archman> any ideas why libpostproc.pc is missing from git or freedom?
[13:40] <archman> after making
[13:42] <iive> archman: is the library built ?
[13:44] <archman> iive: how to check on that? libpostproc.v  Makefile  postprocess_altivec_template.c  postprocess.c  postprocess.h  postprocess_internal.h  postprocess_template.c are in after make
[13:44] <iive> i think you need .so or .a
[13:46] <iive> that lib got removed from libav fork, and some of the commits removing it may have been merged by mistake.
[13:50] <Boon> ubitux: http://pastebin.com/ykyurkzb
[13:50] <Boon> error said: common.mak:17: *** appel à la fonction « foreach » non terminé : « ) » manquant. Arrêt.
[13:55] <ubitux> it looks terminated to me :p
[13:58] <Boon> i checked the ) is work fine
[13:58] <Boon> nothing is missing
[13:58] <Boon> but why that error occur
[13:59] <ubitux> check if there is not known bug with your gmake
[13:59] <ubitux> downgrade or sth
[15:07] <nzo> hi
[15:08] <nzo> is any one can tell me how to extract selected video stream?
[15:18] <burek> nzo, you need to extract video stream only from a file?
[15:18] <burek> can you type: ffmpeg -i file.avi
[15:18] <burek> and use pastebin.com to paste the result
[15:20] <nzo> http://pastebin.com/wFZZh8Us
[15:21] <nzo> it's 3D stereoscopic video from fuji real 3D
[15:27] <nzo> burek, I'd like to know in order to automate the separation of the two streams. And also the command to create the same kind of file from two videos. split and join video stream
[15:28] <burek> well this is simple
[15:28] <burek> just use -map
[15:28] <burek> and choose which stream you want
[15:31] <nzo> burek,  Like this? ffmpeg -map Stream #0.2 1.AVI
[15:33] <relaxed> just the stream umber
[15:33] <relaxed> number*
[15:33] <relaxed> add -c copy unless you want to recode the stream.
[15:58] <JacobS2> hello all, Is it possible to use ffmpeg to stream http on windows ?
[15:58] <JacobS2> as far as ffserver, I didnt find a windows build
[15:59] <JacobS2> I need http to be the output not input
[16:03] <burek> JacobS2, use vlc
[16:05] <qbi_> Hi there :)
[16:05] <qbi_> I made a minimal program using ffmpeg's libs under linux and I'm trying to make it use the hardware accel for h264 (gma500). I have no idea how vaapi works so I'm currenlty digging into a patched mplayer's code, but by any chance could someone point me to a simpler-to-understand piece of example code I could look at?
[16:05] <JacobS2> burek: tried, but it get bad performance and very high cpu usage
[16:06] <JacobS2> burek: but when using ffmpeg with udp I get good results
[16:10] <burek> qbi_, did you try in #libav
[16:10] <burek> JacobS2, ffserver is not ported on Windows and that's the only way to have http server with ffmpeg
[16:10] <burek> sorry
[16:11] <qbi_> I didn't, but that's an idea
[16:17] <JacobS2> burek: maybe you will have a different idea, with udp the switch is showing packet lost.
[16:18] <burek> well, you can always fix your network connection
[16:18] <burek> either by changing cables or providers, etc :)
[16:18] <burek> udp shouldn't generally have packet loss
[16:18] <burek> as shouldn't any other ip protocol per se
[16:19] <burek> but, the good thing with udp is if you are streaming live events
[16:19] <burek> your delay doesn't increase
[16:19] <burek> if some packets are lost, stream will hick up, but it will continue to be real-time
[16:19] <burek> while with tcp you will have hick-ups too + delay which is increasing
[16:19] <burek> as long as the stream is up
[16:20] <JacobS2> burek: you are correct, on most tests it works fine, but on some switches when set to full duplex I get many picture glitches
[16:20] <burek> setup qos on switch/router
[16:29] <JacobS2> burek: could changing to CBR help ?
[16:34] <Duccio> Hi all, I'm trying to re-wrap MTS from a NX70 Sony cam: ffmpeg -i 00000.MTS -vcodec copy -acodec copy test.mov. Why "Could not write header for output file #0 (incorrect codec parameters ?)" ? Thanks...
[16:40] <burek> Duccio, try remuxing to mp4 first
[16:40] <burek> JacobS2, you can try, I'm not sure
[16:40] <Duccio> burek: I have to change test.mov to test.mp4?
[16:40] <burek> JacobS2, but if you are streaming then you definitely want bitrate under control
[16:40] <burek> yes
[16:41] <burek> Duccio, yes, and if that doesn't work
[16:41] <burek> can you please use pastebin.com, to show your command line and its output?
[16:42] <Duccio> burek: http://pastebin.com/y72sinaB
[16:46] <burek> [mp4 @ 0x7fb8ab876a00] track 1: could not find tag, codec not currently supported in container
[16:46] <burek> try mkv
[16:47] <JacobS2> burek: thanks
[16:59] <Duccio> burek: the same
[17:01] <zimbatm> hey guys
[17:01] <zimbatm> i found a very curious thing
[17:02] <zimbatm> when encoding anything to a mpegts output, the start: is always 1.400000
[17:02] <zimbatm> i tried looking up the tickets, apparently it's not a new phenomenon but nothing i found was related
[17:03] <zimbatm> i think it's a bug but i'm not an expert enough to tell
[17:04] <zimbatm> can someone explain to me if it's an expected outcome ?
[17:05] <zimbatm> https://gist.github.com/59e13c20c1f6ae51a246
[17:32] <vivienschilis> i am kind of bored of this fucking segmenter
[17:32] <vivienschilis> oops wrong channel :)
[17:38] <maaaxxiiii> hello
[17:38] <maaaxxiiii> is anyone here?
[17:38] <maaaxxiiii> can somebody help me?
[17:41] <sacarasc> How can we if we don't know your problem?
[17:44] <juanmabc> philosophic
[17:45] <juanmabc> but how he's gonna ask if nobody can help him, or noone is here?
[17:45] <juanmabc> ;D
[18:08] <maaaxxiiii> lol
[18:08] <maaaxxiiii> yes
[18:08] <maaaxxiiii> i will tell my problem
[18:09] <maaaxxiiii> i have a .vob (mpeg version 2)
[18:10] <maaaxxiiii> when i try to encode it to a different container/codec, it encode with some bad quality in the borders
[18:10] <juanmabc> surprisingly ammount of response time, nice reflexes!
[18:10] <maaaxxiiii> i know if you dont see it, you dont understand
[18:10] <maaaxxiiii> thats why im uploading some videos showing you my proble
[18:11] <maaaxxiiii> as soon as i finish uploading i will give you the lnk
[18:17] <mrec> is anyone familiar with av_set_parameters?
[18:17] <mrec> I wonder what the replacement of it is
[18:23] <iive> maaaxxiiii: first, you should be using some profile or preset to set coding complexity, ffmpeg defaults for speed and this doesn't produce good quality results. Second, it is highly recommended to cut/crop all black borders and even some of the mangled border pixels.
[18:24] <maaaxxiiii> iive thanks for ansewring, i was using the slow preset for the mp4 (avc) codec
[18:24] <iive> you mean x264
[18:24] <maaaxxiiii> i am uploading the examplkes whit their respetcives lines that i used in ther encodeing
[18:24] <maaaxxiiii> iive yes
[18:24] <maaaxxiiii> libx264
[18:24] <maaaxxiiii> thats the codec i used
[18:27] <iive> probably the best h.264 encoder out there :) Did you use quality setting like crf or qp , or bitrate based one?
[18:54] <mjevans> maaaxxiiii: are you encoding for a mobile device of somekind; or are you instead encoding for a media center; or is it just 'so I don't need to pop in that stupid disc'?
[18:54] <mjevans> (if a specific device is your target, what make and model?)
[19:00] <maaaxxiiii> iive i used -b 10000
[19:01] <maaaxxiiii> mjevans there is no target device
[19:01] <sacarasc> That is 10kbits per second.
[19:01] <maaaxxiiii> iive i used -b 10000k
[19:01] <maaaxxiiii> that is lol
[19:02] <maaaxxiiii> i just want it to be mp4, because as it is ver comopressed it has same quality with less file size
[19:02] <maaaxxiiii> can i post a lnk here?
[19:02] <maaaxxiiii> here is the link with the exapmles
[19:02] <maaaxxiiii> http://www.mediafire.com/?tlruxw36t21871s
[19:02] <iive> i would recommend you to use quality based setting, value of 18 should be about the same as quant=2 in mpeg2 (aka max quality)
[19:03] <iive> smaller values give better quality for bigger file size/bitrate
[19:03] <mjevans> Link yes.  We probably won't use it.  This is what I'm trying for a targeted bitrate: avconv -i 11_movie.vob -an -vf hqdn3d=2 -c:v libx264 -b:v 2250k -pass 1 -passlogfile av2passhigh10c -threads auto -preset veryslow -profile:v high10 -f rawvideo -y /dev/null && avconv -i 11_movie.vob -an -vf hqdn3d=2 -c:v libx264 -b:v 2250k -pass 2 -passlogfile av2passhigh10c -threads auto -preset veryslow -profile:v high10 11_movie.x264.h2250.mkv
[19:03] <maaaxxiiii> the same result, i just tried, -crf 1
[19:04] <iive> i'll take a look after 1 hour
[19:04] <mjevans> what 'problem' is it in description?  I can't click links
[19:06] <maaaxxiiii> ffs
[19:06] <maaaxxiiii> here is again http://www.mediafire.com/?tlruxw36t21871s
[19:07] <maaaxxiiii> mj are you talking to me?
[19:07] <mjevans> Yes I am.
[19:07] <maaaxxiiii> dont understand the question
[19:07] <mjevans> No, I can't look at links right now -describe- the problem.
[19:07] <maaaxxiiii> ah
[19:07] <maaaxxiiii> the problem
[19:07] <maaaxxiiii> yes
[19:08] <maaaxxiiii> the problem is that in the avi, mp4 and mpeg files, that i encode myself, there is some distorsio in all borders, that can easily be seen in the white ribbon of the guitar.
[19:08] <maaaxxiiii> In the file vob, the there is no distorsion, this is a copy of the original file, but the wiered thing is that the mpeg file if you read the codec infroamtion you will see that
[19:08] <maaaxxiiii> is almost the same codec with same bitrate etc, however the mpeg has the distorsion and the vob does not.
[19:09] <mjevans> Ok, I half-truthed.  I can't look at video links but I CAN pastebin... If you paste your encode script (anonimize as necessary) I can take a look at that.
[19:09] <mjevans> If your encode part is 2 lines or less just dump it here.
[19:09] <maaaxxiiii> ffmpeg -i "D:\Temp\hola.vob" -vcodec mpeg4 -b 20000k -acodec libmp3lame -ab 256k -ar 48000 -vol 512 -ac 2 -y "d:\Temp\hola3.avi"
[19:09] <maaaxxiiii> ffmpeg -i "D:\Temp\hola.vob" -vcodec libx264 -b 10000k -preset:v slow -acodec aac -ab 128k -ar 44100 -q:a 100 -strict experimental -y "d:\Temp\hola3.mp4"
[19:09] <maaaxxiiii> ffmpeg -i "D:\Temp\hola.vob" -vcodec mpeg2video -b 9000k -maxrate 9000k -bufsize 1835k -s pal -r 25 -f dvd -acodec ac3 -ab 448k -ar 48000 -vol 256 -y "d:\Temp\hola3.mpeg"
[19:09] <maaaxxiiii> ffmpeg -i "D:\Temp\hola.vob" -vcodec copy -acodec copy -y "d:\Temp\hola3.vob"
[19:10] <maaaxxiiii> these are the tree files, the avi, mp4 and mpeg have the distorsion
[19:10] <maaaxxiiii> the vob that is a cpoy does not
[19:10] <mjevans> Well of course the vob wouldn't, it's a stream copy
[19:10] <mjevans> have you used ffplay on the vob to see what parts are information and which parts are un-needed gutter?
[19:11] <maaaxxiiii> wow, i have no idea what you are talking about, lol
[19:12] <maaaxxiiii> i use media player classic
[19:12] <maaaxxiiii> the reproducer that comes with matroska
[19:12] <maaaxxiiii> klite codec pack lol
[19:12] <mjevans> also:  ffmpeg -i "D:\Temp\hola.vob" -vcodec libx264 -b:v 10000k -preset:v slow -acodec aac -b:a 128k -ar 44100 -q:a 100  -strict experimental "d:\Temp\hola3.mp4"
[19:13] <mjevans> maaaxxiiii: if you go through the file are there parts on the side that always (or usually) contain non-desired infromation?  IE letterboxing bars, or the thing just doesn't fill the whole frame because idiots mastered it for TVs that overscan by 5%?
[19:14] <maaaxxiiii> no
[19:14] <maaaxxiiii> it has nothing, it is just a video
[19:14] <maaaxxiiii> no add text
[19:14] <maaaxxiiii> nor side black bars
[19:14] <mjevans> So the whole 720x480 or 720x576 box is filled with useful data?
[19:15] <maaaxxiiii> wait a minute
[19:15] <mjevans> Anyway, how does my re-version of your mp4 work?
[19:16] <maaaxxiiii> i think you didnt understood the problem
[19:16] <mjevans> Mostly correct; you did not describe what your 'distortion' is
[19:16] <maaaxxiiii> EXACTLY!
[19:17] <maaaxxiiii> the distorsion is not in the borders of the video
[19:18] <maaaxxiiii> is in the borders of the contrast of the video, for example, when things move little fast, avery thing, (mostly the parts with most contrast) its borders experimnts like a distorsion, like an intrelaced or something, i dont know how to describe it
[19:18] <mjevans> Ahh, ok I need some data from you
[19:19] <maaaxxiiii> yes
[19:19] <maaaxxiiii> tell me
[19:23] <mjevans> http://pastebin.com/mnCGYbcA  << This is a reference.  It contains a line: Video: mpeg2video (Main), yuv420p, 720x480 [PAR 32:27 DAR 16:9], 9400 kb/s, 23.90 fps, 59.94 tbr, 90k tbn, 59.94 tbc  <<< My current disc was detected as having 60000/1001 fields/sec native but 24000/1001 fields/sec after selection of the non-duplicated fields.
[19:23] <mjevans> ffmpeg (renamed avconv from libav) in this case was able to get the film's origional data without a de-interlacing plugin
[19:24] <mjevans> I need that line from you; just that line.
[19:26] <maaaxxiiii> lets see....
[19:27] <maaaxxiiii> this line you mean? Stream #0:0: Video: h264 (![0][0][0] / 0x0021), yuv420p, 720x576 [SAR 64:45 DAR 16:9], q=-1--1, 10000 kb/s, 25 tbn, 25 tbc
[19:27] <mjevans> No, that's your output line
[19:27] <mjevans> I need the input line
[19:28] <maaaxxiiii> here mpeg2video (Main), yuv420p, 720x576 [SAR 64:45 DAR 16:9], 9800 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
[19:28] <burek> maaaxxiiii,
[19:28] <burek> can you please use pastebin.com, to show your command line and its output?
[19:29] <maaaxxiiii> yes
[19:29] <maaaxxiiii> i will paset all the output
[19:30] <mjevans> That seems odd, his tbr == fps, not tbr == tbc
[19:30] <maaaxxiiii> http://pastebin.com/S32ZavDW
[19:30] <maaaxxiiii> there it is
[19:31] <burek> maaaxxiiii, that paste is missing the command line
[19:34] <mjevans> That... has an aweful lot of x264 parameters; did you manually specify all of those? (of course seeing the command line with the output as asked for would tell us)
[19:37] <maaaxxiiii> mj wait a min, im on it
[19:42] <maaaxxiiii> mjevans here it is: http://pastebin.com/m9QD427W
[19:49] <mjevans> that's also for burek
[19:49] <iive> maaaxxiiii: just took a look, it is indeed mice tails, combing, caused by interlace or telecine.
[19:49] <maaaxxiiii> can you see pictures, i have just uploaded a picture showing the problem
[19:49] <maaaxxiiii> https://public.blu.livefilestore.com/y1ptHFApiTRWbTpWQUXlDW8EjR3x-vHcjYeGSU_sgmd92ThoZsnWBa6SZ039hZQztvAc8o5tr7NHVn90CLvsAb5Mg/Distorsion.png?psid=1
[19:50] <maaaxxiiii> there is the link
[19:50] <iive> maaaxxiiii: it is present in the .vob file too
[19:50] <maaaxxiiii> iive what does that means? is it possible to fix?
[19:50] <iive> but my guess is that the player you use is too smart, detects it and deinterlaces it
[19:51] <iive> try -vf yadif
[19:51] <maaaxxiiii> iive im using media player clasic, the player that comes with klite codec pack
[19:51] <maaaxxiiii> im trying...
[19:52] <iive> i don't think that ffmpeg have any useful inverse telecine filter, so ...
[19:52] <mjevans> Actually: iive and maaaxxiiii :: ffmpeg -i test.vob -vf yadif=1,mp=mcdeint:0:0:16,mp=softskip,hqdn3d=2 -vcodec libx264 -crf 16 -threads 0 -x264opts frameref=15:fast_pskip=0 -r 25 -y test.mp4
[19:53] <mjevans> you will note that you -must- specify the output framerate, as well as that I didn't put any audio codec info in there.
[19:53] <iive> mcdeint, is that the one that uses the snow codec?
[19:54] <mjevans> (You should add your aac stuff)
[19:54] <mjevans> iive: not so much in mode 0
[19:54] <mjevans> in mode 2/3 yes it does
[19:54] <maaaxxiiii> mjevans it gives me error
[19:54] <iive> niiiiice
[19:54] <iive> maaaxxiiii: are you using recent ffmpeg?
[19:54] <mjevans> pastebin the error?
[19:55] <iive> oh... i'm blind, it uses the mplayer filters...
[19:55] <mjevans> you probably need to add mp= all the way up to maybe -vf mp=yadif=1,mp...,mp=hqdn3d=2
[19:55] <maaaxxiiii> here http://pastebin.com/VjpwSm82
[19:56] <iive> yadiff is ported to ffmpeg, so it is removed from the mp= pool.
[19:56] <mjevans> maaaxxiiii: the video stream is fine; you forgot to add your audio junk from your MP4 example.
[19:56] <mjevans> Do so and try again
[19:58] <burek> maaaxxiiii, that looks like you need yadif deinterlace, like iive has said, and also, try to target quality with -crf rather than bitrate with -b:v
[19:59] <maaaxxiiii> very slow, but very good result :)
[19:59] <mjevans> maaaxxiiii: -very- slow is mcdeint:3:0:16
[19:59] <mjevans> result is not much better
[19:59] <maaaxxiiii> at least i dont have the "distorsion"
[20:00] <mjevans> IIRC that's called 'combing'
[20:00] <mjevans> Because the interlaced lines make edges look like combs
[20:00] <iive> maaaxxiiii: you can make quick test run without the mcdeint, it is a lot more slower.
[20:01] <sprzybilla> slightly off topic, anyone know of software to open/edit captions?
[20:01] <sprzybilla> I have tried a few different things, but they are all pretty pricey. It though that there might be an open source solution somewhere
[20:02] <maaaxxiiii> ys it is much faster
[20:03] <maaaxxiiii> do you know what parameters use youtube to encode their videos?
[20:03] <mjevans> If you don't want to mcdeint you should probably double the output framerate and remove -all- the filters after mcdeint too
[20:03] <maaaxxiiii> i mean the -vf or preset
[20:05] <burek> sprzybilla, captions=subtitles?
[20:06] <sprzybilla> burek: unforutantely I mean closed captions, specifically .scc files
[20:07] Action: sprzybilla shakes fist at the "americans with disabilities act"
[20:08] <burek> I really don't know what closed captions are :) sorry :)
[20:09] <sprzybilla> np =]
[20:14] <maaaxxiiii> mjevans i saw you told me top use crf 16, is there anyway to caclulate this parameter? 16? to get ceratin quality, or you just said that number because you thought it was going to be ok?
[20:15] <iive> i would recommend you to use quality based setting, value of 18 should be about the same as quant=2 in mpeg2 (aka max quality)
[20:18] <maaaxxiiii> ok but for example, i have this video, with 4965k of bitrate (mpeg version2) and i want to keep same quality but lower file size, so i encode it to mp4 (avc), is there any way to calculate its -crf so i get get the same quality of the mpeg file?
[20:19] <burek> maaaxxiiii, try /j #videolan or #x264
[20:20] <maaaxxiiii> burek ar thos other encoding programs? i use ffmpeg lol
[20:20] <maaaxxiiii> i will see waht can i found at those channles
[20:20] <burek> no, join those channels if you want to ask libx264 specific questions
[20:21] <burek> you'll get more precise answers
[20:21] <mjevans> I find I prefer 2-pass over single pass (because you can shift bits from lower motion scenes to higher motion ones); however if doing 1-pass crf is the current best answer
[20:36] <cbreak> mjevans: crf shifts bits from lower to higher bitrate scenes
[20:37] <mjevans> Then what's the C stand for? Isn't it constant?
[20:37] <JEEB> also, just to note on what mjevans just said about 2pass: <Dark_Shikari> mjevans: what you're saying is incorrect, please don't spread false information. <Dark_Shikari> 2-pass uses the same bit distribution algorithm as CRF. <Dark_Shikari> They are identical.
[20:37] <JEEB> mjevans, rate factor
[20:38] <burek> constant rate factor
[20:38] <mjevans> I see /constant/ and I think the same thing that CBR audio codecs do.  No variation in stream rate ever.
[20:38] <JEEB> nope.avi
[20:38] <JEEB> anyways, the results of a 2pass end up with a certain rate factor :P
[20:38] <cbreak> mjevans: get a clue
[20:38] <mjevans> JEEB: it's time to upgrade, use mkv
[20:39] <burek> crf is NOT constant bitrate
[20:39] <cbreak> crf is more like "constant quality"
[20:39] <burek> exactly
[20:39] <burek> in my streams the bitrate is usually 320kbps but sometimes it goes over 1mbps
[20:39] <burek> so it definitely is not constant
[20:39] <cbreak> if you do a few passes, then you can get close to the quality of a crf encode
[20:40] <cbreak> but with multi pass you get the bitrate you want on average
[20:40] <cbreak> with crf you have no control over the bitrate
[20:40] <cbreak> that's the main reason to use one or the other
[20:40] <mjevans> cbreak: You misunderstant what I was saying though;  JEEB is implying that the quality -per frame- varries with CRF;  While I understood CRF to mean that the QUALITY never varried (but the size did).  I understand 2-pass to varrying the QUALITY PER FRAME to achieve the same average quality over the file in order to match a given bitrate over the whole file.
[20:41] <cbreak> no
[20:41] <JEEB> I haven't said a thing like that yet :P
[20:41] <JEEB> also, cbreak -- you are partially incorrect there. --vbv-maxrate and bufsize work fine with crf
[20:41] <mjevans> er, sorry that was cbreak; 20120403-11:36:09 < cbreak> mjevans: crf shifts bits from lower to higher bitrate scenes
[20:41] <cbreak> JEEB: I know.
[20:41] <JEEB> so you have some control over the bit rate
[20:41] <JEEB> but yes
[20:41] <JEEB> general idea is correct
[20:42] <cbreak> JEEB: but that will still give you no real control over the bitrate, unless you butcher the crf algorithm with limitations so that you get constant bitrate...
[20:43] <JEEB> well, it will give me enough control to keep the stream within certain vbv limitations :)
[20:43] <cbreak> mjevans: crf spends as many bits as it needs to get its "quality". multi-pass tries to estimate the rate factor it needs to stay within the bandwidth goal and get a constant quality.
[23:06] <Unhelpful> you get totally screwed for many things if you don't use repo start :/
[23:06] <Unhelpful> sorry wrong channel
[00:00] --- Wed Apr  4 2012