[Ffmpeg-devel-irc] ffmpeg.log.20120629

Sat Jun 30 02:05:02 CEST 2012

[08:01] <rik316> Am I correct in saying that avio_rl32 will read the next 4 bytes from the stream and return them as an unsigned int?
[11:23] <mikunos> Hi guys I am trying to convert a mov file to an avi file but I get this error: [buffer @ 0x88b0180] Invalid pixel format string '-1'
[11:23] <mikunos> Error opening filters!
[11:23] <mikunos> any idea?
[11:23] <mikunos> this is the command executed: ffmpeg -i Scrivania/matrimonio1.mov video1.avi
[11:29] <saste> mikunos: try to update your ffmpeg, that was a problem which was fixed some time ago
[11:41] <mikunos> ok saste
[11:59] <natrixnatrix89> hi guys.. could you please recommend a lossless video codec for x11 grabbing ?
[11:59] <mikunos> saste: http://pastie.org/4170577
[11:59] <mikunos> nothing to do
[11:59] <mikunos> I have updated the ffmpeg to the latest version
[12:00] <mikunos> compiling all the dependencies
[12:00] <mikunos> but I get this result: http://pastie.org/4170577
[12:00] <JEEB> natrixnatrix89, ffvhuff or libx264. ffvhuff should be quite fast but not really compressible, libx264 depending on your hardware can be quite fast and can offer more compression
[12:01] <saste> mikunos: don't edit the output, the other info are useful as well
[12:01] <mikunos> I have followed this tutorial: https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide
[12:01] <natrixnatrix89> how do I set libx264 to be lossless? because if I set -b or -sameq it will compress.. how do I disable compression then?
[12:01] <JEEB> natrixnatrix89, sameq is not what you think it is
[12:01] <JEEB> and the documentation has since been altered to make it more apparent
[12:02] <JEEB> you use either -crf zero or set however you set quants to zero
[12:02] <natrixnatrix89> ok. but the question is.. how do I use libx264 for lossless recording?
[12:02] <mikunos> ok saste
[12:02] <JEEB> (I don't know how you set quants to zero in ffmpeg, but crf zero should in most cases work)
[12:02] <JEEB> also you most probably want to set a faster preset
[12:02] <mikunos> saste http://pastie.org/4170583
[12:02] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset
[12:03] <natrixnatrix89> yeah. fast preset should help
[12:04] <JEEB> but yeah, ffvhuff is then something that should really be fast and straightforward'ish
[12:04] <JEEB> if libx264 even with the fastest presets is too slow
[12:04] <iTux> Hi
[12:05] <natrixnatrix89> also is ffv1 good?
[12:05] <saste> mikunos: weird, can't find the pixel format in the stream, but it goes on anyway and fails when configuring filters
[12:05] <JEEB> ffv1 is a good codec, but I'm not sure if it's fast enough
[12:05] <saste> you can try to force the pixel format with -pix_fmt ... -i INPUT
[12:05] <iTux> I need some help, I can't configure ffmpeg with libass
[12:06] <saste> and you may open a ticket on trac
[12:06] <natrixnatrix89> but what is -crf ? can't find in documentation
[12:06] <mikunos> how have I to use th e -pix_fmt
[12:06] <JEEB> natrixnatrix89, it's a setting defined by the libx264 encoder
[12:07] <JEEB> it's libx264-specific
[12:07] <iTux> ./configure says "ERROR: libass not found"
[12:07] <JEEB> basically it's "constant rate factor", and tries to be something a la 'constant quality' in a way
[12:07] <JEEB> zero with a 8bit libx264 results in lossless
[12:07] <natrixnatrix89> oh.. I see. thanks.. ok. so can use ffmpeg -I {my screen} -crf 0 -pre fast out.mov right?
[12:08] <saste> JEEB: what about sending a doc patch for -crf?
[12:08] <JEEB> -vcodec/-c:v libx264 -crf 0 -preset veryfast or something
[12:08] <natrixnatrix89> right..
[12:08] <natrixnatrix89> thanks
[12:08] <JEEB> saste, I think it's already there tho, in video codec libx264's help?
[12:09] <JEEB> also I really want to get a first version of this video encoder done this weekend
[12:09] <saste> JEEB: yes, so it is a libx264 option, allright
[12:10] <JEEB> http://git.libav.org/?p=libav.git;a=blob;f=libavcodec/libx264.c;hb=HEAD#l469
[12:10] <JEEB> argh, wrong repo
[12:10] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/libx264.c;h=d56dfe76fd6a42766604bc7444f046f74a481fa1;hb=HEAD#l548
[12:10] <JEEB> there
[12:12] <saste> JEEB: yes you're right, no point into documenting external library options
[12:13] <Vardis> Hello
[12:17] <iTux> ./configure make a log, I paste it here : http://pastebin.com/p3eCiHJq
[12:17] <Vardis> iff I splitting from biger MTS to smaller chunks and encoding to webm how is the smartest way with audio? extract from big all lenght audio then compress to ogg or extract small portion of audio and then compress to ogg and then encode with thet small ogg?
[12:24] <Vardis> And also is there better way to remove interlacing from video or only yadif filter
[12:24] <microchip_> yadif is your "best" option with ffmpeg
[12:25] <Vardis> Oh
[12:27] <Vardis> but somtimes where is sharp contrast like white hand on dark gitar its aniway get jerky
[12:27] <microchip_> iirc, ffmpeg has only 2 deinterlacers; yadif and -deinterlace
[12:28] <Vardis> deinterlace is absolete
[12:28] <Vardis> as I recall correctly
[12:28] <microchip_> it's possible
[12:28] <Vardis> or meybi I'm wrong
[12:28] <JEEB> yes, it's obsolete and shouldn't be used
[12:28] <JEEB> it's very bad
[12:28] <JEEB> (as a deinterlacer)
[12:29] Action: microchip_ lets JEEB take over :p
[12:29] <JEEB> oi, I'm actually working now :P
[12:29] <Vardis> does vpre is always so slow?
[12:31] <microchip_> try a faster preset :p
[12:33] <Vardis> microchip_: I have turion 64x2 1.8GHZ using 720p preset for best quality as possible and got max out 2.9 and min 1.2 frames per/s
[12:33] <Vardis> faster presset will degrade more qulity :(
[12:34] <pron> taa!
[12:34] <microchip_> well, get a faster CPU. Turion 64 is pretty slow and old.
[12:34] <pron> pidarasi spameri zajebalji
[12:34] <Vardis> microchip_: whoos is admin here?
[12:34] Action: pron noslauka asinjainaas rokas un veero kaa pamazam tiiraas 20k mail queue 
[12:34] <microchip_> i don't know
[12:34] <microchip_> pron: GTFO
[12:34] <pron> omm
[12:35] <pron> wrong chan
[12:35] <pron> xD
[12:35] <pron> soz
[12:35] <pron> :}
[12:35] <Vardis> pron: nu skaties kur raksties
[12:35] <pron> haha
[12:36] <pron> microchip_: soz saw similar few names and missed chan :}
[12:36] <microchip_> no problem :)
[12:36] <Vardis> pron: bet 20k pastelja ir shausmas gan piekriitu
[12:36] <pron> Vardis: yep
[12:36] <Vardis> man apolleklii ~15k
[12:36] <Vardis> un kriit bezgala iekshaa
[12:36] <pron> Vardis: but this one is eng only chan
[12:37] <pron> <- tvnet :P
[12:38] <microchip_> what language is that?
[12:38] <pron> latvian
[12:38] <microchip_> ah thanks
[12:38] <pron> no clue wtf is that?:D
[12:40] <jonsen_> hm
[12:40] <Vardis> pron: ok I got Ya :D
[12:41] <microchip_> Vardis: as for the encoding speed, you have 3 options: 1) buy faster hardware, 2) use faster preset or 3) scale down the content
[12:41] <jonsen_> when I want to stream a .mov via ffserver is there anything needed prior to streaming to make it streamable eg. get header and stuff early?
[12:41] <jonsen_> I do have format .asf applied to ffserver, but when connecting to the server the client downloads the whole file and waits until its finished
[12:42] <Vardis> microchip_: whell I have that I got and nobody will sponsor Me for better HW but wideo want at least hd
[12:43] <Vardis> And I only encoding for youtube
[12:44] <microchip_> Vardis: youtube *always* re-encodes the video. So even if you encode it with "perfect" quality, youtube will encode it once more and degrade its quality
[12:44] <Vardis> even webm?
[12:44] <microchip_> i think so
[12:45] <microchip_> JEEB: does youtube *always* re-encode?
[12:45] <JEEB> yes
[12:45] <microchip_> Vardis: there ^^
[12:46] <microchip_> Vardis: so your "perfect" quality video will look "not-so-perfect" after youtube encodes it once more
[12:49] <microchip_> he timed out lol
[13:08] <natrixnatrix89> What is the file format if I want to save video with ffvhuff codec?
[13:09] <JEEB> I would guess you can mux it into avi or matroska (mkv)
[13:10] <natrixnatrix89> thanks
[13:13] <natrixnatrix89> pron: btw which was the channel in latvian on freenode?
[13:19] <pron> natrixnatrix89: i have no clue :}
[13:50] <jonsen_> its kinda frustrating that there is no documentation about how to stream simple things
[14:11] <saste> jonsen: http://ffmpeg.org/trac/ffmpeg/wiki/StreamingGuide
[14:11] <jonsen_> helps nothing as it only covers one very specific case
[14:12] <jonsen_> and finding an answer for "stream .asf from codex .foo could be done that way, consider this and that, problems are..."
[14:12] <jonsen_> i'am doing hard to get simple stream working
[14:47] <natrixnatrix89> ffserver stream?
[14:48] <jonsen_> cool segmentation fault when accessing sdp
[14:51] <jonsen_> natrixnatrix89: ffserver, yes
[15:07] <jzmer> does ffmpeg support any form of hardware-acceled audio decoding?
[15:48] <jonsen_> ffserver seems not to be very ready
[15:48] <jonsen_> it crashs a lot
[20:43] <MarioMey> Hello. I just want to clarify about ffmpeg... because, I used to use -vcodec... then, I had to change to -c:v... now, it says that ffmpeg is depracated...?
[20:43] <MarioMey> Is there any page where I can read about this?
[20:44] <MarioMey> Once I understand how to use it... now, I have to learn again.
[20:44] <MarioMey> I updated to Ubuntu 12.04 (I was on 11.04 with a ffmpeg from who-knows-who's PPA).
[20:45] <sacarasc> You don't have ffmpeg installed, but libav.
[20:46] <JEEB> yeah, and it doesn't say that ffmpeg as a project is deprecated, but that the ffmpeg tool (in libav) is
[20:47] <JEEB> also, -c:v is newer syntax on both libav and ffmpeg
[20:47] <JEEB> s/on/in/
[20:47] <JEEB> you can still use vcodec tho, methinks
[20:48] <MarioMey> What do you suggest...?
[20:48] <MarioMey> Install ffmpeg?
[20:48] <JEEB> uhh, the command line syntax in both should be similar, if you use libav you use avconv and if you use ffmpeg you use the ffmpeg tool
[20:50] <MarioMey> Well... I want to use what isn't depracated...
[20:50] <MarioMey> The one that won't disappear.
[20:50] <JEEB> if you are using libav, avconv is the up-to-date one; if you are using ffmpeg, it has integrated the libav avconv updates into ffmpeg
[20:50] <JEEB> you are now using libav
[20:51] <MarioMey> How to install it?
[20:51] <JEEB> huh?
[20:51] <MarioMey> avconv is the name?
[20:51] <JEEB> yes
[20:51] <MarioMey> sudo apt-get install avconv?
[20:51] <JEEB> you should already have it tho
[20:51] <MarioMey> ah...
[20:51] <JEEB> unless they put it into a separate package
[20:51] <MarioMey> Yes, I have it.
[20:52] <MarioMey> And -c:v or -vcodec?
[20:52] <JEEB> whichever you prefer, -c:v is the newer way of setting a codec, -vcodec is the older
[20:52] <JEEB> I think both still work
[20:52] <MarioMey> Right.
[20:52] <MarioMey> Another question...
[20:53] <MarioMey> I want to concatenate 2 videos with audio.
[20:53] <MarioMey> Can I do "avconv -i one.mod -i two.mod out.mp4?
[20:54] <JEEB> I think the only concateration thingy in both ffmpeg and libav was the filter that basically works as if you were just inputting both as a single file
[20:54] <JEEB> although I'm not sure
[20:54] <MarioMey> Sorry, but I didn't understand.
[20:54] <relaxed> -i concat:1.blah\|2.blah -c copy out.blah
[20:54] <JEEB> relaxed, but wouldn't that just do a simple concateration of input files?
[20:55] <JEEB> as in, not properly decode and put one input after another?
[20:55] <JEEB> so it wouldn't work with stuff that can't be "just concaterated"
[20:55] <relaxed> oh, sorry, I just glanced at his question.
[20:56] <relaxed> I assume you can do it with -map but I don't have time to hold someone's hand.
[20:56] <JEEB> I'm not that sure about that and I don't have enough info on that unfortunately
[20:57] <MarioMey> Thanks guys, I have a work meeting RIGHT NOW.
[21:27] <raptor67782> hi how to play with ffplay without video (only sound)?
[21:28] <raptor67782> ffplay -vo null   "mms://msnbc.wmod.llnwd.net/a275/e1/video/100/vh.asf"  is   not working
[21:29] <sacarasc> -vn
[21:30] <raptor67782> wow cool thanks
[21:30] <raptor67782> http://linuxers.org/tutorial/how-remove-audio-and-video-streams-media-file-using-ffmpeg
[21:31] <raptor67782>    ffplay  -vn   "mms://msnbc.wmod.llnwd.net/a275/e1/video/100/vh.asf"
[21:31] <raptor67782> well, not working
[21:31] <raptor67782>     ffplay  -vn   "mms://msnbc.wmod.llnwd.net/a275/e1/video/100/vh.asf"  teells that the stream is not found
[21:31] <raptor67782> however the stream is well there streamed to inteernet at this url
[21:33] <raptor67782>  ffplay -vn -ab 128     "mms://msnbc.wmod.llnwd.net/a275/e1/video/100/vh.asf"  is not working either
[22:46] <burek> raptor67782, did you ever read ffmpeg documentation?
[22:47] <burek> -vn is an output option
[22:47] <burek> so put it where it should be
[22:47] <burek> also -ab 128 (bits?) is an encoding option, not playback option
[22:47] <burek> and can't be used with ffplay
[22:48] <burek> what exactly are you trying to do ?
[22:58] <MarioMey> relaxed: concat:Sándalo 1º parte.MOD|Sándalo 2º parte.MOD: No such file or directory
[22:58] <MarioMey> It didn't work.
[23:00] <JEEB> see my comment towards relaxed, the concat filter/input thingy most probably isn't what you need. Also, if your file names have spaces and other things you have to quote/escape one way or another you'll have to deal with that.
[23:01] <MarioMey> they have quotes/escape, I used TAB.
[23:01] <MarioMey> What should I use?
[23:04] <intracube> hi
[23:04] <intracube> is it possible to get ffmpeg to compress expand the 0-255/16-235 range of a video?
[23:05] <intracube> I've noticed some camcorders don't peak out at Y=235, but instead go all the way to Y=255
[23:05] <intracube> is there a 0-255->16-235 filter/command?
[23:05] <intracube> the -color_range option got my hopes up:
[23:06] <intracube> http://ffmpeg.org/pipermail/ffmpeg-user/2011-July/001579.html
[23:06] <intracube> but it doesn't seem to work
[23:11] <MarioMey> How to enconde to libx264? "Unknown encoder 'libx264'"
[23:11] <juanmabc> "ffmpeg -codecs | grep 264"  search for "E" in the tags
[23:13] <MarioMey> juanmabc: no tengo codificadores... sólo "D"
[23:13] <MarioMey> Yo se que no está por default, ¿cómo lo instalo?
[23:13] <llogan> MarioMey: use a pastebin site to show your ffmpeg command and the complete console output
[23:15] <MarioMey> Only these:
[23:15] <MarioMey>  D V D  h264            H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
[23:15] <MarioMey>  D V D  h264_vdpau      H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (VDPAU acceleration)
[23:15] <JEEB> then that libavcodec library you are using was built without libx264
[23:15] <JEEB> are you still using a packaged version?
[23:15] <llogan> i meant your encoding command. i want to see the detailed information about your ffmpeg so i don't have to ask as many questions
[23:16] <JEEB> MarioMey, if you are still using a packaged version, try installing the packages x264 and libx264
[23:16] <MarioMey> llogan:  http://pastebin.com/nvX7uSRw
[23:17] <MarioMey> I'm installing x264
[23:17] <llogan> good. install libavcodec-extra-53
[23:17] <llogan> you don't need x264
[23:17] <JEEB> oh, they have a separate libavcodec package?
[23:17] <MarioMey> Should I uninstall it...?
[23:17] <JEEB> didn't know
[23:17] <JEEB> MarioMey, it won't do anything bad
[23:17] <llogan> no, you don't need to uninstall
[23:17] <MarioMey> Ok.
[23:17] <JEEB> just install the libavcodec-extra package as llogan said
[23:18] <smjms> so if I want to encode 2-pass H264, do I need to only specify -pass 1 and -b?
[23:18] <llogan> smjms: basically, yes.
[23:18] <JEEB> smjms, set -pass 1 for first pass with -b and then just switch s/1/2/ for the second pass
[23:18] <MarioMey> llogan: libavcodec53 will be uninstalled... ok?
[23:18] <JEEB> yes
[23:18] <MarioMey> libavutil51 too.
[23:19] <JEEB> it will install newer libraries that have libx264 support
[23:19] <MarioMey> Yes!
[23:19] <llogan> MarioMey: also, you are not using FFmpeg. you are using a fork that is not supported here (but we often help anyway).
[23:20] <JEEB> I already told him that some time ago :)
[23:20] <llogan> because i'm only an asshole 30% of the time
[23:20] <JEEB> and since his problems most probably have a common solution command-line wise I didn't tell him to go to #libav
[23:20] <smjms> okay, so does it matter what output codec I choose on the first pass, or does ffmpeg use some generic stats format libx264 can read?
[23:20] <JEEB> smjms, uhh I wouldn't change the vcodec around
[23:21] <smjms> okay, so libx264 on both passes
[23:21] <llogan> JEEB: ah. i came in late, and i think i've only referred one or two people to #libav
[23:21] <JEEB> stats files and mbtree stats with libx264 especially are libx264-specific
[23:21] <MarioMey> Encoding!
[23:21] <MarioMey> Thanks, gentlemen.
[23:21] <JEEB> smjms, anyways libx264 automatically selects faster settings in case it finds out that you are doing a "first pass"
[23:21] <MarioMey> ffmpeg doesn't have a x264 encoding?
[23:21] <JEEB> (pass 1 is set)
[23:21] <JEEB> MarioMey, the format is H.264 and x264 is the best H.264 encoder around
[23:21] <JEEB> there has been no need to reinvent the wheel
[23:21] <smjms> so I should define -pass after I set the -preset?
[23:22] <JEEB> smjms, it doesn't matter if it's before or after the -preset, it just has to be after -i (thus setting an encoder setting)
[23:22] <MarioMey> Ok, I just want to understand why you said this is not a ffmpeg topic.
[23:22] <smjms> ok
[23:22] <JEEB> MarioMey, you are using libav, which is a fork of ffmpeg
[23:22] <JEEB> (as I said a long time ago)
[23:23] <MarioMey> Ok.
[23:23] <JEEB> and well, in this case the answers to your questions would be the same with both avconv and ffmpeg
[23:23] <JEEB> thus I didn't tell you to go to #libav :)
[23:23] <smjms> is there a setting for "target filesize"?
[23:23] <MarioMey> Ok.
[23:23] <MarioMey> Thanks.
[23:23] <llogan> do users get much help in that channel?
[23:23] <juanmabc> i saw you even take back some libav patches
[23:23] <juanmabc> cool
[23:24] <JEEB> llogan, yes -- it's just that not many ask things there to begin with
[23:24] <JEEB> smjms, bitrate is pretty much target file size
[23:24] <llogan> i guess mru doesn't hang out there then
[23:24] <JEEB> nah
[23:24] <JEEB> he isn't even in libav-devel
[23:24] <JEEB> because his ways weren't exactly liked
[23:24] <smjms> JEEB: okay, thanks for help
[23:24] <llogan> unsurprising
[23:25] <smjms> !beer JEEB
[23:25] <smjms> I guess that was for another channel
[23:25] <JEEB> smjms, bitrate is basically kilobytes or kilobits per second, so you multiply by the length in seconds and you'll get the file size for that bit rate and length
[23:26] <JEEB> also, what I said about the fast settings in libx264 can be better read @ http://mewiki.project357.com/wiki/X264_Settings#slow-firstpass
[23:26] <juanmabc> he's... gone
[23:26] <JEEB> argh
[23:26] <llogan> jim
[23:27] Action: llogan disposes of remaining star trek references
[23:27] <JEEB> lol
[23:28] <JEEB> llogan, thanks for providing me with the info on the libavcodec-extra package, didn't know it worked like that
[23:28] <llogan> it enables other stuff too. not sure what else anymore. i haven't been keeping up with it after the libav adoption obviously
[23:29] <JEEB> probably all of the non-freetard-compatible stuff :)
[23:29] <llogan> and medibuntu also provides it which adds libfaac and probably something else
[23:29] <JEEB> thankfully libfaac should become relatively unneeded soon'ish :)
[23:30] <llogan> what's the news?
[23:31] <JEEB> fraunhofer's AAC encoder was released with a GPL'ish license via the big G, and wbs has written a library of it and is writing the stuff so it can be used with libavcodec
[23:31] <llogan> interesting. where in hell have i been?
[23:31] <JEEB> not LGPL, but better than the nonfree faac and libaacplus :)
[23:32] <intracube> x264 .ffpreset files are broken with recent versions of ffmpeg
[23:32] <JEEB> since it seemingly provides both LC AAC and HE AAC (and v2 of that IIRC)
[23:32] <intracube> does anyone know a fix?
[23:32] <JEEB> intracube, I don't think they're used any more?
[23:32] <JEEB> or come with ffmpeg
[23:32] <llogan> replace -vpre with -preset
[23:32] <llogan> and then use a preset listed in "x264 --fullhelp"
[23:33] <JEEB> or just look at the listing @ http://mewiki.project357.com/wiki/X264_Settings#preset
[23:33] <JEEB> :)
[23:33] <llogan> who is wbs? i have a terrible memory.
[23:33] <JEEB> https://github.com/mstorsjo/
[23:34] <intracube> llogan: JEEB: that works. thanks.
[23:34] <JEEB> np
[23:34] <llogan> JEEB: thanks
[23:34] <intracube> it would be nice if ffmpeg threw up a more helpful error message though... ;)
[23:35] <llogan> what was the message?
[23:37] <intracube> hmm, I can't post the command... trying again
[23:37] <intracube> I mean the error
[23:37] <intracube> "Invalid option or argument: 'directpred=3', parsed as 'directpred' = '3'"
[23:37] <JEEB> oh
[23:37] <JEEB> I guess the ffpreset file contains settings that are in one way or another unusable now
[23:38] <JEEB> as the x264's internal presets got usable
[23:38] <intracube> something like "ffpreset files deprecated" would be helpful
[23:38] <llogan> use a pastebin site to show your ffmpeg command and the complete console output
[23:38] <JEEB> well, the system itself isn't
[23:38] <JEEB> and ffpreset files are still usable for other encoders that don't have pre-made presets
[23:38] <JEEB> (and for custom settings for x264)
[23:39] <JEEB> but I guess since a newer ffmpeg compile won't copy new ffpreset files for x264 I guess it ended up reading an old ffpreset file and failing at parsing it :)
[23:39] <intracube> JEEB: thanks for the explanation
[23:40] <intracube> llogan: yes, I'll use pastebin in future
[23:40] <intracube> the encoding is working fine now
[00:00] --- Sat Jun 30 2012