[Ffmpeg-devel-irc] ffmpeg.log.20120612

[burek]

[01:00] <hwkxx> -r forces a framerate?
[01:12] <Freakshow> sacarasc: what's the difference between -fpre and -vpre?
[01:13] <Freakshow> sorry, my office lost power...
[01:14] <Freakshow> nm... I see in the docs now
[02:00] <ryoohki> what is in "vpre slow"?  what would the cli parameters look like?
[02:02] <llogan> hwkxx: yes. ffmpeg will drop or duplicate frames to reach your desired -r
[02:04] <llogan> ryoohki: ffmpeg now uses -preset slow. see the slow preset in "x264 --fullhelp" for that options that are used.
[02:05] <llogan> (in addition to default)
[02:05] <llogan> which is called "medium"
[02:09] <ryoohki> llogan: thanks!  what does it do?
[02:12] <llogan> ryoohki: a preset is a collection of options that gives you a certain encoding speed/compression ratio.
[02:12] <llogan> for general use use the slowest preset you have patience for
[02:12] <llogan> they are: ultrafast, superfast, veryfast, faster, fast, medium, slow, slower, veryslow, placebo
[02:12] <llogan> ignore placebo as it is a joke and a waste of time
[02:13] <llogan> tweaking various options is a waste of time as well. just stick with the presets.
[02:52] <thisisjet> can anyone help me with overlaying webcam onto a screencast (in ubuntu)?
[02:52] <thisisjet> details here: http://fpaste.org/5a37/
[03:23] <ryoohki> llogan: thanks!  so slow produces high quality video( i have a quad core)
[03:24] <hwkxx> you got slower/slowest
[03:24] <hwkxx> slowest 2 pass
[03:24] <ryoohki> hwkxx: -preset slowest 2 pass( like that on the cli?)
[03:25] <hwkxx> -preset slowest -pass 1
[03:25] <hwkxx> and then -preset slowest -pass 2
[03:26] <hwkxx> this is for two pass encoding
[03:26] <ryoohki> hwkxx: issueing the ffmpeg command twice?
[03:26] <hwkxx> yeah, once for the first pass and for the second pass
[03:26] <hwkxx> the first read the video information and stores into a log
[03:27] <ryoohki> hwkxx: a .log file?
[03:27] <hwkxx> and the second pass optimizes the encoding process by reading the log information
[03:27] <hwkxx> -passlogfile or smth like that
[03:27] <hwkxx> but its generated automatically
[03:28] <hwkxx> but using slowest and 2pass is a waste of time... unless you got a lot of motion
[03:29] <hwkxx> slowest does its job very well
[03:30] <hwkxx> look over the -vprofile as well
[03:43] <llogan> use two-pass with -b:v if you are targeting a specific output file size. use single pass with -crf for everything else.
[03:44] <llogan> in other words: you probably want to use -crf
[03:45] <llogan> ffmpeg -i input -c:v libx264 -preset slow -crf 24 output
[03:45] <llogan> sane range is 18-30ish
[03:45] <llogan> a lower value is a higher quality
[03:59] <ryoohki> i'm using: ffmpeg -i $INPUT -acodec aac -ac 2 -ab 160k -s 376x240 -vcodec libx264 -vpre ipod320 -preset veryslow -b 1200k -f mp4 -threads 0 -strict experimental ipodnano/OUTPUT.mp4
[04:00] <ryoohki> thereis a complaint about "-b 1200k"needing to be -b:a ot -b:v
[04:00] <hwkxx> the audio and video bitrate
[04:00] <hwkxx> some flags have changed,
[04:01] <ryoohki> the problem is that the input file is 5.3b and the output file is 24mb for a 8gb nano ipod
[04:01] <ryoohki> 5.3mb
[04:01] <llogan> -threads 0 is redundant
[04:02] <ryoohki> llogan: isn't there a way to say "-threads auto"?
[04:02] <llogan> try -crf 28 instead of -b 1200k
[04:02] <llogan> -threads 0 is now default
[04:02] <ryoohki> llogan: thanks!
[04:03] <ryoohki> the other complaint is "Incompatible sample format 's16' for codec 'aac', auto-selecting format 'flt'"
[04:04] <llogan> you can ignore that
[04:07] <ryoohki> llogan: i'd like to try two passes but i'd like to keep the output down to the vacinity of the original flv/mp4 from youtube
[04:08] <ryoohki> wow!  -crf 28 made a big difference
[04:08] <ryoohki> instead of 24mb it's 4.7mb
[04:08] <ryoohki> smaller than the otiginal 5.3mb file
[04:13] <llogan> quality will be ok, but it's such a small screen anyway
[04:13] <llogan> generally you use the highest crf value that still gives you an acceptable quality
[04:14] <llogan> then you can use the same preset and crf for the rest of your batch
[04:14] <llogan> and they will all be encoded with the same quality
[04:15] <ryoohki> llogan: what's the max threads for an amd quad core? 4?
[04:15] <hwkxx> 8
[04:15] <hwkxx> 2threads per core
[04:15] <llogan> i don't know. you don't need to declare threads. libx264 will automatically choose an appropriate value.
[04:16] <hwkxx> check out /proc/cpuinfo
[04:16] <hwkxx> and see the cpu numbers there
[04:17] <ryoohki> hwkxx: 4 - it's a quad core
[04:17] <llogan> -threads 0 is cores*1.5 for frame threads
[04:17] <hwkxx> threads not cores
[04:18] <hwkxx> a core is a hardware part :)
[04:20] <ryoohki> hwkxx: is there a way to get it dump the "-preset" values?  mplayer uses "help" like : mplayer -vo help
[04:20] <hwkxx> presets are stores along ffmpeg
[04:20] <hwkxx> stored*
[04:21] <ryoohki> hwkxx: so how can i get it to tell what the acceptable values are?
[04:21] <hwkxx> mine are here: /usr/share/ffmpeg/
[04:21] <hwkxx> they are a bunch of files containing arguments :)
[04:21] <ryoohki> hwkxx: yeah but those are -vpre not -presets
[04:22] <ryoohki> i have slow and slower but slowest doesn't seem to work
[04:23] <hwkxx> its veryslow my bad
[04:23] <hwkxx> :)
[04:24] <ryoohki> hwkxx: yup, thanks!
[04:31] <llogan> hwkxx: if you have a bunch of libx264 presets in /usr/share/ffmpeg/ (other than ipod*) then you're using an old ffmpeg
[04:32] <llogan> ryoohki: you can see the values x264 used with: strings input.mp4 | grep x264
[04:33] <hwkxx> compiled it yesterday
[04:34] <llogan> you can compile old source any day
[04:34] <hwkxx> but i dont think make install removed the existing directory contents
[04:34] <hwkxx> just replaced the needed ones
[04:34] <hwkxx> libvpx-720p50_60.ffpreset
[04:34] <hwkxx> got this
[04:34] <hwkxx> git pull doesnt lie :P
[04:35] <llogan> that's for the libvpx encoder
[04:35] <hwkxx> you were refering to this kind of presets?
[04:36] <llogan> you were telling ryoohki that the (libx264) presets were stored in /usr/share/ffmpeg/
[04:36] <hwkxx> yeah, most of them are from 2011 :)
[04:37] <llogan> recent ffmpeg does not install any libx264 presets files except some ipod presets.
[05:07] <ryoohki> when i switch -vcodec libx264 -vpre ipod32 to -c:v libx264 -vpre ipod32 i get ipod32 not found
[05:12] <hwkxx> -vcodec libx264 is the way to go
[06:08] <frogs> http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/144884
[06:09] <frogs> break build for me: libavcodec/acelp_filters.c:151:8: error: use of undeclared identifier 'HAVE_MIPSFPU' ( i am not using mips, regular amd64 )
[06:13] <drv> rerun configure
[06:23] <frogs> drv: it works when you build ffmpeg standalone, but it fails when mplayer builds libavcodec itself
[06:24] <ryoohki> is -c:a aac the same as "-acodecaac" ?
[06:30] <relaxed> yes, "-acodec aac"
[07:06] <ryoohki> relaxed: thanks!
[07:41] <JJ-Neo> Hi.
[08:28] <JJ-Neo> is there an open source script that allows to pull x amount of thumbnails from wmv videos?
[09:27] <shroomM> am I using libx264 presets right:
[09:28] <shroomM> ffmpeg.exe -i untitled_02.avi -profile:v high -level:v 41 -pix_fmt yuv420p -c:v libx264 -crf 10 -preset slow -c:a libfaac -b:a 128k test1.mp4
[09:28] <shroomM> ?
[09:28] <Mavrik_> yeah
[09:28] <Mavrik_> crf 10 is hellova low though :)
[09:28] <shroomM> I need it for a mezzazine file
[09:28] <shroomM> my question is why aren't the x264 defaults set?
[09:29] <shroomM> oh, I fail
[09:29] <shroomM> apparently they are
[09:29] <shroomM> :S
[09:29] <shroomM> I was sure they weren't
[09:29] <shroomM> i double checked
[09:36] <shroomM> ah, silly me, there's no reason for mbtree when gopsize=1 :)
[10:38] <rdb> I see that URLProtocol and register_protocol etc became private.  What's the currently accepted method to register a protocol in the new version?
[10:40] <rdb> ffurl_register_protocol is in url.h, which doesn't seem to be part of the public API anymore.
[13:24] <sgfgdf> hello, guys! a question about 1080i and 1080p. is it true that 1080i is 30fps and 1080p a 60fps? then AFAIK all network broadcasts today are 1080i and if you capture a 1080i signal is it possible to make it 1080p and stop seeing lines in fast actions?
[13:27] <Tjoppen> "it depends"
[13:27] <JEEB> nope, no frame rates are set in stone for resolutions
[13:28] <JEEB> but yes, most NTSC area broadcasts are 30i
[13:28] <JEEB> (30 "frames per second", interlaced)
[13:28] <Tjoppen> 60 Hz field rate
[13:28] <JEEB> yes
[13:28] <Tjoppen> sgfgdf: look into deinterlacing. there are several options
[13:32] <burek> sgfgdf, i means interlaced
[13:32] <sgfgdf> so if the original video is shooted at 30fps it can't be used as 1080p but only as 1080i?
[13:32] <burek> so 60 fps is not quite fps its 60 fields per second
[13:32] <burek> or 30 odd/even pairs
[13:32] <burek> per sec
[13:33] <burek> your problem is not frame rate but interlacing
[13:33] <sgfgdf> burek, okay as convention when someone says fps what it means actually?
[13:33] <burek> most probably displayed flickering rate :)
[13:33] <sgfgdf> or it depends if it is interlaced ot progressive scan?
[13:34] <burek> interlaced is done in 2 steps, in 1st step only odd lines of the screen are encoded (half the height of the video)
[13:34] <burek> in the 2nd step only even lines
[13:34] <sgfgdf> what make my video interlaced or progressive? my camera or something else?
[13:34] <burek> and the display rate has to be doubled (60) to achieve the "normal" 30 frames/sec display rate
[13:35] <sgfgdf> burek, yes i think i understand that with the odd/even
[13:35] <burek> so you can see entire image
[13:35] <burek> well, the source usually
[13:35] <burek> if you get your stream from a source and it's already interlaced, then you don't have too many choices
[13:35] <sgfgdf> burek, when you see camera that says can shoot 1080 @30fps what it means?
[13:36] <burek> it means that it can shoot at that rate (capture)
[13:36] <sgfgdf> a true fps or not
[13:36] <burek> but it doesn't mean anything about encoding
[13:36] <burek> (interlaced or not)
[13:36] <JEEB> you'll have to see the actual details about the camera to see if it can shoot interlaced or progressive pictures at that resolution/rate
[13:37] <sgfgdf> damn i get lost
[13:37] <burek> what is the model of your cam
[13:38] <sgfgdf> burek, i'm just asking i don't have camera in particular to see if it is 1080p or i. i just want to understand the difference.
[13:38] <natrixnatrix89> guys.. How do I rotate a video?
[13:38] <natrixnatrix89> I'm trying ffmpeg -vf "transpose=1" -I in.mp4 -acodec copy -vcodec copy out.mp4, but it doesn't really change anything.. What am I doing wrong?
[13:39] <burek> natrixnatrix89, you can't use -vcodec copy
[13:39] <burek> when you rotate your video you change everything
[13:39] <burek> you need to re-encode
[13:39] <natrixnatrix89> oh.. so I should use -vcodec libx264 ?
[13:39] <burek> probably
[13:39] <burek> sgfgdf, then look at the camera specs
[13:39] <JEEB> sgfgdf, interlaced is basically two pictures taken at a double rate, both half the height of the "resolution"
[13:39] <natrixnatrix89> and also.. the video filter should be applied before the -I, right?
[13:40] <burek> if it's interlaced, then you'll have double the fps, but that double comes from odd/even pairs of lines, so it's not true fps as with progressive
[13:40] <JEEB> on a progressive screen the two fields are usually shown interleaved
[13:40] <JEEB> which is why you get the lines during movement
[13:40] <sgfgdf> JEEB, but this is based on the capture device, right? or it is something based on the compresssing the video?
[13:41] <burek> if you would manually decode the interlaced stream using 60 frames per second, you would get your output at half the height of the original video
[13:41] <burek> because in each "frame" you would either loose odd or even lines
[13:41] <JEEB> sgfgdf, interlaced/progressive usually depends on whatever your camera/source/application/whatever gives out to you
[13:41] <JEEB> and then most video formats have separate encoding modes for interlaced and progressive
[13:41] <burek> natrixnatrix89, well.. not quite
[13:42] <natrixnatrix89> better after input?
[13:42] <burek> natrixnatrix89, the basic usage of ffmpeg is explained here: http://ffmpeg.org/ffmpeg.html
[13:42] <burek> ffmpeg [global options] [[infile options][‘-i’ infile]]... {[outfile options] outfile}...
[13:42] <burek> so, first you define your inputs, then your outputs
[13:42] <natrixnatrix89> so vfilter should apply to the outfile
[13:42] <burek> filter would go in between
[13:42] <sgfgdf> JEEB, if you have raw video with no compression the interlaced or progressive should depend only on the camera, right? no software and etc.
[13:43] <JEEB> sgfgdf, yes
[13:43] <JEEB> whatever gave you that video is what made it interlaced or progressive
[13:43] <sgfgdf> JEEB, if i have such raw video and it is interlaced i can't make it progressive then, right?
[13:44] <JEEB> you can't magically have the data come back that you have lost when you took half-sized pictures at twice the rate, but good deinterlacers can usually give you a relatively good result
[13:45] <sgfgdf> JEEB, but i have to loose quality in this action?
[13:45] <JEEB> not necessarily
[13:45] <sgfgdf> JEEB, but as you said i have to me a magician to do it.
[13:46] <JEEB> not really
[13:46] <sgfgdf> JEEB, how those deinterlacers do it?
[13:46] <JEEB> also, a deinterlacer is not a person
[13:46] <JEEB> it's a program, a video filter usually
[13:46] <JEEB> sgfgdf, high quality resizing, motion-compensated other stuff etc.
[13:46] <JEEB> stuff like QTGMC in avisynth is pretty amazing
[13:47] <JEEB> (which bases itself on multiple avisynth filters that do a lot of the heavy lifting)
[13:47] <JEEB> if you aren't a perfectionist there's of course yadif
[13:47] <JEEB> which is in ffmpeg as well
[13:47] <JEEB> it can give quite nice results as well
[13:47] <sgfgdf> JEEB, so probably it is something you can't explain here so easily and need more knowledge in this area from my part.
[13:48] <sgfgdf> JEEB, btw can i somehow check if a video is interlaced or progressive using ffmpeg?
[13:48] <JEEB> usually most video formats have the type in some kind of metadata
[13:49] <JEEB> so running ffprobe or something on them would let you check it
[13:49] <JEEB> if there is no metadata in the data you are getting, you will have to make some assumptions and that's always fun
[13:50] <hwkxx> and takes a lot of time :)
[13:51] <sgfgdf> JEEB, "Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 720x544 [SAR 1:1 DAR 45:34], 766 kb/s, 25 fps, 25 tbr, 25 tbn, 50 tbc" is that something to do with this yuv420p?
[13:52] <JEEB> no, the p there is just "planar"
[13:52] <JEEB> 4:2:0 YCbCr, planar
[13:52] <JEEB> (YCbCr is colloquially called YUV)
[13:53] <sgfgdf> JEEB, i don't have any other data about the video stream the other is about the audio. so in my example i can't understand if it is interlaced?
[13:54] <hwkxx> ffprobe -show_streams
[13:57] <sgfgdf> hwkxx, what i'm looking then for? i can't find interlaced or progressive anywhere in the output.
[13:58] <hwkxx> bummer :)
[13:59] <JEEB> sgfgdf, tbc being two times tbr IIRC is a sign of interlaced
[13:59] <JEEB> although I would like if ffprobe/ffmpeg would just tell you the flag like mediainfo does :)
[14:01] <sgfgdf> JEEB, what does they mean? tbc, tbr, tbn?
[14:02] <JEEB> I'm not completely sure, but tbr usually gives you the "frame rate" while tbc might be then giving the rate of the stream or whatever (which would be 50 fields per second in case of interlaced PAL)
[14:04] <sgfgdf> JEEB, so it they are the same it should be a progressive scan since frames and fields match, right?
[14:04] <JEEB> well, progressive pictures have no fields
[14:04] <JEEB> and thus the rate would be a frame rate
[14:04] <JEEB> also do try mediainfo's command line interface
[14:04] <JEEB> that should give you the flag of the stream
[14:04] <JEEB> in a more understandable way
[14:09] <sgfgdf> JEEB, is it enoguh if you have a camera that shoots 1920x1080 with progressive scan at 24fps (so that should be true frames per second)? because i heared that 24 is enoguh for the human eye.
[14:10] <JEEB> it shouldn't look bad, but unless you have compression problems having some extra frame usually isn't a bad thing (unless your end result has to be 24, then you probably want a mod24 resolution)
[14:11] <JEEB> *having some extra frame rate
[14:11] <JEEB> I've seen cameras shooting progressive 24, 25, 30, 50, 60 frames per sec :3
[14:11] <sgfgdf> JEEB, mediainfo tells me this -- "Scan type                                : Progressive" about the same previous video, so probably that with the tbc and tbr is not true.
[14:12] <JEEB> sgfgdf, funny -- if that's true someone herped a derp at encoding it if it still has mismatched lines during playback
[14:12] <JEEB> unless it's your screen and refresh rates derping
[14:12] <burek> or this http://bit.ly/JZ0tYS
[14:13] <torbjorn> what are valid x264 codec parameters for mp4?
[14:13] <sgfgdf> JEEB, ah so may be the metadata can be wrong?
[14:13] <burek> torbjorn, what exactly do you need
[14:13] <JEEB> sgfgdf, well it can be encoded as progressive while having interlaced content
[14:14] <JEEB> which would be pretty lame
[14:14] <JEEB> and other possibilities
[14:14] <sgfgdf> JEEB, what happens in this case? you can't get the true progressive, right?
[14:14] <torbjorn> Im trying to encode a wmv file with ffwmv3 to mp4 with x264
[14:14] <JEEB> torbjorn, let's just say it like this -- pretty much anything from libx264 is ok for mp4 muxing, while your decoding hardware/software usually has more limitations
[14:15] <JEEB> thus, it's up to the overall capabilities of what you're planning to play it back on
[14:15] <torbjorn> i see
[14:15] <burek> torbjorn, usually it will be enough to do something like: ffmpeg -i input.wmv -vcodec libx264 -acodec copy out.mp4
[14:15] <torbjorn> im getting Could not write header for output file #0 (incorrect codec parameters ?)
[14:15] <burek> can you please use pastebin.com, to show your command line and its output?
[14:16] <JEEB> sgfgdf, well it would just lose some possible hints and other stuff from the bitstream if incorrectly encoded
[14:17] <torbjorn> http://pastebin.com/WYSJaj4t
[14:18] <sgfgdf> JEEB, what would "hint" mean in your context? i just can't understand it. (i'm not a native english speaker)
[14:19] <JEEB> well stuff like "bottom/top field first", possible "drop these fields" hints and so forth
[14:19] <JEEB> extra information on the source
[14:20] <sgfgdf> JEEB, so basiacally it is just wrong to do such things, right?
[14:20] <sgfgdf> JEEB, better to leave it as it was interlaced (if it was at first place).
[14:20] <JEEB> yes, if you have not deinterlaced you should leave it as-is
[14:24] <sgfgdf> JEEB, when you mentioned above about cameras which shoot progressive at even 60fps isn't it too much? i assume it could be usefull for some slowmotion.
[14:24] <JEEB> it is more than some PCs and hardware decoding ASICs can decode, yes
[14:25] <JEEB> but as long as you don't have trouble with having bad quality at such a rate (lack of bitrate, bad compression algorithms), it should be OK
[14:26] <sgfgdf> JEEB, and how about 30fps progressive. is it decent nowadays?
[14:26] <JEEB> after all, you can then later cut off half of the frames, or do some motion stuff to make it 30fps
[14:26] <JEEB> why would it be any less decent than any other rate, unless you are trying to make something that in the end hast to be 24fps
[14:28] <sgfgdf> JEEB, so anything above 24fps (for some tolerance because of compression problems) is decent when it is progressive?
[14:29] <JEEB> well, you get more frames per second so the movement should be more fluid than with 24fps + if it's progressive you don't have to deal with fields and all that bullshit
[14:30] <JEEB> the only negative side is when you need to have the end result be 24fps
[14:30] <JEEB> then a mod24 frame rate is needed, for example 48fps or 120fps
[14:30] <JEEB> which are rare
[14:31] <JEEB> of course, you can encode fake-interlaced streams of 1080p30 onto blu-ray and such, so the cases where you need 24fps explicitly are relatively rare
[14:31] <JEEB> (flag the stream as interlaced yet encode progressive)
[14:33] <sgfgdf> JEEB, when you can need to have end result with exactly 24fps? and what is this mod24? is it some standard?
[14:34] <JEEB> no, mod24 is just a way of saying "divisible by 24"
[14:34] <sgfgdf> JEEB, ah i get it. and when it is needed?
[14:35] <JEEB> and you might need to have an end result of exactly 24fps depending on what you're doing. Although, as I said, f.ex. blu-ray would take in progressive 30fps content as long as you flag it interlaced (PAL 25fps is handled in the same way, as the specification doesn't have "progressive, 25fps" in it)
[14:35] <JEEB> but for example if your boss wants it to be exactly 24fps
[14:35] <JEEB> that's one reason :D
[14:36] <sgfgdf> JEEB, hm.
[14:39] <burek> torbjorn, your ffmpeg is old
[14:39] <torbjorn> damn, fedora is always late
[14:39] <burek> also
[14:39] <burek> audio codec
[14:39] <burek> complains with mp4
[14:39] <sgfgdf> JEEB, thank you very much for the help! thanks to burek alsto and all that answered me!
[14:39] <burek> :beer: :)
[14:40] <burek> torbjorn, try instead of "-acodec copy" something like this "-acodec aac -strict -2 -ab 128k -ac 2 -ar 44100"
[14:40] <sgfgdf> burek, btw do you have those lemon/beer thing in your country? it is very strrange mix between beer and lemon :)
[14:41] <burek> yes we do :) it's really cool now in the summer :)
[14:41] <torbjorn> burek: thanks, should have seen that bit
[14:42] <sgfgdf> burek, :) i also like those apple ciders (summersby) it has alcohol, but you can't even notice it.
[14:43] <burek> sgfgdf, yes, and after 2 of those, you don't know the way back home :)
[14:43] <burek> torbjorn, :beer: :)
[14:44] <sgfgdf> burek, if you watch euro 2012 you don't need to go home, just wait for the next game :)
[14:45] Action: sgfgdf afk
[14:51] <burek> :)
[15:13] Action: hwkxx truck thieves fast and furious style (my god) : http://www.youtube.com/watch?v=WpRs8tcBECE&feature=player_embedded
[15:44] Action: sacarasc pokes OHAiTHARU. :O
[15:44] <OHAiTHARU> ohai :D
[15:46] <OHAiTHARU> So, i'm trying to simply remake a .avi with a lower bitrate so that it can be streamed more easily. mp3 audio, ~1000kb/s vid bitrate, i'm thinking about bringing it down to 700. any help
[15:47] <sacarasc> ffmpeg -i input.avi -vcodec mpeg4 -b:v 700 -acodec copy output.avi
[15:48] <sacarasc> That might work...
[15:48] Action: sacarasc shrugs.
[15:53] Action: juanmabc worries why not -vcodec copy to have the same (-b remaining)
[15:54] <sacarasc> copy doesn't reencode, it just copies the stream. You have to encode to change the bitrate.
[15:55] <juanmabc> i would think it copying the codec type and since -b:v is specified it'd do the job
[15:55] <juanmabc> omiting -b:v of course nothing would happen
[15:55] <sacarasc> It doesn't copy the codec, it copies the stream.
[16:10] <Wakou> is there info liked date recorded inside .mp3 file? If so how can I look at it?
[16:11] <hwkxx> meta dta idv3
[16:11] <hwkxx> but you cannot rely on anything
[16:11] <hwkxx> or get the file system create date
[16:11] <hwkxx> which is not available on linux
[16:12] <hwkxx> unix*
[16:12] <Wakou> Can I 'look' at it with ffmpeg?
[16:12] <hwkxx> try ffprobe
[16:13] <Wakou> TY...
[16:13] <Wakou> :) not much there, oh well I'll just have to rename and sort themm manually..
[16:14] <hwkxx> ffprobe -show_format -show_streams can provide more info
[16:17] <Wakou> hwkxx: TY, handy to know..
[16:52] <phraktal77> Hi.. is it possible to stream an mp3 file using ffmpeg & ffserver ? I'
[16:53] <phraktal77> Does anyone have a functional example (ffserver config, and ffmpeg command) that streams an mp3 file ?
[19:30] <burek> phraktal77, you can use ffmpeg only
[19:30] <burek> ffmpeg -re -i input.mp3 -vcodec copy -f mpegts udp://remoteip:port
[20:30] <swkide> Greetings! I am looking for an example to create "http live streaming" segments with ffmpeg - I tried: ffmpeg -v debug -i $input -c copy -map 0 -f segment -segment_time 10 -segment_list out.list out%03d - but only get: Could not write header for output file #0 (incorrect codec parameters ?) - any hints?
[20:33] <swkide> and in Pastebin: http://pastebin.com/j1xBSnYm
[21:09] <ThiefMaster> what's the best coded for a video that needs to play on windows media player (on windows xp)? i tried msmpeg4v2 but the quality is horrible
[21:12] <ThiefMaster> *codec
[21:17] <llogan> ThiefMaster: what was your command?
[21:18] <ThiefMaster> ffmpeg -i whats-your-hobby.flv -vcodec msmpeg4v2 -acodec adpcm_ima_wav whats-your-hobby.avi
[21:20] <ThiefMaster> hm, actually when using wmv2 i also get the bad quality (visible blocks, like when you use very low quality in a jpeg file)
[21:24] <llogan> ThiefMaster: that's because the default settings are being applied and default is crappy for most encoders
[21:25] <llogan> use the -qscale:v option with a value between 2 and 5
[21:27] <llogan> or -qscale if you are using an older ffmpeg.
[21:27] <llogan> a lower value is a higher quality
[21:28] <ThiefMaster> ty, looks good
[21:28] <ThiefMaster> warning, clipping 1 dct coefficients to -127..127:04:49.88 bitrate=3031.0kbits/s - should i care about those warnings?
[21:28] <ThiefMaster> got quite a lot
[21:29] <llogan> only if the output doesn't play or looks shitty
[21:30] <llogan> http://support.microsoft.com/kb/316992
[22:19] <magic_1> hi all, hoping some one can help, im getting the following error, however even when i copy the file into the relevant directory it says it already exists? PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib64/php5/20090626/ffmpeg.so' - libavformat.so.52: cannot open shared object file: No such file or directory in Unknown on line 0 PHP 5.3.2-1ubuntu4.15 with Suhosin-Patch (cli)
[22:19] <magic_1> (built: May 4 2012 00:38:26) Copyright (c) 1997-2009 The PHP Group Zend Engine v2.3.0, Copyright (c) 1998-2010 Zend Technologies
[22:21] <hwkxx> this isnt supported by ffmpeg
[22:22] <hwkxx> you are better off running exec(ffmpeg)
[22:22] <hwkxx> than using the php module
[22:22] <hwkxx> i think you need to compile ffmpeg
[22:22] <hwkxx> because libavformat.so is not found on the system
[22:22] <hwkxx> you can find / | grep libavformat.so.5
[22:23] <hwkxx> to see if its really installed on your system
[22:24] <magic_1> i find libavformat.so.5 in /usr/lib/ dir
[22:24] <hwkxx> well
[22:24] <hwkxx> not in the right place
[22:25] <hwkxx> it should be here: /usr/lib64
[22:25] <hwkxx> since you got a 64 bit distro
[22:25] <magic_1> funny thing though is that in webmin my /usr/lib64 does dir looks like a shorcut and not a dir
[22:25] <hwkxx> compile ffmpeg
[22:25] <hwkxx> i think it will install the libs
[22:25] <hwkxx> in there
[22:26] <hwkxx> its a symlink probably
[22:26] <magic_1> will take a look at it now quick thanks
[22:31] <Wibble-> howdy :) I've got a video that I'm trying to convert to DVD format (with -target pal-dvd).  The video seems to convert just fine, but the audio goes from clear to nothing but random white-noise... See http://pastebin.com/ncyQtxhN - I don't understand the 'incompatible sample format' comment but I'm guessing it's the cause of the problem?
[22:35] <llogan> Wibble-: avconv isn't from FFmpeg, but we can probably help you here anyway.
[22:35] <Wibble-> oh, sorry... I was using ffmpeg, but it told me that I should use avconv now...
[22:35] <Wibble-> the same happens when I use the ffmpeg command
[22:35] <Wibble-> I can upload the paste if it would be helpful
[22:35] <llogan> that ffmpeg isn't from FFmpeg either.
[22:36] <llogan> yes, it's confusing.
[22:36] <llogan> debian/ubuntu switched to a fork of FFmpeg. i dont think any other mainstream distros have done the same
[22:37] <llogan> however the maintainer for the ffmpeg package is also a libav developer, so that has to be considered
[22:37] <llogan> does the audio sound ok for a while and then turn to static?
[22:38] <Wibble-> no
[22:38] <Wibble-> just static from the start
[22:38] <llogan> in all players?
[22:38] <Wibble-> let me test a couple of others
[22:38] <llogan> like ffplay
[22:38] <Wibble-> no audio for xine
[22:39] <Wibble-> static in ffplay
[22:39] <Wibble-> and it said:    Stream #0.1[0x80]: Audio: ac3, 48000 Hz, stereo, s16, 448 kb/s
[22:40] <llogan> does the input file sound normal?
[22:41] <Wibble-> Sorry - not sure how, but I've completely lost my desktop... Can't see anything... have to reboot.
[22:41] <hwkxx> :)))
[22:42] <llogan> i'll blame avconv
[22:42] <hwkxx> this could be the reason for no audio
[22:45] <iive> llogan: you are correct, the package maintainer switched debian to the fork on his own accord, because he was on the fork side.
[22:45] <Wibble-> I swear I'm going to switch away from ubuntu soon... That was the most bizzare thing.  When I clicked on a window it became invisible! Soon I only had the desktop, and no way (that I know of) to get it back...
[22:46] <Wibble-> Sorry if I didn't see any suggestions for the white static?
[22:46] <llogan> iive: any other distros?
[22:46] <iive> Wibble-: can you play that one with another player/library
[22:46] <Wibble-> mplayer and ffplay both have white noise, xine has no audio
[22:46] <llogan> Wibble-: you didn't miss anything
[22:47] <Wibble-> Shall I try reverting to the non-forked ffmpeg following the guide above?
[22:47] <llogan> it's definitelty worth a try
[22:47] <iive> Wibble-: mplayer -ac a52 file
[22:48] <Wibble-> iive, silence - Requested audio codec family [a52] (afm=liba52) not available.
[22:49] <iive> are you with mplayer2 or mplayer 1.1?
[22:49] <iive> doesn't matter
[22:49] <Wibble-> MPlayer svn r34540 (Ubuntu), built with gcc-4.6 (C) 2000-2012 MPlayer Team ... Not helpful for your question though!
[22:49] <llogan> i'm surprised that is in the repo
[22:49] <iive> both have removed that library.
[22:50] <llogan> (mplayer2)
[22:51] <iive> well, there are not only poisonous people, there are also cancerous people... :|
[22:51] <llogan> -loglevel debug -i input -target pal-dvd -vn -t 10 output.mpeg
[22:51] <iive> Wibble-: try to install the library, it may come with some front-end that could at least decode the sample to wav.
[22:52] <iive> Wibble-: does the sample play in any other player? wmp?
[22:52] <Wibble-> I'm just following the instructions for compiling on ubuntu so I'm back on the proper ffmpeg mainline
[22:52] <llogan> i'd be interested in seeing if ffmpeg from FFmpeg produces and/or plays it normally or not
[22:53] <llogan> in other words: i don't know what is causing the issue
[22:53] <iive> if it doesn't, please fill a bugreport and provide sample.
[22:54] <llogan> you can skip sections of that guide if you don't expect to use them (such as libvpx)
[23:16] <Wibble-> Many thanks iive and llogan - compiling from source fixed the problem.  Yet another reason to move away from Ubuntu!
[23:16] <iive> hehehe
[23:17] <hwk> this should be added in the topic
[23:17] <hwk> most users come with issues which are related to a distro package
[23:17] <Wibble-> it _also_ fixed the other major problem I was having that specifying -b changed audio and video bitrate, then because the audio bitrate changed it said it was invalid!
[23:20] <llogan> Wibble-: glad it worked for you. using recent ffmpeg is always a good first step when experiencing issues
[23:24] <Wibble-> What would your mpeg2 bitrate guess be to have very little quality loss from 740kbps in h264?
[23:25] <llogan> -target already sets the bitrates, IIRC
[23:25] <burek> Wibble-, also, you can try in the future with static version, if you don't like to compile things often: http://ffmpeg.gusari.org/static/
[23:25] <burek> just to check if something is a bug or not
[23:25] <Wibble-> thanks burek :)
[23:26] <llogan> burek: are those your builds?
[23:26] <Wibble-> It does llogan - but I'd like to squeeze more on to a DVD.  I've got about 3.5 hours of material :)  It's okay, I'll do some trial and error
[23:26] <llogan> oh, duh. they are
[23:27] <llogan> there are also these ones http://dl.dropbox.com/u/24633983/ffmpeg/index.html
[23:27] <llogan> burek: want a link to your builds on ffmpeg.org/download.html?
[23:28] <burek> well, feel free to put it there
[23:28] <burek> it's automated anyway
[23:28] <burek> I just check it from time to time to see if it works fine
[23:28] <burek> :)
[23:29] <llogan> ok, i'll add it. do you have a preference for the link name? Burek's Static FFmpeg Builds?
[23:29] <burek> well, no need for credits from my side, I'd prefer just Linux static builds
[23:29] <burek> or something similar
[23:30] <burek> but if the name is required for some reason, then ok, you can put it too
[23:30] <hwk> :)))
[23:44] <RobertNagy> what behaviour should I expect from avformat_seek_file when seeking past the end?
[23:48] <danielr> hi there, i was wondering if someone can help with my problem, i am trying to work out how to make thumbnails of a video using the interval command, however every time it will generate 2 black frames at the start which is messing up the system which uses the thumbnails via indexes. is there a better command for this
[23:49] <burek> danielr,
[23:49] <burek> -pb
[23:49] <burek> oops
[23:49] <burek> can you please use pastebin.com, to show your command line and its output?
[23:49] <burek> :))
[23:50] <danielr> absolutely
[23:50] <danielr> its got variables in a shell script but you get the idea
[23:51] <danielr> http://pastebin.com/KaMKwzSu
[23:51] <danielr> they are going into a contact sheet using montage
[23:51] <danielr> ill show some output
[23:51] <burek> -r ${INTERVAL} is not verz clear
[23:51] <burek> very*
[23:52] <burek> should be -r {FRAME_RATE}
[23:52] <danielr> http://static.electroteque.org/images/big_buck_bunny_480p_h264-0.jpg
[23:52] <burek> btw, what is your overall goal
[23:52] <danielr> interval is 0.1
[23:53] <LittleFoot> hi gaize
[23:53] <danielr> i mean 1/10
[23:53] <burek> ok
[23:53] <LittleFoot> why r my colors still messed up
[23:53] <LittleFoot> http://www.youtube.com/watch?v=41sht008sIU
[23:53] <danielr> overall goal is the contact sheet in the link
[23:53] <hwk> i got something whichi creates tile thumbs
[23:53] <burek> danielr, there was a video filter that detects black frames
[23:53] <hwk> someone from here helped
[23:53] <burek> let me check
[23:54] <burek> danielr, is your goal to avoid first 2 black frames
[23:54] <burek> or something else
[23:54] <danielr> ill check if its black
[23:55] <danielr> yeh its a fade in
[23:56] <danielr> is there a way to simply set an offset time
[23:56] <llogan> -ss
[23:56] <burek> yes, use -i ... -ss seconds
[23:56] <danielr> just a sec that might help
[23:56] <burek> or hh:mm:ss.mmm
[23:56] <burek> LittleFoot, what colors? i don't see anything messed up there
[23:56] <danielr> ill show the actual system I've built in a sec ;)
[23:56] <hwk> "$1" -i "$3" -an -vf drawtext="timecode='00\:00\:00\:00' :rate=24 :fontcolor=white :fontsize=21 :shadowcolor=black :x=5 :y=5",scale='min(300\, iw*3/2):-1',select="not(mod(n\,$5)),tile=4x12" -vsync 0 -frames:v 1 "$4"
[23:57] <LittleFoot> the colors are all psycodelic
[23:57] <hwk> $5 the frame interval
[23:57] <burek> LittleFoot, not on my display :) they look ok :)
[23:58] <burek> typical for CS capture
[23:58] <LittleFoot> let me check on my mac
[23:58] <llogan> i think he is referring to the purple and junk like that at 00:00:25
[23:58] <LittleFoot> but the textures flash weird colors sometimes
[23:58] <LittleFoot> yes
[23:58] <LittleFoot> see
[23:59] <llogan> how do you know youtube didn't take a dump on it?
[23:59] <llogan> did the file you uploaded look like that too?
[23:59] <burek> oh I see now
[00:00] --- Wed Jun 13 2012