[Ffmpeg-devel-irc] ffmpeg.log.20120114

burek burek021 at gmail.com
Sun Jan 15 02:05:01 CET 2012

[00:00] <llrcombs> but not on x264, yeah?
[00:00] <JEEB> with everything
[00:00] <JEEB> it's something you only use when you're developing something that needs the same qp values
[00:00] <llrcombs> heh, someone told me it didn't work with x264
[00:00] <JEEB> because qp is not something you use for normal encodes .-.
[00:00] <JEEB> well, in any case it's not something to use
[00:00] <llrcombs> how do you force it to be the same actual _quality_?
[00:01] <JEEB> you'll have to buy a few thousand chinese child laborers for that
[00:01] <llrcombs> or do you just have to mess around until you find a good compromise between size and quality?
[00:01] <JEEB> because there is no automated way to find the same quality for your eyes
[00:01] <JEEB> the closest alternative is to use -crf
[00:01] <JEEB> and find the highest crf value that still looks good for you
[00:01] <JEEB> start at around 24
[00:01] <JEEB> and if it looks bad, lower it
[00:02] <JEEB> values lower than 18 aren't usually needed, so don't lower it too much from 24 before trying stuff in the middle
[00:02] <llrcombs> 0 is lossless, right?
[00:02] <JEEB> 0 is lossless mode with 8bit libx264
[00:02] <JEEB> with >8bit libx264 the crf values start to go negative as well
[00:03] <JEEB> thus it's better off to use -qp 0 for lossless if you think you ever will use >8bit libx264
[00:03] <JEEB> if you use only 8bit libx264 you can use crf zero just fine
[00:03] <llrcombs> what would you reccomend for MJPEGs?
[00:04] <llrcombs> if I want the highest quality possible and the original file is MJPEG
[00:04] <JEEB> no idea, I only use ffvhuff and libx264
[00:04] <llrcombs> I used -sameq at first to go to libx264, since I didn't really understand what it did
[00:04] <JEEB> well, if you want to have a lossless copy of the file in H.264 then you can use -crf 0
[00:04] <JEEB> and have a lossless copy
[00:05] <llrcombs> won't that be ridiculously huge?
[00:05] <JEEB> well, depends
[00:05] <JEEB> it will be big of course, as it is lossless
[00:05] <JEEB> but whether or not it is bigger than the original is a separate matter
[00:05] <JEEB> of course, MJPEG is not lossless so I might even leave it as-is :P
[00:05] <llrcombs> true
[00:06] <JEEB> oh, highest quality possible as in output and not lossless?
[00:06] <JEEB> as in, "export"
[00:06] <JEEB> (last stage of whatever workflow)
[00:06] <JEEB> something that end users watch, and that isn't edited
[00:06] <llrcombs> as in, highest quality I can get from this original without lossless
[00:06] <JEEB> what are you going to be using this for?
[00:07] <JEEB> end users?
[00:07] <JEEB> editing?
[00:07] <JEEB> as in, only for watching, or a working copy / whatever?
[00:07] <llrcombs> sharing on Mediafire and YouTube (though I think no matter what my quality settings are, it won't make YouTube look any different)
[00:07] <llrcombs> watching
[00:07] <JEEB> k
[00:07] <JEEB> then what I just said
[00:08] <JEEB> highest crf that still looks good
[00:08] <JEEB> and setting preset to the slowest you can take should give you compression
[00:09] <llrcombs> http://www.youtube.com/watch?v=cUwjrIU9AsE&list=UU1qXejKCgpHeK-_DR8NDdkw&index=7&feature=plcp <-- I used -sameq and didn't specify crf, bitrate, etc... and ended up with this
[00:09] <llrcombs> looks a little better not on YT, MOVs are here: http://www.mediafire.com/?w4d6rz648sg8y#19mj8p79zcjje
[00:09] <JEEB> well, youtube'll re-encode your stuff in any case so to maximize profit there you might as well upload the original MJPEG there (or in case of you doing editing/whatever lossless H.264)
[00:10] <llrcombs> not bad for YT, IMHO
[00:10] <llrcombs> I'd have to reencapsulate the files for YT if I wanted to up the originals, as I'm pretty sure YT doesn't like THPs
[00:11] <llrcombs> and I'm not sure if the mjpeg format here is standard
[00:11] <JEEB> yeah, lossless H.264 via libx264 or ffvhuff or something
[00:11] <JEEB> should be the "thing as-is"
[00:12] <JEEB> also, -sameq's quality depends on the QPs of the original thing, and how they work on the output codec >_> When you mismatch codecs it becomes even more fun
[00:13] <llrcombs> just out of curiosity now, as you seem to know a fair amount about codecs, can you identify what format these captions are: http://www.mediafire.com/?w4d6rz648sg8y#a4j6s38nq6mz2
[00:13] <llrcombs> I've been trying to figure them out; I've never seen binary captions before
[00:13] <llrcombs> (mostly because one usually sees simple SRTs or SSAs)
[00:14] <llrcombs> they were all compressed with LZ77, but the files up'd at that link are extracted
[00:15] <JEEB> if this is out of some game there's a good chance they're in some weird proprietary format
[00:15] <llrcombs> yeah, that's what I've been thinking
[00:15] <llrcombs> ahh well
[00:15] <llrcombs> the video files they correspond to are a weird proprietary format too, but ffmpeg supports it!
[00:16] <JEEB> luckily enough :D
[00:16] <llrcombs> they're THP files, which are used solely in GameCube + Wii games
[00:16] <llrcombs> strangely-encapsulated not-quite-standard MJPEG with custom PCM audio
[00:16] <JEEB> looking at the size of the captions I'll guess it's image-based
[00:17] <llrcombs> I was guessing that it was in multiple languages, but image-based is possible too, I suppose
[00:17] <JEEB> (easier for the game console to deal with too, since it doesn't have to render text)
[00:19] <llrcombs> ahh well, it was worth a shot
[00:19] <llrcombs> I can manually remake subs for them, there aren't that many lines
[00:20] <JEEB> reminds me that I should see if I could continue the development of the ARIB caption app an internets friend of mine started
[00:20] <JEEB> (used in Japanese digital TV)
[00:21] <llrcombs> I've got a bug in trac right now about some rainbow artifacts in a few of the THP files on that page
[00:21] <llrcombs> (most of the ones from FE9)
[00:36] <dualbus> hi, I'm trying to stream live audio using ffserver; I'm capturing the audio from a network camera with a python script I made; the script outputs pcm_s16le, 4096 hz, 2 channels, and I'm piping that to ffmpeg
[00:36] <dualbus> the exact command is " python -u audio.py |  ffmpeg -vn -f s16le -codec pcm_s16le -sample_rate 4096 -channels 2 -i - -vn -f mp2 -codec libmp3lame -sample_rate 8000 -channels 2 http://localhost:8090/feed1.ffm"
[00:37] <dualbus> http://dpaste.org/oEH9T/
[00:37] <relaxed> why -f mp2?
[00:37] <dualbus> thats my ffserver.conf
[00:38] <dualbus> relaxed: I don't really know, I've tried ogg/mp2 as an output format
[00:38] <dualbus> and liblamemp3 and libvorbis as output acodec
[00:39] <dualbus> but all the ffmpeg command does is to capture some blocks, and then it exits (breaking the pipe)
[00:39] <dualbus> if I do the same command, but instead of pushing to the ffm location, I do record.mp3 it works correctly
[00:40] <dualbus> so I'm guessing i'm doing something wrong in the ffserver config, but I don't know what
[00:41] <dualbus> at first I thought it was my python script, since there is a delay between each output block (about 1sec)
[00:42] <dualbus> but if I transcode directly to hard-drive it works correcly
[00:48] <llrcombs> OHSHIT
[00:48] <llrcombs> ohwait, this is freenote
[00:48] <llrcombs> *freenode
[00:48] <llrcombs> I forgot what net I was on
[00:49] <llrcombs> I'm used to foonetic's netsplits, which take out about a quarter of whatever channel I'm in
[01:23] <AkiraBot> wow 0.8.10 released
[02:04] <PovAddict> does the mp4 container support embedded subtitles?
[02:05] <microchip_> yes
[02:05] <PovAddict> what codec should I use
[02:05] <PovAddict> ?
[02:06] <PovAddict> I have a .mp4 with h264 video and aac audio, and separately, a subtitle file in SRT format
[02:06] <PovAddict> what codec should I tell ffmpeg to transcode the subtitles into?
[02:08] <llrcombs> SSA?
[02:09] <PovAddict> I thought both SSA and SRT were things created by the "community"?
[02:10] <PovAddict> http://en.wikipedia.org/wiki/MPEG-4_Part_17 apparently this is what mpeg4 uses, does ffmpeg have any support for it?
[02:13] <muken> there is no software supports mpeg-4 part17 (mp4s with OTI=0x08)
[02:14] <PovAddict> "no software" is a bit of a strong statement, that wikipedia page says VLC can decode...
[02:15] <muken> i think it is 3gpp timed text. timed text != mpeg-4 part17
[02:19] <muken> fourcc for mpeg-4 streaming text is 'mp4s', and for 3gpp timed text is 'tx3g'. i have never seen 'mp4s' case
[02:20] <PovAddict> is there any subtitle format, embedded or separate-file, that is supported by Windows / WMP out of the box?
[03:10] <shintaku> thank goodness... there's a channel
[03:11] <shintaku> i'm trying to convert my massive 25GB movie into a smaller, sub 300MB mp4.  the current resolution of the movie is 1680x1050.  i'd like to half that.
[03:12] <shintaku> not really familiar with ffmpeg, but am trying:  ffmpeg -i ~/Desktop/iShowU-Capture.mov -s 840x525 -vcodec mpeg4 -an ~/Desktop/out.mp4
[03:12] <shintaku> swScaler: Unknown format is not supported as input pixel format
[03:13] <PovAddict> what does "ffmpeg -i ~/Desktop/iShowU-Capture.mov" alone say? put full output in a pastebin
[03:14] <shintaku> thanks
[03:15] <shintaku> http://pastebin.com/UgxWaeXF
[03:16] <shintaku> have a feeling it'll probably have something to do with the Apple Intermediate Codec
[03:16] <PovAddict> hm
[03:17] <PovAddict> try removing the -s
[03:17] <PovAddict> use "-t 10" to only convert the first 10 seconds
[03:17] <shintaku> as a test run?
[03:17] <PovAddict> yeah, trying without scaling first
[03:17] <Arthur_D> how do I use ffprobe to display width&height of all video files in a folder?
[03:18] <shintaku> same thing
[03:18] <shintaku> w/o the -s
[03:22] <PovAddict> I think that codec isn't supported by ffmpeg (yet)
[03:24] <shintaku> poo
[03:44] <CalimeroTeknik> how do I drop id3 tags when transcoding a file?
[04:04] <shintaku> ok, converted to m4v w/ quicktime.  i can now convert m4v to mp4.  when i do, though, it looks like crap.  any idea how to control quality.  i tried -b 1200.
[04:04] <shintaku> current settings are:  Stream #0.0(eng): Video: h264, yuv420p, 1152x720 [PAR 1:1 DAR 8:5], 629 kb/s, 14.86 fps, 1500 tbr, 1500 tbn, 3k tbc
[04:04] <shintaku> in m4v
[04:04] <PovAddict> that's 1200 bits per second
[04:04] <PovAddict> I don't think any video codec can give you sane quality at that bitrate ;)
[04:04] <PovAddict> use -b 1200k
[04:08] <shintaku> oh.. lol
[04:08] <shintaku> thanks
[04:08] <shintaku> still looks pixelated
[04:09] <PovAddict> are you sure the original video isn't like that already?
[04:10] <shintaku> yeah
[04:10] <shintaku> it looks good
[04:11] <shintaku> what is acodec?
[04:11] <shintaku> audio?
[04:11] <PovAddict> yes
[04:11] <PovAddict> what's the bitrate of the input video?
[04:11] <PovAddict> the m4v I mean
[04:14] <shintaku> 629 kb/s
[04:14] <shintaku> last q
[04:14] <shintaku> what's the diff between libx264 and mpeg4 codec?
[04:14] <shintaku> this looks great now:  ffmpeg -i ~/Movies/iShowU-Capture2/Resources/iShowU-Capture2\ -\ Computer.m4v -r 15 -t 10 -b 2M -bt 4M -vcodec libx264 -vpre hq -an ~/Desktop/out.mp4
[04:15] <shintaku> this looks like crap
[04:15] <shintaku> ffmpeg -i ~/Movies/iShowU-Capture2/Resources/iShowU-Capture2\ -\ Computer.m4v -t 10 -b 1629k -vcodec mpeg4 -an ~/Desktop/out.mp4
[04:15] <shintaku> i have no idea what the first set of options do
[04:15] <shintaku> i just copied it from something i saw online
[04:15] <PovAddict> err if your m4v was 629 kb/s it would look like shit too
[04:16] <shintaku> it's readable
[04:18] <PovAddict> "ffmpeg -i ~/Movies/iShowU-Capture2/Resources/iShowU-Capture2\ -\ Computer.m4v" => pastebin
[04:18] <shintaku> http://pastebin.com/zbQG6B2t
[04:19] <shintaku> i'm encoding it w/ that libx264 line i pasted earlier.  that looks fine
[04:19] <PovAddict> isn't it small enough already, then?
[04:19] <shintaku> yeah, just want it in mp4
[04:19] <PovAddict> it already is
[04:20] <shintaku> m4v = mp4?
[04:20] <shintaku> i don't know
[04:20] <PovAddict> ok, try this: ffmpeg -i ~/Movies/iShowU-Capture2/Resources/iShowU-Capture2\ -\ Computer.m4v -vcodec copy out.mp4
[04:20] <shintaku> will do.  thanks a bunch.
[04:40] <shintaku> well, thanks a lot for your help, PoVAddict.  heading to bed
[05:49] <PovAddict> hi
[05:49] <PovAddict> I have a 720x306 video which I want to transcode into a 16:9 DVD
[05:51] <PovAddict> the original resolution is something like 2.35:1, so I need either padding or cropping to fit 16:9
[05:52] <PovAddict> however, if I use -target pal-dvd and pad it, ffmpeg first scales it from 720x306 to 720x576 and then pads *that* into 1024x576, which is not an allowed resolution in DVD
[05:52] <PovAddict> dvdauthor somehow manages to create a .VOB but I don't think it's exactly compatible...
[06:32] <PovAddict> took me a while but I managed to do what I wanted - wheew
[06:59] <PovAddict> what's a reasonable bitrate for AC3 audio in a DVD?
[07:00] <PovAddict> I don't know about ac3 quality, but using the default 448 kb/s for AC3 seems way overkill considering my source material is 160 kb/s MP3
[12:47] <mvip> I'm having a somewhat odd issue. I need to turn a sequence of one or two images into a video. Let's say I have image1.png, then I will need to turn that into a 7 second long video. Any points on how to best go about to do that?
[12:48] <mvip> If i just do 'ffmpeg -i foo.png foo.mpg' that will generate a video that is a fraction of a second long
[13:19] <mvip> For the record, the command is "ffmpeg -loop_input -t $SECONDS -r 1 -i $INPUT -s hd1080 -vcodec libx264  $OUTPUT"
[17:55] <praedo> hello
[17:56] <praedo> i use crf and -t to specify a time limit
[17:56] <praedo> is it safe to predict the final bitrate of a 2 hour video considering the bitrate of a -t 60 chunk?
[17:57] <JEEB> nothing is really safe if you encode a part of a thing, but you should get a better view if you encode 500-1500 frame chunks from various parts of the video
[17:58] <JEEB> instead of just the first 60sec
[17:58] <relaxed> is there a reason why you wouldn't use 2 pass instead?
[18:03] <praedo> i don't know how
[18:03] <praedo> i learned a command line which is gives quite a good quality
[18:04] <praedo> and i understood that 2 pass would take a very long time
[18:04] <praedo> ffmpeg -i file -deinterlace -vcodec libx264 -preset medium -crf 20 -threads 0 -acodec libfaac -aq 100 output.mp4
[18:04] <praedo> this is what i always use
[18:04] <JEEB> if you want all encodes to have the same size for the same length (same average bitrate), then you use 2pass. If you just want to set quality, crf is what you want.
[18:05] <praedo> but sometimes i have to repeat everything by changing 20 to 22 or 24
[18:05] <JEEB> also, some places do a crf 1st pass, and then set a bitrate and do a second pass if the crf encode is too big
[18:05] <JEEB> (mostly video hosting sites)
[18:06] <praedo> i want good quality for all lenghts with a result of 1300-1400 kbps
[18:06] <JEEB> if you're that set for a bitrate
[18:06] <JEEB> just use 2pass .-.
[18:07] <praedo> okay... please tell me the basic commands
[18:07] <praedo> with this example:
[18:07] <praedo> ffmpeg -i file -deinterlace -vcodec libx264 -preset medium -crf 20 -threads 0 -acodec libfaac -aq 100 output.mp4
[18:07] <JEEB> -pass 1 -b:v 1400k for the first pass and -pass 2 -b:v 1400k for the second?
[18:07] <JEEB> and preset should be set to the slowest you can take
[18:07] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset <- listing
[18:07] <praedo> and both passes will take the same?
[18:08] <JEEB> nah, second pass is faster
[18:08] <JEEB> or well
[18:08] <relaxed> also, remove -deinterlace and add -vf yadif
[18:08] <JEEB> first pass will use faster settings
[18:08] <JEEB> and second pass will be faster because it doesn't have to do frame type decisions
[18:08] <JEEB> IIRC
[18:08] <praedo> so i don't specify medium? i get a good quality with medium but it takes 8 hours for a 45min video
[18:08] <praedo> pentium 4
[18:08] <llrcombs> JEEB: what's placebo, anyway?
[18:08] <llrcombs> "supermegaincrediblyslow"?
[18:09] <JEEB> llrcombs, everything up to eleven
[18:09] <JEEB> yeah
[18:09] <praedo> i want the same good quality i get with medium
[18:09] <JEEB> for benchmarks
[18:09] <praedo> how should the 2 pass commands look like
[18:09] <praedo> ?
[18:09] <llrcombs> JEEB: is it ever worthwhile?
[18:09] <JEEB> praedo, if you don't specify medium it will get set as the default .-.
[18:09] <JEEB> llrcombs, not really -- it's there for benchmarks
[18:09] <JEEB> just like ultrafast
[18:09] <llrcombs> JEEB: gotcha, so veryslow is usually your best bet if you can stand it
[18:10] <JEEB> although some people really don't care as they get double-digit frame rates with SD content with placebo
[18:10] <JEEB> lol
[18:10] <JEEB> praedo, take away -crf and add -pass 1 -b:v 1400k for the first pass and -pass 2 -b:v 1400k for the second pass?
[18:12] <praedo> good, and i don't specify medium?
[18:12] <JEEB> medium is default .-.
[18:12] <JEEB> so if you don't specify it, it will get automatically used
[18:13] <llrcombs> I've heard conflicting statements on this: does setting the preset adjust video quality, filesize, or both?
[18:13] <praedo> i see
[18:13] <Mavrik> llrcombs, mostly quality, filesize is dependant on bitrate
[18:14] <JEEB> llrcombs, it sets basically how much time is given to compression algorithms (read: how slow algorithms are used)
[18:14] <JEEB> when you use crf it can affect the file size, with bitrate encoding it affects the quality
[18:14] <relaxed> praedo: http://pastebin.com/raw.php?i=ERVw0xZd
[18:15] <JEEB> (note that the quality that crf values give are not exactly the same between settings because different algorithms change how the crf value gets calculated)
[18:18] <Mavrik> ah.
[20:07] <XeonBloomfield> Hello
[20:07] <XeonBloomfield> "-acodec vorbis" and "-acodec libvorbis" are two different things, right?
[20:18] <saste> XeonBloomfield: vorbis -> internal codec, libvorbis -> wrapper around libvorbis
[20:19] <XeonBloomfield> saste: I see
[20:19] <XeonBloomfield> saste: vorbis (internal) is saying little... to ass...
[20:20] <XeonBloomfield> saste: it convert about 5 times faster, but quality is terrible
[20:27] <praedo> JEEB, why -y in the 2nd pass? that will overwrite the file... so where are the stats of the 1st pass stored?
[20:28] <JEEB> I haven't said ANYTHING about -y >_>
[20:28] <JEEB> ask the guy who actually gave you that
[20:28] Action: JEEB smacks praedo for highlighting the wrong guy
[20:28] <praedo> http://pastebin.com/raw.php?i=ERVw0xZd
[20:28] <praedo> isn't that correct?
[20:28] <JEEB> yes, that is not me
[20:29] <praedo> i know
[20:29] <praedo> please have a look
[20:29] <praedo> isn't -y needed then?
[20:29] <JEEB> I don't care, I am watching video now so you can take the lower priority
[20:30] <JEEB> and the -y is probably in case you are actually outputting the 1st pass into a file instead of NUL or /dev/null
[20:30] <JEEB> that's all
[20:42] <diegoviola> hi
[20:43] <diegoviola> i want to extract part of some movie from 00:03:15 to 00:03:30, and encode this to ogg/theora/webm into another file, is this possible with ffmpeg?
[20:52] <falip> what is a standard video and audio codec for FLV?
[20:52] <Hyperi> New one is pretty much the same as mp4 iirc
[20:53] <Hyperi> falip: http://en.wikipedia.org/wiki/Flash_Video
[20:53] <falip> what about old one?
[20:53] <JEEB> falip, FLV can take VP3, VP62, Sorenson Spark, H.264 IIRC video-wise
[20:53] <JEEB> audio wise mp3, aac at least
[20:53] <JEEB> probably more
[20:53] <falip> ta
[20:53] <Hyperi> Oh right, I only use H.264 in my mp4 FLV
[20:54] <Hyperi> As it works for HTML5 browsers aswell nicely :P
[20:54] <Hyperi> <video> tag that is, and not restricted to Flash
[20:54] <JEEB> mp4 is what the container format in mpeg-4 specs is normally called .-.
[20:54] <falip> does ffmpeg set a container format based on extension, or is there a special switch?
[20:54] <JEEB> there is a format switch IIRC as well
[20:55] <Hyperi> like -vcodec parameter? :>
[20:55] <JEEB> nah, that's video codec
[20:55] <JEEB> but yeah, similar switch
[20:55] <falip> JEEB: there is -formats but that just lists formats.
[20:56] <JEEB> yeah
[20:56] <JEEB> I don't remember what the format setting switch was :P
[20:56] Action: JEEB usually sets the output file name in a way that ffmpeg can guess
[20:57] <Hyperi> 'man ffmpeg' might help :P
[20:58] <ioni> jeeB did you find out the problem?
[20:59] <JEEB> ioni, nope -- I don't have access to the box and helping the guy was too herp derp
[20:59] <JEEB> http://ffmpeg.org/ffmpeg.html is another good place to check stuff :3
[20:59] <ioni> k
[20:59] <falip> how do you distinguish between mpeg the container format and the codec?
[20:59] <falip> is it only a container format?
[20:59] <JEEB> uhh
[21:00] <JEEB> there are lots and lots of mpeg-specified container formats :<
[21:00] <falip> ah I see
[21:00] <JEEB> mpeg-ts , mpeg-ps , iso media | mp4
[21:00] <JEEB> etc. etc.
[21:00] <JEEB> usually they are distinguished by extensions, but of course that isn't always completely correct
[21:01] <JEEB> (someone can make an mkv file f.ex. and then rename it avi)
[21:02] <falip> there is such a thing as a mpeg audio codec(mp3), but no mpeg video codec is there? only codecs for the mpeg video format
[21:03] <falip> i mean
[21:03] <falip> there is such a thing as a mpeg audio codec(mp3), but no mpeg video codec is there? only codecs for the mpeg container format
[21:03] <falip> actually i suppose there is like mpeg1video
[21:04] <falip> so is mpeg set of container formats, and video codecs, and audio codecs?
[21:05] <falip> set=a set
[21:08] <JEEB> mpeg is an entity that does specifications of formats
[21:08] <JEEB> those can be a lot of stuff
[21:08] <JEEB> ranging from audio, video codecs f.ex. to containers and other similar stuff
[21:09] <JEEB> some mpeg video codecs: mpeg-1, mpeg-2, mpeg-4 part 2, mpeg-4 part 10 (also called H.264/AVC)
[21:11] <JEEB> audio wise you have: mpeg-1 layer 1, layer 2, layer 3 (layer 3 is also what you call "mp3")
[21:11] <JEEB> and then some others, including ALS in mpeg-4
[21:12] <JEEB> (ALS is part of mpeg-4 part 3, which also contains a lot of other formats .-. )
[21:25] <falip> thanks btw  I found a force format option -f fmt
[21:29] <JEEB> yeah
[21:29] <JEEB> that was it
[21:38] <diegoviola> anyone please?
[21:40] <sacarasc> diegoviola: -ss and -t
[21:40] <diegoviola> sacarasc: thanks
[21:41] <diegoviola> can i encode h264/mkv to webm?
[21:41] <diegoviola> with ffmpeg
[21:45] <sacarasc> Yes.
[21:46] <diegoviola> ok ty
[22:10] <falip> why do I get this error
[22:10] <falip> C:\>ffmpeg -i w.flv -vcodec svq1 -acodec libmp3lame w2.flv
[22:10] <falip> it says video codec not compatible with flv
[22:10] <falip> but surely sorenson spark is
[22:10] <falip> compatible with flv
[22:16] <JEEB> falip, then your ffmpeg version just doesn't support muxing it there or something .-.
[22:16] <JEEB> I'd personally use H.264/AAC within flv tho
[22:22] <falip> what about for mp4?
[22:23] <falip> would you do -vcodec mpeg4 -acodec libmp3lame ?
[22:24] <JEEB> nope
[22:24] <JEEB> at least if I wanted compression/good control of quality
[22:24] <JEEB> lol
[22:24] <JEEB> libx264 <3
[22:25] <falip> when do you use libx264 and when h264?
[22:26] <JEEB> h264 is the H.264 decoder, libx264 is a mighty good H.264 encoding library
[22:26] <falip> howcome one only specifies one then when doing -vcodec?
[22:26] <falip> aren't both a decoder and encoder necessary to convert from one to another?
[22:27] <JEEB> yes, but the decoder is usually automatically used if the format is identified
[22:27] <sacarasc> If you want to use h264 for decoder, you can use -vcodec h264 in the input section.
[22:27] <JEEB> ^
[22:27] <JEEB> pretty much "set -vcodec before -i", right?
[22:27] <JEEB> after -i it becomes the encoder selection
[22:28] <falip> is it better to do -ss and -t before -t or after?
[22:28] <falip> I mean before -i or after?
[22:31] <falip> i think i got it.. as an output is slower but more accurate
[22:31] <falip> for -ss anyway is what i've got noted down from some time.
[22:31] <falip> to do with keyframe or something. like whether it cuts at that.
[22:36] <falip> I just tried to convert an mpg to mp4, libx264  and I got the errors
[22:36] <falip> [libx264 @ 01e0d010] broken ffmpeg default settings detected
[22:36] <falip> [libx264 @ 01e0d010] use an encoding preset (vpre)
[22:37] <JEEB> you have an old ffmpeg >_>
[22:37] <JEEB> get a new one
[22:37] <falip> ok but out of interest what would I have done?
[22:37] <JEEB> back then you had to use the ffmpeg's presets that would override ffmpeg's then-broken defaults
[22:37] <JEEB> now ffmpeg uses libx264's defaults
[22:38] <JEEB> and has -preset to set libx264's presets
[22:38] <falip> what defaults was it set to back then?
[22:38] <JEEB> Something Quite Awful
[22:38] <JEEB> awful enough to make libx264 block their usage :P
[22:38] <falip> was it the settings for another codec or something?
[22:38] <JEEB> well
[22:39] <JEEB> they pretty much had one line of defaults for ALL codecs
[22:39] <falip> heh
[22:39] <JEEB> I think it was either 200kbps or 300kbps, with some very basic features used >_>
[22:39] <falip> so now it autodetects?
[22:39] <falip> or does it piss off all the other codecs?
[22:40] <falip> piss off   = slang, for annoy
[22:40] <JEEB> nah, now it just uses libx264's defaults for libx264
[22:40] <falip> british slang
[22:40] <JEEB> which is the sane thing to do
[22:40] <JEEB> no idea about other codecs
[22:40] <falip> k
[22:40] <JEEB> also, overall libx264 usage got a lot easier
[22:41] <JEEB> you got -preset to set libx264's presets, -profile:v to set libx264's profile, and -level for libx264's level setting
[22:42] <JEEB> before those additions got in, I would've never used ffmpeg to encode libx264 .-.
[22:42] <JEEB> because you basically had to write down the settings in a very herpy derpy way
[22:43] <diegoviola> ffmpeg is awesome
[22:43] <diegoviola> -t and -ss worked great for me, thank you all
[22:44] <Mavrik> JEEB, is  it -profile:v now? :)
[22:45] <JEEB> Mavrik, just setting -profile works too unless you're using aac encoding >_>
[22:45] <JEEB> because nowadays profile tries to get applied to both, and aac encoding seems to have "profile"
[22:45] <JEEB> thus, setting -profile:v will only have it affect video
[22:45] <Mavrik> yeah, I thought it was "-vprofile" so I'mnot really sure which is the "last" version :P
[22:46] <JEEB> -profile:v seems to be the final way
[22:46] <JEEB> (for now)
[22:46] <Mavrik> ^^
[22:47] <Mavrik> eh, I'll deal with it when my tests fail with -vprofile -_-
[22:47] <JEEB> yeh
[22:48] <huglester> guys where I can find information on how to put watermark over video? I'm using 0.9.. so searchign for fresh docs... I trie to look in the archive, but it has some files which probably need to be compiled somehow to open them
[22:49] <Mavrik> huglester, http://ffmpeg.org/libavfilter.html
[22:49] <huglester> ah.. so it's done via libavfilter? thanks
[22:50] <Mavrik> huglester, yeah, it's a video filter
[22:50] <JEEB> the overlay video filter
[22:50] <JEEB> IIRC
[22:50] <Mavrik> huglester, you're looking for the overlay filter
[22:51] <Mavrik> oh, I'm late -_-
[22:51] <huglester> thank you :)
[22:51] <huglester> Yeah I was going to ask it.. because could not find the word 'watermark' in the docs
[22:51] <huglester> but after I searched for '.png' I foudn out about overlay :)))
[22:52] <falip> I think I just got the ffmpeg command to convert an mpg to mp4  -vcodec libx264 without any -profile or -level
[22:53] <JEEB> sure
[22:53] <JEEB> you don't need to set those unless you are encoding for some limit
[22:53] <JEEB> just make sure you have a new'ish ffmpeg and you'll start getting sane settings >_>
[22:53] <falip> yeah this is the latest one
[22:53] <falip> the previous one gave the error
[22:53] <JEEB> yup
[22:53] <falip> when you say some limit, what kind of limit?
[22:54] <JEEB> hardware devices, etc.
[22:54] <JEEB> profiles = feature limits
[22:54] <JEEB> level = memory limits
[22:54] <falip> oh ok
[22:54] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset <- also, a listing for libx264's presets
[22:54] <JEEB> (basically use the slowest preset you can take speed-wise :))
[22:54] <falip> is there anything that lists recommended video and audio codecs for each common container format?
[22:55] <JEEB> not that I think
[22:55] <falip> when ffmpeg wasn't good for libx264, what did you use?
[22:56] <Mavrik> falip, instead of checking which codec for container, rather check which codec will work best for content and target devices and then choose container accordingly
[22:56] <JEEB> x264's command line encoder
[22:56] <JEEB> I still am better with controlling x264cli than libx264 via ffmpeg :D
[22:57] <falip> Mavrik: well i can find like one piece of software isn't good at playing a file so I need to convert it to something else
[22:58] <Mavrik> *shrug*
[22:58] <Mavrik> pretty much anything can play H.264/LC-AAC in MP4 nowdays, so that's a very safe bet
[22:58] <falip> they all seem just as good in terms of quality, when they play!
[22:58] <Mavrik> unless you have some special demands
[22:59] <falip> well, i have some old windows machines with a WMP that won't see mp4
[22:59] <falip> and i have some software to split videos, that won't play mp4 it seems
[22:59] <falip> xilinx, they have a piece of software for combining them that works properly really well
[22:59] <falip> so i like to use their splitter too
[23:00] <Mavrik> huh
[23:00] <Mavrik> then you'll have to check what those WMPs can play
[23:00] <falip> I mean xilisoft is the software.
[23:00] <falip> WMP=windows media player
[23:00] <falip> no WMPs ;-)
[23:01] <falip> sometimes I play things in an early WMP that won't accept mp4 at all.
[23:03] <huglester> guys, anything here does seem invalid for you here? /usr/local/bin/ffmpeg -y -i 4_4734464ac1fc386a1a79c4f036e2e7af.avi_240.mp4 -vf "movie=watermark_video.png [logo]; [in][logo] overlay=10:main_h-overlay_h-10 [out]" watermark.mp4
[23:04] <huglester> I am hetting these errors: http://d.pr/iD3X
[23:06] <Mavrik> hmm... either you don't have png support compiled into ffmpeg or you confused it with it
[23:07] <huglester> I'll just try try jpeg  now, to see if there are any changes
[23:07] <Mavrik> huglester, check ffmpeg -codecs to see if your ffmpeg supports png
[23:10] <huglester> Mavrik: seems no :( giving .jpg file worked!
[23:10] <huglester> need to search on how to enable png support
[23:11] <huglester> Thank you very much
[23:14] <huglester> Mavrik: do I need ffmpeg with png support, or avfilter..?
[23:16] <falip> what does this line mean?
[23:16] <falip> EV    libx264         libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
[23:16] <falip> encoder, video OK
[23:16] <falip> but what's the H.264 / AVC thing mean?
[23:17] <falip> AVC is a different video codec, so what does it mean in that context?
[23:17] <microchip_> falip: AVC = advanced video coding, which is another name for H.264
[23:23] <falip> and sorenson is another name for H.263?
[23:23] <JEEB> nah, sorenson spark is a H.263-based codec
[23:23] <microchip_> yes, although it's a modification of .263
[23:23] <microchip_> H.263*
[23:23] <JEEB> http://wiki.multimedia.cx/index.php?title=Sorenson_Spark
[23:28] <falip> so when -codecs has lines like this
[23:28] <falip> DEVSD  flv             Flash Video (FLV) / Sorenson Spark / Sorenson H.263
[23:28] <falip> EV    libx264         libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
[23:28] <falip> the right hand side is a mix really
[23:29] <falip> 'cos for the libx264 line AVC is another name for H.264
[23:29] <falip> but for the flash line,  Sorenson is not another name for flash
[23:29] <falip> so what does that right hand side set out to achieve?
[23:30] <falip> also is there a quick way to cut 5 seconds off a long video in ffmpeg?
[23:31] <falip> using -ss 0 -t .. takes quite a while.
[23:31] <falip> i'd rather have a cut function maybe it'd be quicker
[23:31] <falip> than an extract function
[23:33] <JEEB> sorenson spark was IIRC the first codec in 'flv' flash video, no?
[23:33] <JEEB> thus it's like mpeg4
[23:33] <huglester> I'm lost :( how do I compile png support in? maybe I need to compile ffmpeg with: --enable-decoder=png ?
[23:34] <JEEB> which is really mpeg-4 part 2
[23:34] <JEEB> (the first video codec in mpeg-4)
[23:40] <diegoviola> how do i tell ffmpeg not to encode audio while encoding some video
[23:41] <falip> Is AVC a name for the encoder and the decoder?
[23:41] <sacarasc> diegoviola: -an or -acodec copy
[23:41] <diegoviola> sacarasc: i mean i don't want audio at all in the output file
[23:41] <diegoviola> i want to exclude it
[23:41] <sacarasc> Then the first one.
[23:41] <diegoviola> ty
[23:42] <JEEB> falip, the video format in itself is called MPEG-4 Part 10 AVC / ITU-T H.264 or whatever
[23:42] <JEEB> the name of the decoder in ffmpeg is h264
[23:42] <JEEB> and the encoder available to encode in that format is libx264
[23:43] <falip> diegoviola: -an means no audio
[23:43] <falip> if i recall. and -vn means no video
[23:44] <diegoviola> falip: thanks
[23:44] <falip> wow, so h264 and h.264 are different things, such fiddly syntax!
[23:44] <microchip_> falip: no they're not
[23:44] <JEEB> nope, ffmpeg's codec's name just doesn't have the dot there :P
[23:45] <falip> yeah JEEB I call that fiddly syntax!
[23:46] <JEEB> well, it's usual to drop stuff like that on command line usually .-.
[23:46] <falip> and do any decoders have a name like h264.. or only libx264?
[23:47] <falip> doh i mean
[23:47] <falip> and do any encoders have a name like h264.. or only libx264?
[23:47] <JEEB> external encoders have a name like "libSOMETHING"
[23:47] <microchip_> falip: nope, only libx264
[23:47] <JEEB> internal encoders/decoders have a name set inside ffmpeg
[23:47] <falip> well, libmp3lame-yes. But mp2 is just mp2.
[23:48] <JEEB> yes, because it's internal
[23:48] <JEEB> lib = library
[23:48] <falip> ok
[23:48] <falip> so i guess the ones that were around when ffmpeg was built!
[23:48] <falip> and the others are external?
[23:48] <microchip_> falip: whay do you even care how it's called? just use it :P
[23:49] <microchip_> falip: it has little to do with when ffmpeg was built
[23:49] <falip> microchip_: you mean just call it ;-)
[23:49] <microchip_> falip: it has to do with whether there's an ffmpeg implementation of a specific codec or not
[23:50] <falip> well maybe initially they reinvented the wheel then later they decided to use libraries more.
[23:51] <falip> though their philosophy as to which they made internal and which they made external isn't important and certainly wouldn't be of interest to you microchip_ if you think even what it is called doesn't matter much!
[00:00] --- Sun Jan 15 2012

More information about the Ffmpeg-devel-irc mailing list