[Ffmpeg-devel-irc] ffmpeg.log.20131213

burek burek021 at gmail.com
Sat Dec 14 02:05:01 CET 2013


[00:00] <SirCmpwn> thanks a bunch
[00:00] <SirCmpwn> will this fail if no audio streams are present
[00:00] <llogan> negative mappings are also do-able: -map 0 -map 1 -map -0:s -map -1:s
[00:01] <llogan> iDunno. try it.
[00:01] <SirCmpwn> yeah
[00:07] <SirCmpwn> the answer is yes, if you were wondering
[06:29] <SirCmpwn> is there a way to convince ffmpeg to keep a similar output bitrate compared to the input bitrate when converting ogg to mp3 (and back again)
[06:38] <relaxed> SirCmpwn: why not use a lossless codec like flac or alac?
[06:38] <SirCmpwn> I need to play it in the browser
[06:39] <SirCmpwn> and I want it to get as close to the original quality as possible
[06:39] <SirCmpwn> it's worth noting that an ogg file can exceed the maximum mp3 bitrate
[06:40] <relaxed> -c:a libmp3lame -q:a 0 will give you the highest quality
[06:41] <SirCmpwn> will that waste space if the input quality is lower than 320 kbps?
[06:43] <relaxed> test it and see
[06:43] <SirCmpwn> yeah
[06:45] <SirCmpwn> relaxed: looks like it tries to choose the best bitrate to represent the actual content in, so far as I can tell
[06:46] <SirCmpwn> the maximum bitrate for an mp3 is 320kbps, and it gave me a file with 257kbps
[06:47] <relaxed> yes, you can do the same with libvorbis
[06:47] <SirCmpwn> excellent, tahnks for the help
[06:48] <SirCmpwn> thanks*
[06:48] <relaxed> for libvorbis the quality scale is reversed: 10 highest - 1 lowest
[06:48] <relaxed> -q:a 10
[06:48] <LithosLaptop> lwas just going to say that :)
[06:48] <SirCmpwn> thanks, I would have screwed that up
[06:54] <relaxed> according to `man oggenc` the default is 3
[07:09] <SLAYBoz> what is better  ffmpeg aac encoder or faac aac encoder
[07:12] <relaxed> libfdk_aac
[07:12] <SLAYBoz> huh
[07:12] <relaxed> libfdk_aac is the recommended aac encoder
[07:13] <SLAYBoz> what is that
[07:13] <relaxed> but between the two options you gave, faac
[07:13] <SLAYBoz> i see
[07:13] <SLAYBoz> what is libfdk_aac then
[07:14] <relaxed> "libfdk_aac is the recommended aac encoder"
[07:14] <relaxed> but you have to compile ffmpeg with support for it if it's not already
[07:14] <SLAYBoz> who made libfdk-aac
[07:15] <relaxed> https://trac.ffmpeg.org/wiki/UbuntuCompilationGuide
[07:16] <SLAYBoz> i don't get it
[07:17] <relaxed> what don't you get?
[07:17] <SLAYBoz> who made libfdk-aac
[07:17] <relaxed> Fraunhofer
[07:18] <relaxed> use google for more info
[07:18] <SLAYBoz> then why would ffmpeg support it
[07:18] <SLAYBoz> ffmpeg has their own aac
[07:18] <relaxed> ffmpeg has their own vorbis too, and it sucks
[07:18] <relaxed> so we recommend libvorbis
[07:19] <SLAYBoz> then what vorbis doesn't suck
[07:19] <relaxed> LIBVORBIS
[07:19] <SLAYBoz> i heard nero aac is best
[07:19] <relaxed> then use that
[07:20] <SLAYBoz> which libnero_aac  is that it
[07:21] <relaxed> ffmpeg has no support for nero's encoder
[07:21] <SLAYBoz> i see, is it because it's closed?
[07:25] <SLAYBoz> what is libfdk-aac
[07:26] <SLAYBoz> sorry i mean libvo_aacenc
[07:27] <relaxed> an aac encoder
[07:27] <SLAYBoz> who made that one
[07:28] <relaxed> google
[07:28] <SLAYBoz> but fraunhofer one is better?
[07:28] <SLAYBoz> and what is this one?  libaacplus
[07:28] <relaxed> for fuck's sake
[07:29] <SLAYBoz> what's wrong?
[07:35] <relaxed> I've told you the recommended one three times.
[10:50] <BoR0> I have one video with the following parameters: Stream #0:0[0x21]: Video: h264 (High) ([27][0][0][0] / 0x001B), yuv420p(tv, bt470bg), 720x576 [SAR 16:11 DAR 20:11], 25 fps, 25 tbr, 90k tbn, 50 tbc
[10:50] <BoR0> let that be video A
[10:50] <BoR0> I have video B with the following parameters: Stream #0:0[0x100]: Video: h264 (High) ([27][0][0][0] / 0x001B), yuv420p, 400x300 [SAR 1:1 DAR 4:3], 29.97 fps, 29.97 tbr, 90k tbn, 59.94 tbc
[10:50] <BoR0> how can I change video B to video A so that resolution, SAR/DAR fps and tbr are equivalent?
[12:49] <stonie_> hi.. can i produce mp4 with interlacing tff?
[12:55] <brontosaurusrex> stonie_, probably using x264 options http://mewiki.project357.com/wiki/X264_Settings#tff
[12:56] <stonie_> tried -top 1 with -flags ildct+ilme
[12:56] <stonie_> ty lemme read
[12:57] <stonie_> how can i set "-tff"
[12:58] <brontosaurusrex> try just -ttf after other x264 settings
[13:09] <brontosaurusrex> stonie_, working?
[13:13] <stonie_> mh i dont know exactly.. it shows MBAFF only (with mediainfo) ... is that correct?
[13:46] <benbro> does mp4 have meta info in it?
[13:47] <benbro> such as markers of a timeline
[13:53] <Mavrik> benbro, depends what do you mean
[13:53] <Mavrik> but yeah, mp4 has meta data like most formats :)
[13:57] <benbro> Mavrik: I have a video created from a slideshow and I want to check if there is info of when there is a slide transition
[14:06] <brontosaurusrex> stonie_, should be fine
[14:33] <benbro> how can I extract all keyframes?
[14:34] <benbro> I've tried the following but get a muxing overhead error
[14:34] <benbro> ./ffmpeg -i test.mp4 -vf "select=eq(pict_type\,I)" -qscale:v 2 frame-%03d.png
[14:55] <rcombs> https://trac.ffmpeg.org/ticket/1582 <-- so, what exactly is +full_chroma_int, and why isn't it on by default?
[14:57] <rcombs> best explanation I've found is "full chroma interpolation", but that seems rather vague, and I can't find anything documented that says it severely reduces banding and fixes discoloration
[15:13] <stonie_> brontosaurusrex, ok thanks .. the receiver cant import those files directly, the gotta define the field order by hand before ingesting
[15:13] <stonie_> what command line would you use for mp4 full hd (1920x1080) in good quality?
[15:15] <stonie_> right now i came up with "ffmpeg -i testmaterial.avi -y -pix_fmt yuv420p -acodec aac -b:a 192k -vprofile high -preset slower -ac 2 -strict experimental -threads 0 -vcodec libx264 -b:v 10000k -aspect 16:9 -r 25 -s 1920x1080 -wpredp 2 -top 1 -flags ildct+ilme -coder 1 -flags +loop -me_method umh -subq 8 -me_range 16 -g 250 -keyint_min 25 -sc_threshold 40 -i_qfactor 0.71 -b_strategy 2 -qcomp 0.6 -qmin 10 -qmax 51 -qdiff 4 -bf 3 -refs 4 -trellis 1 -par
[15:15] <stonie_> titions +parti8x8+parti4x4+partp8x8+partb8x8 outfile.mp4"
[15:16] <JEEB> ugh
[15:16] <JEEB> stop setting settings manually that can be handled with presets :|
[15:17] <JEEB> many of those things are either current or previous defaults, and you shouldn't have to touch them at all
[15:17] <stonie_> ok - how would u optimize that command? or, which are via presets
[15:18] <JEEB> most
[15:19] <stonie_> which preset would you use?
[15:20] <JEEB> just use the slowest one that is fast enough for you, and limit the reference frames if some hardware needs that
[15:20] <JEEB> (because setting the level only sets the flag, doesn't limit refs [unlike with x264cli])
[15:22] <stonie_> ok ty
[15:22] <rcombs> stonie_: also, you probably want -crf <something> and -maxrate 10000k instead of setting a constant bitrate (which is wasteful)
[15:22] <JEEB> it's not even constant :P
[15:22] <JEEB> it's just one-pass ABR
[15:22] <stonie_> ok ill try that tyvm
[15:22] <rcombs> yeah
[15:23] <rcombs> either way, -b:v is rarely what you're looking for with x264
[15:24] <JEEB> basically ffmpeg -i testmaterial.avi -y -c:a aac -ac 2 -b:a 192k -strict experimental -c:v libx264 -preset your_favourite_preset (your possible profile/level/ref LIMITATIONS here) (your rate control stuff here) (possible interlacing-related stuff here) out.mp4
[15:24] <JEEB> I separated the audio to the front because you had it all over the place in your example :P
[15:25] <JEEB> also you don't need to SET high profile, it WILL get used IFF it is needed, you only set the profile when you want to LIMIT the encoder to a feature set
[15:25] <DaFrameSkipper> what does "-skip_threshold" "-skip_factor" "-skip_exp" do? and how do use them
[15:25] <grkblood13> I got that thing working rcombs
[15:25] <JEEB> if your preset doesn't use the features, then it doesn't really help if you try to set an overly high profile flag
[15:25] <rcombs> grkblood13: good!
[15:25] <stonie_> JEEB, ok thanks.. ill try some more variations...
[15:25] <stonie_> but, how do i set the interlaced stuff then? it needs to be tff
[15:27] <DeadSix27> someone is really called brontosaurusrex on here?
[15:28] <rcombs> DeadSix27: yup. There's also someone called DeadSix27.
[15:28] <DeadSix27> :O
[15:28] <DeadSix27> *shocked*
[15:29] <DeadSix27> well need to go.
[15:30] <DaFrameSkipper> anyone have any what "-skip_threshold" "-skip_factor" "-skip_exp" does and how do use them?
[15:30] <JEEB> stonie_, that would go into the (possible interlacing-related stuff here) place :P You will have to excuse me for not knowing those by heart. Looking at http://ffmpeg.org/ffmpeg-all.html there are multiple "interlace" filters, there's plenty of "make this stuff interlaced" filters at least. Although if your input is already interlaced I wouldn't be surprised if it was already being encoded as such?
[15:30] <rcombs> ack, interlacing
[15:30] <rcombs> please, kill me now
[15:31] Action: saste kills rcombs
[15:32] <JEEB> also I recommend not taking mediainfo's information at too much face value, even libx264's terminal output when it starts encoding is much better than that
[15:32] <stonie_> ill try some more.. thanks alot
[15:32] <stonie_> ok ye.. thats what i found out too
[15:33] <stonie_> mediainfo just gives the header info
[15:33] <JEEB> not even that :P
[15:34] <JEEB> also I'm going to take this moment of you using the word 'alot' and paste this for you http://hyperboleandahalf.blogspot.com/2010/04/alot-is-better-than-you-at-everything.html
[15:34] <rcombs> ^this
[15:35] <stonie_> haha nice .. ty
[15:35] <stonie_> ALOT
[15:35] <JEEB> anyways, if libx264 is saying it's using interlaced encoding in its line that it blurfs out, then I'm just going to note that you're getting interlaced output just fine
[15:35] <JEEB> and it's something else stumbling your beloved piece of crap in plastic :)
[17:01] <stonie_> what would you use indead of mediainfo?
[17:01] <stonie_> to get as detailed info as possible
[17:03] <DaFrameSkipper> anyone have any what "-skip_threshold" "-skip_factor" "-skip_exp" does and how do use them?
[17:04] <ubitux> stonie_: ffprobe
[17:04] <JEEB> DaFrameSkipper, http://ffmpeg.org/ffmpeg-all.html at least has some info on them :P Seem to be libvpx-specific options
[17:05] <JEEB> or not
[17:05] <DaFrameSkipper> its really vague
[17:05] <JEEB> there's two skip_thresholds on that page
[17:05] <JEEB> well, unfortunately then you'll have to go and grep the source code :)
[17:06] <JEEB> you are not going to get more info on how those can be used otherwise
[17:07] <jeeger> Hello! I'm currently recoding old mst3k episodes from avi to x264. Does the preset actually matter? Is the quality difference from -preset slow to superfast noticeable? Any other recommendations?
[17:08] <JEEB> How much difference the difference presets make depends on a lot of stuff, but generally such a large difference does affect the result quite a bit. My recommendation is the one I use generally: Use the slowest preset that is fast enough for you
[17:10] <stonie_> ty
[17:22] <jeeger> JEEB Thanks.
[18:35] <viric> mm I fail to use "-ss" with the concat filter.
[18:36] <viric> if I put it prev to -i, it is unprecise.
[18:36] <viric> If after -i, it's applied to the output before encoding.
[19:42] <Bombo> i got a stream "0:0 video 0:1 audio1 0:2 audio2 0:3 subs1 0:4 subs2" how do i select everything but subs2? i tried "-c:v libx264 ... -map 0:0 -c:a libmp3lame ... -map 0:1 -c:a libmp3lame ... -map 0:2 -c:s none -c:s copy -map 0:4"
[19:42] <Bombo> tried -sn for 0:3 too
[19:42] <Bombo> complains about missing parameters for 0:3
[19:44] <Bombo> it works when i just copy 0:3 too "-c:s copy -map 0:3" instead of "-c:s none", but it contains the subs2 then
[19:54] <Bombo> ah nevermind. got it ;)
[20:33] <DaFrameSkipper> is there a way to skip frames automatically in ffmpeg?
[20:36] <DaFrameSkipper> for example when im livstreaming?
[20:38] <svm_invictvs> Heya
[20:39] <svm_invictvs> I'm looking for the source to a specific version of libavformat
[21:02] <alesan> hi, how can I pass a specific option to libx264, for example the --keyint infinite
[21:03] <JEEBsv> -g was the ffmpeg keyint-like setting
[21:03] <JEEBsv> now find out which value ~= infinite
[21:04] <alesan> hi JEEBsv :)
[21:07] <DaFrameSkipper> JEEBsv, any ideas how to dynamically drop frames for livestreaming?
[22:36] <svm_invictvs> So I'mg etting this in a stack trace.
[22:36] <svm_invictvs> http://mysticpaste.com/view/NlgUIBKYSs?2
[22:37] <svm_invictvs> I've read some people reporting crash bugs for those fucntions.  Is tehre a known workaround for this?
[23:42] <slorbast> Hey alls, I was wondering if it's possible to take a video and display text on the side based on timestamp information from say an IRC log?
[23:43] <slorbast> apo_, any luck with twitch streaming?
[23:43] <sacarasc> I think it is, slorbast, but I don't know how.
[23:44] <slorbast> sacarasc, thanks, I'm perusing the webs for info but hoped someone had experience in here while I looked :)
[23:45] <apo_> slorbast: nope
[23:46] <apo_> slorbast: I can easily upload over 100 KB/s to my own server, but twitch won't even do 66 KB/s.
[23:46] <slorbast> apo_, and still having issues getting to AMS server?
[23:48] <apo_> slorbast: not with ffmpeg 2
[23:48] <apo_> but it didn't really change anything
[00:00] --- Sat Dec 14 2013


More information about the Ffmpeg-devel-irc mailing list