[Ffmpeg-devel-irc] ffmpeg.log.20160315

[burek]

[00:00:14 CET] <Dan0maN> my first example had a %05, this example has %03
[00:00:17 CET] <llogan> try the glob pattern. 3rd example that furq linked. hopefully it will order them correctly.
[00:01:23 CET] <Dan0maN> k.  ty
[00:01:26 CET] <llogan> then add -pix_fmt yuv420p as an output option; otherwise it will try to preserve the chroma subsampling, but most non-FFmpeg based players only like YUV 4:2:0
[00:01:40 CET] <Dan0maN> is that a standard type of pattern descriptor?
[00:01:48 CET] <llogan> and use -framerate instead of -r for image file demuxer
[00:01:55 CET] <Dan0maN> k
[00:02:16 CET] <furq> [mjpeg @ 0x5649cc0] Changeing bps to 8
[00:02:20 CET] <furq> where do i report this terrible bug
[00:06:51 CET] <drv> the "Changeing" part?
[00:06:58 CET] <furq> yes
[00:07:33 CET] <llogan> was just about ti fix it, but some crazy old lady entered my office.
[00:07:40 CET] <llogan> go away
[00:10:19 CET] <llogan> ok. gave her coworkers number... must have been fixed before. not in git master.
[00:11:25 CET] <daslicht> not ffmpeg related but maybe anyone like to help
[00:11:31 CET] -daslicht:#ffmpeg- http://stackoverflow.com/questions/35999691/is-there-any-icecast-cli-source-client-which-can-play-cue-sheets-and-handle-hand
[00:11:31 CET] -daslicht:#ffmpeg- http://stackoverflow.com/questions/35999718/is-there-any-icecast-cli-source-client-which-offers-the-current-playback-positio
[00:12:52 CET] <llogan> yeah, 3885ef0
[00:13:45 CET] <Dan0maN> worked great.  ty llogan.
[00:14:52 CET] <llogan> Dan0maN: also add "-movflags +faststart" if your viewers will watch via progressive download
[00:15:24 CET] <llogan> daslicht: those don't look like programming questions. SO is for programming questions.
[00:15:41 CET] <daslicht> which is teh correct place ?
[00:16:12 CET] <daslicht> stack exchange ?
[00:16:16 CET] <llogan> within SE probably Super User or video.stackexchange.com
[00:16:44 CET] <llogan> you can get the questions migrated instead of making crossposts which is annoying
[00:17:00 CET] <daslicht> i move them
[00:17:08 CET] <daslicht> video ?
[00:17:16 CET] <daslicht> adio
[00:17:17 CET] <daslicht> audio
[00:18:06 CET] <daslicht> hm theer is no audio
[00:20:41 CET] <daslicht> no idea
[00:20:57 CET] <daslicht> I leave them there until we found teh correct place to migrate
[00:58:34 CET] <cbsrobot> daslicht: tried using python ?
[01:02:45 CET] <daslicht> nope
[01:02:55 CET] <daslicht> I have very little expericne with python
[02:49:46 CET] <nadermx> When running ffmpeg if I use the -threads option should I set it to number of threads on server (4) or 0?
[02:51:39 CET] <c_14> 0 is the default
[02:52:04 CET] <nadermx> but does 0 use more than one thread if its available? Or does 0 use just 1?
[02:52:22 CET] <furq> it depends on the encoder
[02:52:27 CET] <c_14> and decoder
[02:52:36 CET] <nadermx> libmp3lame specifically
[02:52:42 CET] <furq> i don't think you can multithread lame
[02:52:44 CET] <c_14> If they support threading without additional options it should do it automatically.
[02:53:05 CET] <furq> if you're encoding multiple files you can do them in parallel with something like xargs or parallel
[02:53:59 CET] <nadermx> I'm calling ffmpeg as a subprocess in a golang program
[02:55:08 CET] <nadermx> Just trying to figure out how to make it go faster I suppose for longer songs
[02:55:33 CET] <c_14> Get a faster cpu
[02:56:02 CET] <c_14> Or try cutting the file into parts and encoding them in parallel and then concatting them
[03:03:28 CET] <nadermx> ok thank you
[03:46:46 CET] <wellen> hi
[03:47:46 CET] <wellen> can i check this result... ffmpeg -i test720pav.mp4 -c:a copy test192ka.m4a
[03:48:48 CET] <wellen> is that good for just getting the audio from the input file eg quickest method?
[03:52:56 CET] <c_14> add -vn and -sn
[03:53:05 CET] <c_14> or -map 0:a:0
[03:53:11 CET] <c_14> Assuming you only want 1 audio stream
[03:53:12 CET] <johnnny22-afk> In ffserver, what happens to a Feed's File once it reaches the FileMaxSize  ?
[03:53:16 CET] <c_14> or -map 0:a if you want all of them
[03:53:42 CET] <c_14> johnnny22-afk: I assume it's a ring buffer, but I don't know
[03:53:45 CET] <johnnny22-afk> I'm also pondering about how that Truncate option works.. it's not too clear.
[03:54:12 CET] <wellen> the input file has audio adn video but only 1 of each and i am just trying to get it to output the ony audio it has as an m4a
[03:54:56 CET] <c_14> Then either -vn or -map 0:a. In this case they'll do the same thing
[03:55:33 CET] <wellen> based on my first attempt is that why it is encoding the video (because i need -vn ?
[03:55:39 CET] <c_14> yes
[03:56:03 CET] <wellen> thanks c_14 - i used to be a lot sharper than this...then i got ill doh!
[04:01:25 CET] <wellen> im still pondering a ffmpeg gui btw
[04:03:47 CET] <johnnny22-afk> c_14: a ring buffer on disk ?
[04:04:34 CET] <wellen> im done, night all thanks again c_14
[05:50:49 CET] <jookiyaya> what is better audio codec ?  vorbis or opus?
[06:17:48 CET] <sling_> Hi
[06:18:12 CET] <sling_> i am running the ffmpeg shell command using the java process builder
[06:18:32 CET] <sling_> when ffmpeg fails with error how do i get the error code
[06:18:43 CET] <sling_> bcaz in the log i dont find the error code
[06:20:02 CET] <sling_> 	ProcessBuilder pb = new ProcessBuilder(command.split(" ")).inheritIO();
[06:20:17 CET] <sling_> final Process p; 		try { 			p = pb.start();  			TransProgressThread transProgressThread = new TransProgressThread(transcodingContext); 			transProgressThread.start();  			p.waitFor(); }
[06:24:09 CET] <sling_> [aac @ 0000000002b71c80] The encoder 'aac' is experimental but experimental codecs are not enabled, add '-strict -2' if you want to use it. [aac @ 0000000002b71c80] Alternatively use the non experimental encoder 'libvo_aacenc'.
[06:24:28 CET] <sling_> how do i can get that there is an error happened using above error
[07:13:35 CET] <relaxed> sling_: if ffmpeg returns anything other than zero, there was an error.
[07:14:51 CET] <johnnny22-afk> can ffmpeg skip holes in a sparse file ?
[07:53:00 CET] <johnnny22-afk> if i wanted ffmpeg to simply append to a output file that already exists, should I output to stdout and >> to the file ?
[07:53:16 CET] <sling_> @relaxed THeN I NEED TO PARSE THE LOGS TO GET THE ERROR STATEMENT?
[07:53:47 CET] <sling_> do you have a list of error statements?
[07:53:57 CET] <sling_> based on that i can handle errors differently
[07:59:48 CET] <relaxed> sling_: you'll have to check the source
[08:16:23 CET] <johnnny22-afk> lets say ffmpeg fails to read the input stream, is there any tricks to resume consumption of the input when reinvoking ffmpeg again after ?
[08:16:53 CET] <johnnny22-afk> to resume at the same location
[09:54:57 CET] <johnnny22-afk> what would be the best way to throttle ffmpeg externally ?
[09:55:27 CET] <johnnny22-afk> ideally throttle it's bandwidth usage.
[09:56:13 CET] <JEEB> depends on what you want to do
[09:56:26 CET] <johnnny22-afk> I was thinking of using iptables possibly
[09:57:35 CET] <johnnny22-afk> or possibly achieve limiting it's output, thus impacting it's input !? I'm guessing.
[10:00:02 CET] <JEEB> output limiting is possible with most sane video encoders with maxrate and bufsize
[10:04:05 CET] <johnnny22-afk> right, right, but i'm using ffmpeg to consume an online stream and writing it to disk with -c copy.  I might even have two instance of that running. Ideally, I'd want to be able to set the maximum bandwidth that whole group of instances of ffmpeg are using as a group.
[14:46:12 CET] <zuloyd> hi
[14:47:39 CET] <zuloyd> Is it possible with ffserver to record a live stream, play it back via HTTP and be able to seek inside it?
[14:48:06 CET] <zuloyd> I managed to get it to record an RTSP stream, and this RTSP stream is forwarded to a WEBM stream accessible with the built-in HTTP server
[14:48:10 CET] <zuloyd> however no seeking is possible
[14:48:23 CET] <zuloyd> I'd like to be able to jump back in time
[14:54:30 CET] <flux> so it say on the webs that -crf drops video quality for high-action scenes, because it's difficult to see what's happening so clearly anyway
[14:54:47 CET] <flux> however, my goal would be to use this for security video footage, where this might be counter-productive
[14:55:23 CET] <flux> is my only recourse to use the x264 library interface directly to adjust the quality manually? or perhaps use the same quality value instead of trying to increase it for high-motino scenes?
[14:56:10 CET] <J_Darnley> You could try constant quantiser as an alternative
[14:57:39 CET] <flux> right, I suppose that's the only "pre made" solution?
[14:58:23 CET] <J_Darnley> If you want decent frame-by-frame quality then yes (or just go for high quality)
[14:58:58 CET] <flux> though for pictures that don't have a lot of going on (ie. only camera noise), I would prefer not to waste space :)
[14:59:34 CET] <J_Darnley> const. quant. isn't going to waste much in those areas
[15:00:01 CET] <J_Darnley> inter-frame prediction will reduce the cost of static scenes and areas towards nothing
[15:00:05 CET] <flux> is the bitrate adjustment algorithms smart enough to understand it is compressing noise?
[15:01:09 CET] <J_Darnley> x264 does have a perceptual feature to reduce the cost of high noise (usually film grain or dithering)
[15:01:19 CET] <flux> nice
[16:43:58 CET] <eduarte> hello
[16:44:18 CET] <eduarte> am i alone here ?
[16:44:34 CET] <eduarte> can i make a question ?
[16:48:00 CET] <J_Darnley> Yes.  Just ask your question.
[16:49:34 CET] <eduarte> I am trying to cut a video with ffmpeg, cutting it down to 30 milliseconds, tho the result always have more than 32 centiseconds, and always different sizes
[16:49:40 CET] <eduarte> is that influenced by the video quallity ? how can that be happening ?
[16:52:26 CET] <kepstin> eduarte: are you re-encoding or using copy mode?
[16:53:22 CET] <kepstin> keep in mind also that 30ms is (depending on the frame rate) only 1-2 frames, or sometimes not even that
[16:55:46 CET] <eduarte> i mean -i videofile
[16:57:33 CET] <eduarte> http://pastebin.com/LNMSErJ3
[16:58:08 CET] <J_Darnley> What's this about fractions of a second?
[16:58:14 CET] <kepstin> eduarte: with that command, you're asking for 30 seconds of video...
[16:58:16 CET] <J_Darnley> That is 30 seconds
[16:58:22 CET] <eduarte> in the case it is not even 1-2 frames, it wont work ?
[16:58:30 CET] <eduarte> oh rly ?
[16:58:39 CET] <kepstin> yes, it's HH:MM:SS
[16:58:42 CET] <J_Darnley> HH:MM:SS
[16:58:46 CET] <eduarte> and can i add more : ?
[16:58:50 CET] <eduarte> hh:mm:ss:cs:ms ?
[16:58:55 CET] <kepstin> no, but you can use fractional seconds with a .
[16:59:02 CET] <kepstin> so -t 0.030 is 30ms
[16:59:09 CET] <eduarte> oh ok
[16:59:14 CET] <eduarte> thank you :D
[17:01:10 CET] <eduarte> another question is:
[17:02:07 CET] <eduarte> imagine i join 1000 videos with 30 ms each with ffmpeg, does the quantity will affect the video performance? create any glitch or somth ?
[17:03:04 CET] <kepstin> eduarte: you're probably going to have rounding errors that add up, due to the fact that 30ms is close to, but not quite, the same length as a frame.
[17:05:04 CET] <eduarte> hm
[17:05:42 CET] <eduarte> is there someway to be sure the minimum i can cut from a video with ffmpeg ?
[17:06:03 CET] <J_Darnley> No
[17:06:05 CET] <J_Darnley> Depends on the video format
[17:06:12 CET] <eduarte> i see
[17:06:13 CET] <J_Darnley> and the framerate
[17:06:19 CET] <kepstin> eduarte: depends mostly on your desired output framerate, I think
[17:06:34 CET] <J_Darnley> You can't cut a P or B frame out of the middle of a video
[17:06:54 CET] <kepstin> J_Darnley: looks like eduarte is doing a re-encode, so grabbing an arbitrary frame is possible
[17:06:58 CET] <eduarte> than again, i would still have a problem if i want to cut like 1 and half frames right ?
[17:07:14 CET] <J_Darnley> What?  The use the image2 muxer and save every frame with ease.
[17:07:27 CET] <kepstin> eduarte: i'd recommend that you pick a framerate for the final result, then work based on frames rather than time
[17:07:42 CET] <eduarte> hm i see
[17:07:47 CET] <furq> eduarte: use -frames:v if you want a particular number of frames
[17:08:58 CET] <eduarte> ok I will try that
[17:10:45 CET] <eduarte> just to give you context, what im trying to do is, 1- i have videos of me playing each note of an instrument ,,, 2-i am interpreting midi files and generating with ffmpeg a final video cutting and concatenating the videos of each note, the reason of those questions is because i need to get the timings right :), and i get each notes timing in milliseconds
[17:11:44 CET] <J_Darnley> I hope you're using a 1000fps camera then
[17:11:44 CET] <eduarte> the final goal is to have an output of you playing a music given a midi file, kinda weird but soudns fun :)
[17:12:09 CET] <eduarte> not at all :D
[17:12:37 CET] <eduarte> i will see the max precision i can get from the camera in time and reduce the minimum note time to that
[17:12:42 CET] <J_Darnley> Then good luck with 40/33/16 ms precision.
[17:13:21 CET] <kepstin> yeah, you're probably just gonna have to make your conversion program drop frames to try to keep the video approximately in sync
[17:14:43 CET] <eduarte> thanks
[17:14:46 CET] <osense> hi
[17:15:08 CET] <osense> is there some recent issue with the 'copy' codec?
[17:15:16 CET] <osense> I'm getting 'Protocol not on whitelist 'none'!'
[17:15:47 CET] <osense> when trying to concatenate videos with the demuxer
[17:16:04 CET] <kepstin> osense: that's complete unrelated to the copy codec, but it rather a demuxer/format issue
[17:17:36 CET] <osense> hmm, any idea what could be wrong? is it illegal to try to concat a single video?
[17:17:38 CET] <osense> http://pastie.org/10760945
[17:19:10 CET] <kepstin> osense: you're hitting a security protection. by default, ffmpeg only allows playlists (and concat is a playlist, basically) to read local files if the playlist is itself a local file
[17:19:22 CET] <kepstin> but you're sending the playlist on stdin, which isn't a file
[17:20:41 CET] <kepstin> to override that, you can use the "-safe 0" input option, which disables the security check
[17:21:33 CET] <kepstin> (or at least, I think that's the problem? could be wrong)
[17:23:11 CET] <osense> hmm, nothing's changed
[17:23:26 CET] <osense> (input option means after -i right?)
[17:23:32 CET] <kepstin> no, before
[17:23:46 CET] <osense> still nothing
[17:23:51 CET] <kepstin> hmm.
[17:25:32 CET] <kepstin> oh, the -safe option only prevents access to e.g. absolute paths. it's probably not needed here, actually
[17:25:41 CET] <kepstin> I have no idea how to override that protocol issue
[17:25:54 CET] <kepstin> (aside from writing the playlist to a file, which would probably work)
[17:29:49 CET] <kepstin> ah, I was looking at the help for the wrong ffmpeg version
[17:29:58 CET] <kepstin> try adding "-protocol_whitelist file" as an input option
[17:43:56 CET] <osense> kepstin: now I get Protocol not on whitelist 'file'!
[17:44:52 CET] <osense> bbl
[17:53:23 CET] <mr_pinc> Hello  I am trying to follow this example - https://trac.ffmpeg.org/wiki/Scaling%20%28resizing%29%20with%20ffmpeg#no1 to convert a 4:3 image into a resized 16x9 image with a matined aspect ratio.  Thing is even though I specify 16x9 dimensions, the resulting image is still 4:3 - so if my input image is 1024x768 and I run the following scale="'if(gt(a,16/9),403,-1)':'if(gt(a,16/9),-1,2227)'"
[17:53:23 CET] <mr_pinc> output_320x240_boxed.png - the result is an image that is 303x227 not 403x227
[18:06:45 CET] <mr_pinc> So the image does maintain the aspect ratio but it does not fill in the desired 16x9 (403x227)
[18:27:53 CET] <J_Darnley> You need to pad if you want black bars
[18:28:36 CET] <mr_pinc> Thank you - I just got it working using this suggestion - http://stackoverflow.com/questions/8133242/ffmpeg-resize-down-larger-video-to-fit-desired-size-and-add-padding
[18:39:11 CET] <cimbakahn> Hello!
[18:39:20 CET] <cimbakahn> Does anyone know where there is a list of all the formats that mpv plays?
[18:39:39 CET] <J_Darnley> ffmpeg -formats?
[18:41:10 CET] <cimbakahn> You know like does it play .avi, .ogg, .mp4, .wax etc..
[18:41:59 CET] <cimbakahn> I know it plays ffmpeg formats, but what are those?
[18:44:02 CET] <furq> cimbakahn: http://sprunge.us/CSBW
[18:44:28 CET] <furq> note the first line
[18:45:16 CET] <cimbakahn> Thank you furq!  You are so sweet!
[18:45:36 CET] <furq> i guess `ffmpeg -formats ^\ D` is a better answer
[18:45:39 CET] <furq> er
[18:45:45 CET] <furq> i guess `ffmpeg -formats | grep ^\ D` is a better answer
[18:46:48 CET] <cimbakahn> HAHAHA!
[18:50:02 CET] <J_Darnley> I doubt there is an E without a D
[18:51:15 CET] <J_Darnley> oh well there are some
[18:52:14 CET] <J_Darnley> oh now most of those come from slight naming differences
[19:15:47 CET] <note> Hello there! I wish for ffmpeg transcoded videos from the cameras of Android and iPhones to be viewable in each other's devices, but so far whatever arguments I use to ffmpeg, the resulting videos are only viewable on the same phone/platform the video was captured from. Might you know what is the problem?
[19:17:06 CET] <note> I should add that I would like them to be viewable in the HTML5 WebView component of each other's platform.
[19:24:00 CET] <pzich> note: I unfortunately don't know too much about what might be causing that, but whoever does is going to want to see your commands and their full output, i.e.
[19:27:00 CET] <note> Ok thanks, the commands ( http://pastebin.com/cQgNH1kz ) have always succeeded. It's just that the resulting videos don't work on the opposite platforms.
[21:23:31 CET] <jookiyaya> what is better audio codec ?  vorbis or opus?
[21:25:30 CET] <kepstin> opus.
[21:27:57 CET] <jookiyaya> why are ffmpeg people so bias to opus
[21:28:15 CET] <furq> why do you keep asking the exact same questions and then complaining about the answers
[21:29:00 CET] <jookiyaya> i am not complaining
[21:29:18 CET] <J_Darnley> Is it not true that opus is better?
[21:29:31 CET] <kepstin> based on listening tests that i've seen, opus provides better quality at the same bitrate as vorbis. therefore it's a better audio codec.
[21:30:00 CET] <jookiyaya> j_darnley  it seems like people here are very pro opus, which is not the same case for different #channel
[21:30:59 CET] <furq> why are you listening to what people in #handbrake have to say
[21:31:10 CET] <bencoh> haha
[21:31:13 CET] <jookiyaya> i like to listen to everybody
[21:31:22 CET] <jookiyaya> bencoh what is funny?
[21:32:09 CET] <bencoh> the idea implied by furq's question :)
[21:32:40 CET] <jookiyaya> furq people in #handbrake seem to respect people in #ffmpeg
[21:32:58 CET] <furq> that's nice
[21:36:40 CET] <jookiyaya> kepstin what about vorbis vs aac
[21:37:10 CET] <furq> 02:53:25 ( furq) jookiyaya: http://listening-test.coresv.net/s/scores_by_tracks_en.png
[21:38:00 CET] <jookiyaya> there is no  ffmpeg-aac on that list
[21:38:14 CET] <furq> i wonder if that guy who disagrees has been reading those garbage "soundexpert" listening tests with artifact amplification
[21:38:17 CET] <jookiyaya> i don't think ffmpeg support qaac
[21:38:37 CET] <kepstin> jookiyaya: qaac is quicktime's encoder, not open source.
[21:38:49 CET] <furq> nobody can say with any authority how good ffaac is because nobody has done a reasonably wide-scale listening test
[21:39:04 CET] <furq> if i had to guess i'd say it's probably a bit worse than vorbis but not enough to be noticeable at 128k
[21:39:08 CET] <kepstin> jookiyaya: compared to fdk_aac, the (arguably) best aac encode ffmpeg supports, I think vorbis is slightly worse?
[21:39:18 CET] <kepstin> hard to tell for most people yeah
[21:39:33 CET] <furq> fdk was slightly worse than qaac in 2011
[21:39:43 CET] <furq> i've not seen any more recent tests
[21:40:08 CET] <kepstin> I mean, if you're running all of these codecs at well above 128kbit (stereo), you're probably not gonna be able to tell them apart
[21:40:31 CET] <furq> i don't see any reason to encode audio unless you need player compatibility
[21:40:35 CET] <kepstin> (you'd probably be able to notice vo-aacenc tho, even at those bitrates. Don't use v-aacenc)
[21:40:39 CET] <furq> in which case your choice is determined by the player
[21:41:11 CET] <jookiyaya> is v-aacenc really that bad?
[21:41:19 CET] <bencoh> (iirc ffmpeg dropped support anyway :)
[21:41:21 CET] <furq> vo is even worse than faac
[21:41:26 CET] <furq> and yeah vo isn't even in 3.0 anyway
[21:41:34 CET] <jookiyaya> is vo worse than mp3?
[21:41:34 CET] <bencoh> 3.x already? pfew
[21:41:55 CET] <bencoh> oh right, it's been released
[21:42:12 CET] <furq> someone should tell freebsd that it's been released
[21:42:32 CET] <jookiyaya> people these days have smartphones now  and they all have players like VLC<  so why is  player compatiblity an issue?
[21:42:34 CET] <TD-Linux> this is the most recent listening test with ffmpeg's aac encoder https://hydrogenaud.io/index.php/topic,109716.0.html
[21:42:45 CET] <kepstin> but yeah, if your source is e.g. a dvd, there's probably a 192-256kbps stereo ac3 track which you'd just want to passthrough.
[21:43:12 CET] <furq> TD-Linux: that's from one guy isn't it
[21:43:22 CET] <kepstin> that's just one listener, and an older version of the encoder, yeah
[21:43:29 CET] <bencoh> jookiyaya: because people also have smartTVs/settopboxes/phones with hardware/DSP decoders
[21:43:34 CET] <furq> and yeah that's a pre-3.0 release
[21:43:44 CET] <TD-Linux> furq, yes, and it's out of date too.
[21:43:57 CET] <bencoh> and in some cases you just have to use what your devices can play
[21:44:23 CET] <furq> it still does better than faac
[21:44:31 CET] <jookiyaya> bencoh  you cannot install  player like vLC on  smarttv/settopbox ?
[21:44:58 CET] <furq> jookiyaya: there's no opus/vorbis support in mp4 and no aac/mp3 support in webm
[21:45:04 CET] <bencoh> even if you could, you often cant decode in software anyway
[21:45:09 CET] <BtbN> VLC decodes in software, you don't realy want that on mobile devices.
[21:45:21 CET] <furq> apparently opus is coming to mp4 but it's not going to be much use to anyone who hasn't got an up-to-date player
[21:45:23 CET] <BtbN> And on SmartTVs the CPU might not be up to the task to decode anything
[21:45:31 CET] <TD-Linux> in 2016 audio is nearly always decoded in software.
[21:45:33 CET] <jookiyaya> btbn  why? what's wrong with software decodes?
[21:45:47 CET] <BtbN> They use way more energy than a DSP doing it for free.
[21:45:48 CET] <bencoh> BtbN: software audio decoding on mobile isn't that bad actually compared to some hw pipeline
[21:45:57 CET] <TD-Linux> DSPs don't decode for free
[21:46:04 CET] <bencoh> but still :)
[21:46:10 CET] <BtbN> More than a CPU having to stay awake
[21:46:33 CET] <BtbN> *Less than
[21:46:35 CET] <TD-Linux> BtbN, reminds me I need to redo my measurement of this
[21:46:38 CET] <bencoh> BtbN: my tests on n900 says otherwise for sw vs dsp cpu decoding
[21:46:42 CET] <bencoh> -s
[21:46:50 CET] <bencoh> s/cpu/mp3/ wetf
[21:47:04 CET] <bencoh> alright, looks like I cant type properly tonight
[21:49:23 CET] <jookiyaya> does ffmpeg support  converting 7.1 audio to  5.1 ?
[21:51:49 CET] <jookiyaya> according to this:   http://oi57.tinypic.com/15g2cxz.jpg    Neroaac is better than opus
[21:52:07 CET] <bencoh> neroaac, seriously?
[21:52:16 CET] <furq> firstly, no it isn't
[21:52:22 CET] <furq> secondly, that test is by one guy
[21:52:26 CET] <furq> but thirdly, and most importantly, no it isn't
[21:52:35 CET] <jookiyaya> furq  nero is using 96k
[21:53:13 CET] <jookiyaya> opus is using 102
[21:53:36 CET] <furq> well i guess you'd better use nero then
[21:53:37 CET] <furq> hf
[21:54:10 CET] <jookiyaya> hf?
[21:55:35 CET] <pzich> it means "have fun" :)
[21:57:20 CET] <jookiyaya> oh
[21:57:26 CET] <kepstin> jookiyaya: look closer - the legend of that graph says that the average over multiple albums of opus was 94k, nero was 99k. Basically what happened is that opus makes better use of vbr to give extra bitrate to hard samples and save it on easy ones.
[21:57:57 CET] <jookiyaya> oh
[21:58:06 CET] <jookiyaya> is Nero better than aac-fdk ?
[21:58:34 CET] <kepstin> jookiyaya: I haven't seen any well-done comparisons between the two
[21:58:53 CET] <kepstin> i suspect they're fairly close.
[21:59:27 CET] <furq> http://listening-tests.hydrogenaud.io/igorc/aac-96-a/nonblocked_means_all.png
[22:00:16 CET] <jookiyaya> what are "ct" "tvbr"
[22:00:29 CET] <furq> cvbr and tvbr are qaac
[22:00:34 CET] <jookiyaya> oh
[22:01:06 CET] <furq> CT is another proprietary encoder
[22:01:25 CET] <jookiyaya> and it stands for?
[22:01:43 CET] <jookiyaya> never heard of it
[22:02:14 CET] <furq> coding technologies
[22:02:27 CET] <furq> it was the winamp aac encoder before they ditched it in favour of fraunhofer
[22:02:36 CET] <furq> fdk is the open-source version of fraunhofer aac
[22:03:25 CET] <jookiyaya> i love how fdk is open source yet it's not included in ffmpeg
[22:03:42 CET] <furq> it has an incompatible license
[22:06:48 CET] <TD-Linux> the license does not meet the OSI definition of open source
[22:07:42 CET] <jookiyaya> i tried fdk-aac.exe  , it doesn't even transcode
[22:10:21 CET] <jookiyaya> does ffmpeg support  converting 7.1 audio to  5.1 ?
[22:16:22 CET] <kepstin> jookiyaya: of course. The question is whether it has a predefined matrix for it, or if you need to use the pan filter for that. You could just drop the extra channels of course...
[22:17:03 CET] <jookiyaya> kepstin what does default option do?
[22:18:15 CET] <jookiyaya> i just found a bug for opus. it's misreporting its bitrate
[22:18:27 CET] <jookiyaya> or is that ffmpeg bug
[22:24:34 CET] <kepstin> jookiyaya: not sure what you mean
[22:25:08 CET] <jookiyaya> kepstin what is the difference betwen  predefined matrix  and  pan filter
[22:25:35 CET] <kepstin> jookiyaya: with pan filter, you can decide exactly how to downmix the channels manually.
[22:25:44 CET] <jookiyaya> i see
[22:26:33 CET] <kepstin> if you're using unconstrained vbr mode in opus (the default), the bitrate parameter is basically a quality level, calibrated such that if you encode a large number of files of varying types of audio, the average bitrate will be what you requested.
[22:26:46 CET] <kepstin> average over all the files
[22:27:11 CET] <jookiyaya> kepstin i understand that but  it's reporting as 3072 kbps
[22:27:21 CET] <jookiyaya> when it's close to 128kbps
[22:28:07 CET] <kepstin> are you encoding audio + video? the bitrate ffmpeg prints in the status line is the combination of both.
[22:28:21 CET] <jookiyaya> it's misreporting when it's inside mkv file
[22:28:50 CET] <jookiyaya> audio only, it's fine
[22:29:22 CET] <kepstin> jookiyaya: are you going to paste some command lines and example output? I have no idea what you are talking about here.
[22:30:34 CET] <jookiyaya> http://www.imagebam.com/image/9db148471937577
[22:31:09 CET] <furq> that's mpc-hc's fault
[22:31:09 CET] <jookiyaya> it should be 640/128/128
[22:31:22 CET] <furq> try looking at the mediainfo tab instead
[22:31:31 CET] <jookiyaya> mpc-hc use mediainfo
[22:31:39 CET] <furq> it does in the mediainfo tab
[22:31:43 CET] Action: llogan loves images of text
[22:31:56 CET] <furq> the main tab is from directshow or some other inaccurate bullshit
[22:32:17 CET] <jookiyaya> it doesn't report the bitrate from mendiainfo tab
[22:32:20 CET] <kepstin> yeah, that's not ffmpeg, what does ffprobe say the stream bitrates are? :)
[22:32:34 CET] <furq> that's because mkv doesn't store bitrates in the metadata
[22:32:44 CET] <furq> you can only get track bitrates from mkv if there's 0 or 1 vbr tracks
[22:32:53 CET] <furq> or by probing the entire file
[22:33:27 CET] <jookiyaya> you are right mediainfo does display correctly
[22:34:58 CET] <jookiyaya> what is wrong with this tab
[22:35:11 CET] <jookiyaya> it's misreporting it completely
[22:35:14 CET] <furq> i guess it's there for legacy reasons
[22:35:20 CET] <furq> it should really just use mediainfo for that
[22:35:49 CET] <furq> the mediainfo tab is there specifically because the main tab is always either missing information or completely wrong
[22:36:00 CET] <jookiyaya> latestversion of mediainfo finally fixed not reporting bitrate with mkv too
[22:36:05 CET] <furq> oh that's nice
[22:36:18 CET] <jookiyaya> it used to not report it for some reason
[22:36:29 CET] <furq> it didn't report it for the reason i just said
[22:36:37 CET] <furq> you need to scan the whole file if there are multiple vbr tracks
[22:37:23 CET] <jookiyaya> it (always) works fine with mp4 though
[22:37:36 CET] <furq> mp4 stores stream bitrates in the metadata, mkv doesn't
[22:37:38 CET] <kepstin> jookiyaya: that's because mp4 is a different format that works in a different way
[22:38:05 CET] <jookiyaya> problem with mp4 is , i cannot preview the video while it's encoding, with mkv, i can
[22:38:17 CET] <jookiyaya> i hate that
[22:44:43 CET] Action: kepstin keeps getting annoyed that mkv doesn't preserve video timebase
[22:49:07 CET] <jookiyaya> kepstin what is that?
[23:26:49 CET] <NeedHelp> Hello is there anyone available to answer a couple questions?
[23:27:28 CET] <Guest33272> Is there anyone around who can answer a couple questions for me?
[23:27:38 CET] <kepstin> generally works better on irc to ask your question, then hang around to see if anyone answers it.
[23:27:57 CET] <Guest33272> Oh sorry, I'm not used to irc
[23:28:00 CET] <Guest33272> I'll do that then
[23:28:26 CET] <Guest33272> so I have a .flv file that I'm trying to put into lightworks, but the file extension isnt supported
[23:28:38 CET] <Guest33272> can ffmpeg help me out with that?
[23:28:50 CET] <Guest33272> I'm trying to convert the .flv file into a readable/importable format
[23:29:04 CET] <kepstin> yes, ffmpeg can do help with that
[23:29:08 CET] <J_Darnley> Just rename it if it doesn't like the filename
[23:29:10 CET] Action: J_Darnley runs
[23:29:21 CET] <pzich> my little brother used to think that would actually work :-/
[23:29:22 CET] Action: kepstin doesn't know offhand what containers and codecs lightworks supports
[23:30:12 CET] <kepstin> it could be that something as simple as 'ffmpeg -i whatever.flv -c copy whatever.mp4' would be enough, depending on the codecs used in the flv file.
[23:30:33 CET] <Guest33272> Yeah, see, It's a video I recorded with OBS
[23:31:47 CET] <kepstin> probably h264 video and aac or mp3 audio then. give the simple ffmpeg copy command a try.
[23:31:56 CET] <furq> http://www.lwks.com/index.php?option=com_content&view=article&id=100&Itemid=211
[23:32:34 CET] <furq> it looks like you'll have to either transcode it or use an editor which isn't openly hostile to non-professional formats
[23:33:03 CET] <Guest33272> hm, I'll try what kepstin suggested
[23:33:09 CET] <kepstin> hmm? that list includes h.264
[23:33:15 CET] <Guest33272> it's my first time using this thing
[23:33:22 CET] <kepstin> does say avchd tho
[23:33:28 CET] <kepstin> might need a different container
[23:33:41 CET] <Guest33272> argh this is pretty complicated.
[23:33:54 CET] <Guest33272> stupid obs...
[23:33:56 CET] <kepstin> should handle h264 in mpeg-ts at the very least, i think
[23:34:26 CET] <kepstin> really, lightworks should just use ffmpeg to read input files, then this problem wouldn't happen ;)
[23:34:27 CET] <Guest33272> stupid question, how do I execute/run ffmpeg?
[23:34:36 CET] <Guest33272> just extracted it to a sep folder but i dont see an executable file
[23:34:47 CET] <kepstin> Guest33272: what did you download?
[23:34:57 CET] <furq> if you got the zeranoe build then ffmpeg.exe is in bin/
[23:34:58 CET] <kepstin> I assume you're on windows, so you'd have to grab a binary
[23:35:12 CET] <Guest33272> yeah i'm on windows
[23:35:20 CET] <kepstin> https://ffmpeg.zeranoe.com/builds/
[23:35:25 CET] <furq> https://ffmpeg.zeranoe.com/builds/
[23:35:32 CET] <furq> bah
[23:35:53 CET] <Guest33272> i'm running a 64 bit so...
[23:35:58 CET] <Guest33272> which of the above 3 would I select?
[23:36:01 CET] <Guest33272> all of them or..?
[23:36:15 CET] <kepstin> Guest33272: read the page, it says right on it
[23:36:23 CET] <furq> If you are confused about what build type you need just download a static build.
[23:36:28 CET] <furq> it makes no difference whether you get 32 or 64 bit
[23:36:38 CET] <Guest33272> Alright, thanks, ill do that first
[23:37:06 CET] <jookiyaya> https://sourceforge.net/projects/ffmpeg-hi/   this claims it comes with fdkaac and  10bit x264 ?  what is 10bit
[23:38:04 CET] <jookiyaya> does regular build not come with 10 bit?
[23:38:11 CET] <J_Darnley> 10 bit sample depth
[23:38:14 CET] <jookiyaya> is that against license too?
[23:38:26 CET] <J_Darnley> and it better not come with fdkaac
[23:38:33 CET] <furq> someone's getting reported
[23:38:48 CET] <furq> 10-bit isn't against license but it prevents you from encoding 8-bit x264
[23:38:48 CET] <jookiyaya> furq lol are you serious?
[23:39:33 CET] <furq> and very few players support decoding 10-bit h264
[23:39:46 CET] <jookiyaya> so what does extra 2 bit exactly mean
[23:39:53 CET] <J_Darnley> Only rubbish ones that don't use ffmpeg
[23:40:01 CET] <furq> yeah but that's most of them
[23:40:25 CET] <furq> on the desktop it'll be fine but if you need browser or mobile device support then good luck
[23:40:25 CET] <jookiyaya> why can't they build one that does both?
[23:40:32 CET] <furq> ask the x264 guys
[23:40:39 CET] Action: J_Darnley points back to his last line
[23:40:41 CET] <furq> libvpx and x265 support multiple sample depths
[23:40:53 CET] <jookiyaya> no i will ask the ffmpeg guys
[23:40:55 CET] <furq> in the same build, i mean
[23:41:07 CET] <Guest33272> alright so I got the exe loaded @kepstin
[23:41:08 CET] <J_Darnley> Because you haven't written a patch for it yet, jookiyaya
[23:41:18 CET] <Guest33272> where do I go from here :O
[23:41:41 CET] <jookiyaya> i don't see how ffmpeg cannot include 2 x264  when it can support like 50+ different encoders
[23:42:04 CET] <furq> Guest33272: are you actually invested in using lightworks
[23:42:19 CET] <Guest33272> I was recommended the software by a friend
[23:42:22 CET] <furq> it might be easier to use an NLE which can actually import your files
[23:42:26 CET] <Guest33272> but unfortunately, because I streamed on OBS for a while
[23:42:27 CET] <furq> http://www.openshotvideo.com/
[23:42:28 CET] <furq> like that
[23:42:43 CET] <Guest33272> it's saved all my stream VODs with .flv file extensions
[23:42:55 CET] <Guest33272> so uploading to youtube is rather painful because youtube's editor is terrible.
[23:43:08 CET] <furq> what are you actually hoping to do
[23:43:24 CET] <furq> if you just want basic editing tools then avidemux is probably your best bet
[23:43:26 CET] <Guest33272> I want to take my stream vods (which are .flv file extensions) and convert them into a format where
[23:43:32 CET] <Guest33272> I can edit them with a video editing software
[23:43:41 CET] <furq> i mean what do you want out of your video editing software
[23:44:16 CET] <Guest33272> More or less something that offers me alot of flexibility in how I can customize my video
[23:44:26 CET] <Guest33272> Lightworks seems pretty good, especially for a free software
[23:46:05 CET] <Guest33272> Does that openshotvideo program support .flv file extensions though? cause otherwise I'd be back to square 1
[23:46:26 CET] <Guest33272> http://www.lwks.com/index.php?option=com_kunena&func=view&catid=23&id=110394&Itemid=81#110479 <- this link told me to use ffmpeg but now I'm pretty confused haha
[23:46:40 CET] <jookiyaya> i don't understand the 8/10 bit: i thought rgb is 24bit
[23:47:28 CET] <J_Darnley> And how many bits does each red, green, and blue sample use?
[23:48:02 CET] <jookiyaya> oh okay: so 8 bit = 24bit
[23:48:26 CET] <furq> Guest33272: i've never used openshot but it uses ffmpeg for decoding and encoding so it should support everything ffmpeg supports
[23:48:27 CET] <jookiyaya> so 10 bit = 30bit RGB ?
[23:48:47 CET] <J_Darnley> Basically, yes
[23:48:55 CET] <furq> "Also, you can import and edit H.264 directly in Lightworks if you use the ultrafast preset of the OBS Studio H.264 encoder."
[23:49:01 CET] <jookiyaya> but in windows, it claims it uses 32bit color
[23:49:13 CET] <J_Darnley> 8 bits of padding or alpha!
[23:49:14 CET] <furq> that's troubling
[23:49:20 CET] <Guest33272> hm?
[23:49:31 CET] <J_Darnley> Learn yourself some stuff
[23:49:39 CET] <Guest33272> I'm kind of new with the whole video editing software thing, but I recognize the importance of it for my viewers
[23:49:59 CET] <Guest33272> so I'm doing my best to work with what I have and I've got alot of footage in .flv format
[23:50:13 CET] <Guest33272> that I'd like to be able to export to youtube after editing in a video editing software
[23:50:32 CET] <Guest33272> if openshot can do that, then I'd use it gladly
[23:50:48 CET] <TD-Linux> youtube can parse tons of stuff so that shouldn't be an issue
[23:52:09 CET] <jookiyaya> what does  8 bits of padding or alpha  exactly do?
[23:52:20 CET] <c_14> magic
[23:52:31 CET] <J_Darnley> Fills up the rest of the doubleword
[23:52:31 CET] <Guest33272> I mean, I could put it into youtube but then
[23:53:13 CET] <J_Darnley> and alpha is for transparency
[23:53:40 CET] <jookiyaya> "doubtword"?
[23:54:02 CET] <J_Darnley> no, doubleword
[23:54:08 CET] <jookiyaya> what is doubleward?
[23:54:12 CET] <J_Darnley> 4 bytes
[23:54:12 CET] <jookiyaya> word*
[23:54:58 CET] <J_Darnley> Wow, I could swear some people have never looked at assembly on x86(!)
[23:55:57 CET] <jookiyaya> i remember in windows, you had option to choose  32bit color or 16bit color:   how many bits of rgb does 16bit color use and how much of it were part of alpha
[23:56:10 CET] <J_Darnley> Probably none
[23:56:22 CET] <J_Darnley> Idon't remeber whether Windows used 565 or 555
[23:56:55 CET] <jookiyaya> so  16bit color =  only 5.333 bit
[23:57:25 CET] <jookiyaya> i could easily tell the difference between 16bit and 32 bit color settings
[23:57:39 CET] <jookiyaya> in windows
[23:58:00 CET] <jookiyaya> also it support 256 colors too
[23:59:00 CET] <furq> jookiyaya: https://gist.github.com/l4n9th4n9/4459997
[23:59:28 CET] <furq> just ignore all the stuff about animes
[00:00:00 CET] --- Wed Mar 16 2016