[Ffmpeg-devel-irc] ffmpeg.log.20151224

[burek]

[04:28:15 CET] <x1> I intend to record high-quality screencasts over several hours without losing too much diskspace. How could I pick my codecs effectively?
[04:29:00 CET] <x1> My criteria are 1) lossless 2) comparitively modern 3) support only required for Linux machines
[04:35:38 CET] <jigg> use h264
[05:46:34 CET] <garth> I'm looking to do an intermediate encode of a mpeg2 source (yuv 4:2:0) with as little loss of quality as possible (disk space/access speed is no object), am I right to assume that yuv4 would be the best choice for the video codec? (I'm doing this as I have to use multiple aggressive filters and I'll end up encoding it in a lot of different formats/quality and I don't want to waste time with the filters processing)
[06:30:09 CET] <shibly> Hi, anyone there?
[06:32:07 CET] <shibly> I have a question, may i ask?
[06:32:53 CET] <llogan> shibly: don't ask to ask. just ask.
[06:33:42 CET] <shibly> Can anyone post a good link for ffmpeg c programming tutorial?
[06:33:49 CET] <shibly> ffmpeg programmin in C
[06:47:22 CET] <_6U54N0_> i am tryin to play a acc audio but, only get noise and distortion.
[06:47:22 CET] <_6U54N0_>  i am using SDL_AudioSpec for sound device
[06:47:22 CET] <_6U54N0_>  the mp3 and wav audio sound ok.
[06:47:50 CET] <_6U54N0_> i have tried with resample instructions too, but i get a error, Exception float point and my program die.
[06:48:24 CET] <_6U54N0_> my code base is this tutorial:  http://dranger.com/ffmpeg/tutorial03.html
[06:48:24 CET] <_6U54N0_>  i have made a little modifications to resample and convert audio into the audio_decode_frame() funtion
[06:48:46 CET] <_6U54N0_>  this is the part of code where i made a modification to play a aac audio:
[06:48:46 CET] <_6U54N0_>  http://pastebin.com/mGxciMCJ
[07:07:10 CET] <_6U54N0_> i am using ffmpeg 2.0.4 version from 2014
[07:10:07 CET] <thomedy> hello
[07:10:13 CET] <thomedy> is anyone working
[07:10:16 CET] <thomedy> im out of my depth
[07:10:20 CET] <thomedy> and google isn't doing much
[08:47:28 CET] <pinPoint> erm?
[09:04:41 CET] <thomedy> help please
[09:04:57 CET] <thomedy> i cant seem to figure out how to encode to avc1 and mpeg4
[09:08:32 CET] <t4nk896> What should be to blame if the audio through a stream is skipping badly. YTDL-CORE, ffmpeg, or my internet?
[09:35:40 CET] <thomedy> i have encoded into so many variations
[09:54:58 CET] <fritsch> thomedy: does it work if you download the file?
[10:01:27 CET] <thomedy> fritsch: are you still availabel to help?
[10:01:40 CET] <fritsch> if you answer my question
[10:01:51 CET] <fritsch> does the file work from local disk?
[10:01:52 CET] <thomedy> i have downloaded and run web
[10:02:00 CET] <thomedy> most have been my l ocal storage
[10:02:27 CET] <thomedy> i have tried encoding multiple variations and the closest i have come is aac and mp4v
[10:02:30 CET] <fritsch> so then, provide logfiles of ffplay / ffprobe and so on
[10:02:40 CET] <fritsch> does the original file work from local disk?
[10:02:40 CET] <thomedy> the only file that works right is aac and avc1
[10:02:58 CET] <thomedy> it plays i am just trying to step frame by frame through the vid with javvascript
[10:03:14 CET] <thomedy> and for some reason only that particular codecruns real time
[10:03:27 CET] <fritsch> so you live encode?
[10:03:32 CET] <fritsch> I don't get your usecase yet
[10:03:36 CET] <fritsch> you stream?
[10:03:44 CET] <thomedy> here watch this one sec
[10:03:52 CET] <fritsch> try to describe your use case and what you are doing ....
[10:04:05 CET] <fritsch> looking at resulting files is worthless if one does not know what you do
[10:04:51 CET] <thomedy> okay my use case is very specific....
[10:04:56 CET] <thomedy> i have a website im buildin
[10:05:11 CET] <thomedy> the desin incorporates a video paused
[10:05:28 CET] <thomedy> and then as i scroll left or right it either moves forward frame by frame or backwards
[10:05:50 CET] <thomedy> i can only iterate through one type of video
[10:05:52 CET] <fritsch> so you "seek frame wise"?
[10:06:04 CET] <thomedy> yyes
[10:06:07 CET] <fritsch> hehe
[10:06:19 CET] <fritsch> try mjpeg codec
[10:06:21 CET] <fritsch> just for fun
[10:06:35 CET] <thomedy> oh my god i hope your right
[10:06:36 CET] <thomedy> one sec
[10:06:58 CET] <fritsch> you know h264 and such have reference frames
[10:07:06 CET] <fritsch> that are needed in forward and backward direction
[10:07:14 CET] <fritsch> to properly decode something
[10:07:41 CET] <fritsch> and what about the audio?
[10:07:49 CET] <fritsch> how is that related?
[10:08:23 CET] <thomedy> i dont know if it is
[10:08:27 CET] <fritsch> "moves backward" frame by frame ... moving backward is hard for a hw decoder
[10:08:35 CET] <thomedy> logically i would say it isn't but i want to start one step at  a time
[10:08:38 CET] <thomedy> with a working model
[10:08:41 CET] <thomedy> its muted anyway
[10:08:46 CET] <thomedy> and i think its aac
[10:08:49 CET] <fritsch> yeah don't unmute it
[10:08:54 CET] <thomedy> nope
[10:08:56 CET] <thomedy> not at all
[10:09:00 CET] <thomedy> its purly visual
[10:09:17 CET] <thomedy> but what ever variable that i can match i just want a working verison
[10:09:25 CET] <thomedy> i know my fps is 29.97
[10:09:33 CET] <thomedy> and i hae test ed that
[10:09:43 CET] <fritsch> 29.97 sounds interlaced
[10:10:08 CET] <thomedy> ill be honest i dont know what interlaced means
[10:10:16 CET] <fritsch> you are writig a video editor?
[10:10:27 CET] <thomedy> but i have to assume the pixels are woven into the frames together
[10:10:28 CET] <thomedy> ad no
[10:10:28 CET] <fritsch> cause I don't see a point at all to go back and forth frame wise
[10:10:29 CET] <thomedy> not at all
[10:10:31 CET] <fritsch> per click
[10:10:43 CET] <fritsch> you need to click 29.97 times to get one second of video
[10:10:54 CET] <thomedy> nope... essentially all i want is viusally a side scrolling parallax website for portfolio work
[10:11:29 CET] <fritsch> and how long is that video?
[10:11:32 CET] <thomedy> yeah that i get
[10:11:35 CET] <thomedy> 7 seconds
[10:11:38 CET] <thomedy> that is it
[10:11:39 CET] <fritsch> ah!
[10:11:42 CET] <thomedy> but i refuse to give up
[10:12:03 CET] <fritsch> okay - then why not use some js + created pngs?
[10:12:17 CET] <fritsch> 29*7 ~ 210 images
[10:12:28 CET] <fritsch> then you can jump from index to index
[10:12:37 CET] <fritsch> and preload them
[10:13:02 CET] <fritsch> or better said preload 20 back and 20 forth a time and dynamically load the next chunk into browser memory
[10:13:12 CET] <thomedy> okay
[10:13:18 CET] <fritsch> just an idea
[10:13:27 CET] <thomedy> you know what i did that
[10:13:30 CET] <thomedy> and i liked it almost
[10:13:34 CET] <thomedy> but it doesn't load right
[10:13:40 CET] <thomedy> it only works in firefox
[10:13:41 CET] <fritsch> hehe - that's another issue
[10:13:48 CET] <thomedy> i know
[10:13:56 CET] <thomedy> i m jus ttrying not to give up
[10:13:58 CET] <fritsch> try mjpeg now - and see if works better
[10:14:04 CET] <thomedy> i am its encoding
[10:14:13 CET] <thomedy> while i was typing that it finished
[10:14:14 CET] <thomedy> one sec
[10:14:29 CET] <thomedy> im goingg to run ffmpeg -i file real quick to get my info
[10:14:41 CET] <fritsch> you seem you do like a hackjob :-)
[10:14:45 CET] <fritsch> hehe
[10:15:00 CET] <fritsch> oki - afk, I'd personally would go with the image stack for a 7 second demonstration
[10:15:07 CET] <fritsch> and use JS to correctly cash that stuff
[10:15:13 CET] <thomedy> the thing is their big images
[10:15:20 CET] <fritsch> browser video capabilities in the sense you want them is hard
[10:15:35 CET] <fritsch> which dimensions does the video need to have?
[10:15:59 CET] <fritsch> you can add quite a lot compression to pngs
[10:16:21 CET] <thomedy> at min i would think 75% of screen height/width and that should keep an okay resolution
[10:16:24 CET] <fritsch> but yeah obviously the codec will be much better video wise, as you can save a lot data cause of not so fast changing content
[10:16:24 CET] <thomedy> i want good resolution
[10:16:31 CET] <thomedy> i even optinpg'd the pngs
[10:16:40 CET] <thomedy> and reduced their size by 40 %
[10:16:53 CET] <fritsch> how large is the mjpeg video?
[10:17:05 CET] <fritsch> and how large the h264 version?
[10:17:17 CET] <thomedy> actually let me check one sec
[10:17:55 CET] <thomedy> mjpeg seems about 3x larger...  have i got that right
[10:17:59 CET] <thomedy> ls -al is all i did
[10:18:12 CET] <fritsch> yes
[10:18:19 CET] <fritsch> as every frame is a full frame
[10:18:24 CET] <fritsch> just to test your code
[10:18:32 CET] <fritsch> try it from local browser and local disk
[10:19:13 CET] <fritsch> at the end there is still gif and apng :-) and such things
[10:19:17 CET] <thomedy> i usually run local host its just easier and saves on data
[10:19:24 CET] <thomedy> but it seems to be right right now
[10:19:28 CET] <fritsch> okay - so does the video display?
[10:19:30 CET] <thomedy> good point
[10:19:34 CET] <thomedy> im checking the js next
[10:20:11 CET] <fritsch> you can reduce the jpeg quality a bit further (depending on the params you gave it)
[10:20:22 CET] <fritsch> the good about mjpeg - no further codec is needed
[10:21:42 CET] <thomedy> well that has to be the right codec
[10:21:49 CET] <thomedy> the sample rate is correct
[10:22:02 CET] <thomedy> but i have no idea why i still cant step through the frame by frame
[10:22:02 CET] <fritsch> what is sample rate in regard to video?
[10:22:17 CET] <fritsch> as I have no idea how you step
[10:22:20 CET] <fritsch> I cannot help you there
[10:22:30 CET] <fritsch> here is another idea:
[10:22:33 CET] <fritsch> make the pngs
[10:22:37 CET] <fritsch> make multiple of them
[10:22:41 CET] <fritsch> in different qualities
[10:22:50 CET] <thomedy> 48000 and 44000
[10:22:50 CET] <fritsch> measure the bandwidth with the onload method of the images
[10:22:53 CET] <fritsch> and change accordingly
[10:22:56 CET] <thomedy> i am really not sure what that means
[10:22:56 CET] <fritsch> that's audio
[10:23:12 CET] <thomedy> so that really should not matter
[10:23:16 CET] <fritsch> if bandwidth is good load hq frames, if not lower
[10:23:20 CET] <thomedy> i cant tell what im missing
[10:23:21 CET] <fritsch> and let the browser scale them
[10:23:22 CET] <fritsch> done
[10:23:36 CET] <fritsch> yeah me neither :-)
[10:23:36 CET] <thomedy> what is hq?
[10:23:41 CET] <fritsch> high quality
[10:23:46 CET] <fritsch> lq: low quality
[10:23:49 CET] <fritsch> mq: medium quality
[10:23:53 CET] <thomedy> i might have to go back to the frame by frame images
[10:23:58 CET] <fritsch> fsnq: fritsch's super nice quality
[10:24:00 CET] <thomedy> but i already know its jumpy that way
[10:24:12 CET] <fritsch> yeah - writing a "player" which is what you do in js
[10:24:16 CET] <fritsch> is a problem
[10:24:23 CET] <fritsch> if it's jumpy your pre caching does not work
[10:24:39 CET] <fritsch> use an image array and cache 10 images before and 10 images after the current point you are
[10:24:43 CET] <thomedy> my players ihave written in the past have neer been a problem
[10:24:43 CET] <fritsch> which is dynamically loading
[10:24:49 CET] <fritsch> then use those :-)
[10:25:12 CET] <thomedy> sure right now i havce pre loaded about a 150 of them
[10:25:22 CET] <thomedy> okay its late i hae to work in the morning
[10:25:26 CET] <thomedy> im really diggin gthis design
[10:25:49 CET] <thomedy> thank you by the wayor the life of me i never would have thought mjpeg
[13:49:22 CET] <_6U54N0_>  http://pastebin.com/mGxciMCJ
[13:51:59 CET] <fritsch> _6U54N0_: check the swr_alloc_set_opts method
[13:52:40 CET] <_6U54N0_> ok,
[13:52:48 CET] <fritsch> wait
[13:52:58 CET] <fritsch> convert dies, right?
[13:53:02 CET] <fritsch> what's the backtrace?
[13:53:35 CET] <fritsch> the memcpy most likely misses the sample_size, but out_samples is in bytes
[13:53:38 CET] <fritsch> check your variable names
[13:53:39 CET] <fritsch> :-)
[13:54:57 CET] <fritsch> check &frame.data[0]
[13:55:13 CET] <fritsch> that looks wrong, or is that interleaved audio input?
[13:57:25 CET] <_6U54N0_> when put NULL by frame.dat[0]  then my program run but in silence.
[13:58:42 CET] <fritsch> hehe
[13:58:54 CET] <fritsch> i think so - it seems it's not filled correctly then
[13:59:12 CET] <_6U54N0_> but i dont know what put in this placer.
[13:59:25 CET] <_6U54N0_> place*
[13:59:50 CET] <fritsch> if you have floatp
[13:59:56 CET] <fritsch> input it won't work like that
[14:00:59 CET] <fritsch> av_samples_get_buffer_size
[14:02:55 CET] <fritsch> post the backtrace of the crash (that code is quite unreadable) if I don't study it for 15 minutes
[14:03:17 CET] <_6U54N0_> av_samples_get_buffer_size value is equal to frame.linesize[0]   that work ok.
[14:03:30 CET] <fritsch> need to go afk
[14:04:25 CET] <_6U54N0_> my code base is this tutorial:  http://dranger.com/ffmpeg/tutorial03.html
[14:05:20 CET] <_6U54N0_> i only made a little modification to resample and convert audio, ... not more.
[14:11:49 CET] <jonascj> Hi guys. when you specify 3500k as bitrate in ffmpeg, is it 3500*1024 bit/s or 3500*1000 bit/s?
[14:12:11 CET] <JEEB> 1000
[14:12:26 CET] <jonascj> IEC has defined KiB to mean 1024 bytes and KB to mean 1000 bytes...
[14:12:29 CET] <JEEB> there are separate suffixes for 1000 and 1024, IIRC
[14:12:35 CET] <JEEB> jonascj: yeah but this is bits
[14:12:45 CET] <JEEB> 1000 is more used with bits (per second)
[14:13:12 CET] <JEEB> also IIRC there are separate suffixes for both :P
[14:13:22 CET] <jonascj> alright, bits/bytes is just a matter of 8, but k (1000) vs ki(1024) adds up when you have many many giga
[14:13:25 CET] <waressearcher2> 300 baud/s
[14:13:56 CET] <waressearcher2> jonascj: du hast rechts
[14:14:18 CET] <jonascj> alright, audio bitrate of 256k in ffmpeg means 256*1000 bit/s, great, now I know how to calculate target bitrate for a given target file size
[14:14:40 CET] <jonascj> I got screwed by assuming that both DVDs and ffmpeg took G, k, m etc. to mean 1024^n, not 1000....
[14:17:43 CET] <JEEB> well now you'll know what to expect from other video related tools regarding bit rates :)
[14:18:38 CET] <jonascj> yup, and I've reeducated myself as to what hdd and optical media manufactureres specify
[14:18:41 CET] <jonascj> :-/
[14:18:45 CET] <jonascj> maybe it has been so for 15 years :P
[14:20:02 CET] <jonascj> when FFMPEG reports 252 kb/s for an audio stream it is also kilobit (1000 bit)?
[14:21:25 CET] <jonascj> or is it kilobyte?
[14:23:00 CET] <DHE> lowercase b is bits
[14:23:32 CET] <jonascj> "duration 04:01:24.40, start 0.00, bitrate: 20237 kb/s", "stream #0:0(eng) Audio: acc... 252 kb/s)" and "stream #0:1 Video: h264 .... 19979 kb/s"
[14:25:08 CET] <jonascj> Good, thanks
[14:36:28 CET] <jonascj> Anything critical I miss in this "ffmpeg -i input.mp4 -c:v libx264 -preset veryslow -b:v 4100k -vf scale=-1:720 -pass 1 -c:a copy -f mp4 /dev/null" and the corresponding "-pass 2" command?
[14:37:29 CET] <jonascj> It is some h264 1080p video with bitrate 20237 kb/s. I am trying to compress it to 8GB, by scaling to 720p and decreasing the bitrate. Those numbers now work out because you guys telling me k=1000 in ffmpeg.
[14:37:52 CET] <jonascj> But have I forgotten an important switch in my command, the quality parameter of x264 or similar?
[14:38:20 CET] <DHE> it looks good enough. just understand that the second pass will run a lot slower
[14:40:13 CET] <jonascj> than pass1? I know, I don't know how slow, but it will be done in 12 hours or something like that
[14:40:39 CET] <jonascj> (i have run it once before where I messed up the sizes/rates because of 1000 vs 1024)
[14:45:31 CET] <DHE> think of it this way. if you want to make any changes you can usually just re-run pass 2 by itself.
[14:46:42 CET] <jonascj> oh okay, I didn't know. I thought the parameters of pass 1 and pass 2 had to match
[14:48:42 CET] <jonascj> does -preset matter on pass 1?
[14:55:19 CET] <DHE> minimally. pass 1 usually tweaks the settings in favour of speed
[14:55:55 CET] <jonascj> will this be variable bitrate, or constant bitrate?
[14:56:07 CET] <jonascj> the commands I gave that is.
[14:59:54 CET] <jonascj> constant since it is with -b:v
[15:06:42 CET] <DHE> for constant bitrate you need to also set -maxrate, -minrate and -bufsize to something
[15:06:48 CET] <DHE> otherwise it will just target a final average bitrate
[15:38:27 CET] <jonascj> DHE: targeting a final average bitrate seems like what you want in a mixed video? Lower the bitrate where you can to be able to have higher bit where it is needed most.
[15:46:10 CET] <DHE> typically. for a personal media collection this is usually fine
[15:50:26 CET] <jonascj> alright, I'll try to think about when it is more appropriate or wanted with a really constant video bitrate.
[15:57:46 CET] <DHE> CBR is usually used when there's a maximum bitrate target (eg: streaming sources)
[15:57:55 CET] <DHE> where variance isn't really tolerable
[16:26:38 CET] <_6U54N0_> does somebody knows a example to convert aac audio to audio S16
[16:26:55 CET] <_6U54N0_> for SDL_AudioSpec
[17:23:38 CET] <tapas_> i want to see how ffmpeg is autthenticating with rtmp server. how do i enable log for that ?
[18:52:25 CET] <_6U54N0_> does somebody knows a example to convert aac audio to audio S16, cause i am using SDL_AudioSpec audio device.
[18:54:07 CET] <_6U54N0_> this is a part of code:   http://pastebin.com/mGxciMCJ
[18:55:27 CET] <_6U54N0_> on the line 75 my program die
[18:56:14 CET] <_6U54N0_> error Exception floating point.
[19:16:55 CET] <ashka> hi, I am having issues with writing to a file from a pipe. I am using ffmpeg 2.5.4-1. I am running this command : ffmpeg -i - -vcodec copy -acodec copy out.flv, the audio is AAC 48k ( http://pastebin.com/EbfNxQYf ), however inspecting the saved file says AAC 44.1k, and the audio is pitched down when played ( http://pastebin.com/QBjXvR3v )
[19:17:45 CET] <ashka> if this is a bug and not me doing something wrong I'll try to upgrade but how can I fix the files I already have ? doing ffmpeg -i in.flv -vcodec copy -ar 48000 out.flv resamples without pitching the audio back up
[19:25:15 CET] <waressearcher2> ashka: hallo, wie geht's ?
[19:47:41 CET] <nyuszika7h> I would like to remove non-English audio tracks from files, but the tracks may not always be in the same order, is there a way to select based on language?
[19:48:28 CET] <JEEB> try adding the language code into a map command and see if that works, -map 0:a:en f.ex.
[19:50:11 CET] <nyuszika7h> doesn't seem to work, it just chooses the first audio track
[19:50:35 CET] <JEEB> which is not english?
[19:50:54 CET] <JEEB> I guess it doesn't work like that then... welp
[19:52:50 CET] <nyuszika7h> yeah it's not English
[19:52:59 CET] <JEEB> alright
[19:53:07 CET] <JEEB> I had only used type:number so far :)
[19:53:15 CET] <JEEB> I guess you'll have to script it with ffprobe or so then :)
[19:53:32 CET] <nyuszika7h> hmm, -map_metadata is a thing
[19:53:53 CET] <JEEB> that copies specific metadata, IIRC
[19:53:55 CET] <furq> nyuszika7h: -map 0:m:language:eng
[19:53:59 CET] <nyuszika7h> the docs seem to suggest that's something else though, yeah
[19:54:10 CET] <nyuszika7h> furq: ah, thanks
[22:21:30 CET] <metRo_> hello, I'm trying to cross compile ffmpeg but it keeps give an error about smbclient. I have installed it and I can confirm libsmbclient.h is in sysroot. I'm cross compiling on an ubuntu machine to run on a raspberry pi
[22:21:39 CET] <metRo_> can you help me identify the problem_
[22:21:40 CET] <metRo_> ?
[22:21:50 CET] <metRo_> I tried to set pkp_config path
[22:22:00 CET] <metRo_> I have the ./configure log if you want to take a look
[22:22:02 CET] <metRo_> thank you
[22:22:07 CET] <metRo_> and merry christmas all :)
[22:23:03 CET] <waressearcher2> metRo_: hallo, wie geht's es dir ?
[22:24:11 CET] <metRo_> waressearcher2: sorry but I don't understand you
[22:26:49 CET] <poseidon1157> Hi all
[22:26:55 CET] <waressearcher2> poseidon1157: hallo
[22:26:58 CET] <poseidon1157> Merry christmast
[22:27:02 CET] <poseidon1157> *Christmas
[22:27:23 CET] <poseidon1157> Also, does anyone know how to turn several moof segments back into a mp4?
[22:27:34 CET] <poseidon1157> Or is that regarded as a 1 way transformation?
[22:29:39 CET] <metRo_> waressearcher2: here is the congif output logfile if you want to take a look http://pastebin.com/LqHuea74
[00:00:00 CET] --- Fri Dec 25 2015