[Ffmpeg-devel-irc] ffmpeg.log.20151116

[burek]

[01:05:11 CET] <Soni> how can I use paletteuse with no dithering?
[01:13:57 CET] <PhoenixBR> Hi guys! Here I am again. Just to be sure and sleep well.
[01:14:22 CET] <PhoenixBR> uyvy422 -> yuv422p conversion does not lose information, right?
[01:14:56 CET] <PhoenixBR> using -f rawvideo -vcodec rawvideo
[01:16:06 CET] <PhoenixBR> JEEB: ?
[01:56:44 CET] <Soni> how can I use paletteuse with no dithering?
[02:03:11 CET] <c_14> I don't think you can.
[02:23:51 CET] <Soni> I kinda can
[02:24:10 CET] <Soni> but mpeg artifacts >.<
[02:29:52 CET] <Soni> ffmpeg should learn multithreading and OpenCL (if it already knows that, how do I enable it?)
[02:31:40 CET] <Soni> also truecolor gif
[02:33:37 CET] <c_14> There's a couple of filters with opencl support iirc
[02:33:50 CET] <c_14> And it does thread (though it depends on the en and decoders)
[03:02:18 CET] <Soni> removegrain is slow :/
[04:43:46 CET] <Soni> c_14, apparently I can use dither=none
[06:18:34 CET] <Bray90820> Is there an exe of ffmpeg for windows 7
[06:22:04 CET] <Bray90820> Would this be the right code to install ffmpg on windows
[06:22:07 CET] <Bray90820> setx /M PATH "C:\Users\Scott\Desktop\ffmpeg-2.8.2;%PATH%"
[06:22:14 CET] <Bray90820> Assuming you put the folder on your desktop
[06:28:44 CET] <waressearcher2> Bray90820: hallo
[06:29:07 CET] <Bray90820> waressearcher2: Hi
[06:29:28 CET] <waressearcher2> Bray90820: wie geht's ?
[06:29:45 CET] <Bray90820> What?
[06:32:35 CET] <Bray90820> waressearcher2: ?
[06:35:02 CET] <waressearcher2> Bray90820: warum hast du 90820 Nummer aber nicht 90210 ?
[06:43:11 CET] <Bray90820> waressearcher2: Ich bin mit einem Wandler Ich spreche nur Englisch, aber die Zahlen beziehen sich auf meinen Geburtstag nicht Postleitzahl, wenn ich Sie richtig verstanden
[06:47:16 CET] <Soni> so uhh can anyone help me?
[06:48:12 CET] <Soni> I have this pixel art thingy in a 1080p mp4 and I wanna convert it to a 90x72 apng
[06:48:35 CET] <Soni> (1350x1080 is the mp4 resolution)
[06:48:53 CET] <Soni> the thing is whatever I do introduces nasty dots everywhere
[07:01:39 CET] <Prelude_Zzzzz> hey everyone.. question ... i am using ffmpeg to pull in a stream and i get errors like " Failed to open segment of playlist 0 " ..frame rate is also low.. when i use vlc on my pc to play it , it plays fine
[07:01:42 CET] <Prelude_Zzzzz> any ideas ?
[07:01:53 CET] <Prelude_Zzzzz> some param i am missing or something on the input ?
[07:03:01 CET] <Prelude_Zzzzz> it seems the links to the .ts files inside the m3u8 are full URL:// streams.. mabye something in the url stream is & and not under " " hence it can't play correctly ?
[07:04:54 CET] <Bray90820> Can someone explain to me this error
[07:04:55 CET] <Bray90820> http://pastebin.com/raw.php?i=zQfSN7HR
[07:11:02 CET] <Bray90820> Here is the full output
[07:11:02 CET] <Bray90820> http://pastebin.com/raw.php?i=JnRf9du0
[07:13:01 CET] <matt___> Hello, if you have a few minutes, I'm looking for some command syntax help.  I have a mpegts stream with no audio that is being written to a named pipe.  I can verify that the source feeding the pipe is producing data, but ffmpeg appears to be attempting to process audio.  How do I specify that there is no audio channel in the input stream?
[07:13:47 CET] <matt___> Also, are there configuration options to drop bad input but otherwise continue processing the stream?  For example, if there is a bad chunk, I would want ffmpeg just resume at the next good one; maybe it does this by default for mpegts.
[07:16:46 CET] <waressearcher2> matt___: hallo
[07:19:36 CET] <matt___> Hi
[07:20:19 CET] <waressearcher2> matt___: wie geht's es dir ?
[07:29:04 CET] <Soni> can anyone help me fix compression artifacts in a pixel art video?
[07:30:09 CET] <matt___> So so. Können Sie mir helfen mit dem Problem?
[07:34:58 CET] <waressearcher2> matt___: nein, leider nicht
[07:38:16 CET] <matt___> kein problem. danke
[08:19:28 CET] <matt___> Does anyone know how I can take a mpegts stream as input when it has no audio?  Ffmpeg appears to be trying to process the video channel as an audio stream.  I attempted to pass -an, but to no avail.  Thanks in advance
[08:20:24 CET] <Soni> so splitting a video into multiple files makes ffmpeg much faster (I guess because it can do the files in parallel instead of sequentially)
[08:29:59 CET] <fritsch> Soni: no ... most codecs can use threading
[08:30:44 CET] <Soni> fritsch, lossless?
[09:11:05 CET] <flux> matt___, to me it sounds like the stream is buggy somehow..
[09:19:54 CET] <matt___> flux, thanks for your response; I'm figuring that may be the case
[09:20:41 CET] <matt___> unfortunately I don't have much control over the format
[09:22:33 CET] <matt___> how does ffmpeg handle bad chunks in a mpegts stream?  does it just drop them and continue processing from stdin/a fifo [in my case]?  I'm trying to setup a simple UDP livestream, and if there's buggy input, I would like it to be skipped ideally
[12:36:47 CET] <zotherstupidguy> is it possible to use ffmpeg for taking screenshots of a daemon process?
[12:39:48 CET] <flux> ..usually daemons don't have screens to take shots of?
[12:47:15 CET] <zotherstupidguy> flux true, but does that mean that the GUI data doesn't exist at all?
[12:48:42 CET] <zotherstupidguy> it is the same as if you are running xwindows on a remote server, my guess and i am not sure, is that the x protocol sens the data to the client , so in the same sense you can run a deamon and record the data via a script into a video or somthing
[12:49:07 CET] <BtbN> ... what?
[12:49:19 CET] <BtbN> Taking a screenshot of what exactly?
[12:49:49 CET] <zotherstupidguy> of a program that runs as a deamon
[12:50:27 CET] <zotherstupidguy> does it have to be displayed really to take a screenshot of it?
[12:51:30 CET] <BtbN> What program?
[12:51:41 CET] <BtbN> You can't take a screenshot of your webserver oO
[12:52:20 CET] <furq> is this even a program that normally has a gui
[12:52:34 CET] <flux> I guess zotherstupidguy means a minimized program?
[12:52:53 CET] <flux> but typically programs that aren't displayed aren't drawn anywhere..
[12:53:29 CET] <flux> unless perhaps there is a virtual surface provided where the program is 'visible', but that's clearly out of ffmpeg's scope :)
[12:58:36 CET] <DHE> well how would ffmpeg produce a picture of it then? I don't think there's a matrix-view indev available. :)
[13:00:45 CET] <flux> dhe, not bad, ffmpeg pid://`pidof httpd` ;-)
[13:00:54 CET] <chungy> There's some programs you'd run with a dummy Xorg driver.... certain game servers that are Windows-only and draw a GUI, you can shove them off to an X server not connected to any display and run them via Wine.
[13:01:05 CET] <chungy> It's an obscure use-case, it exists... I doubt that's what zotherstupidguy wants though. :P
[13:01:07 CET] <flux> I guess ffmpeg works just fine with those
[13:01:58 CET] <chungy> (and mind you that Windows itself even allows proper daemons without any sort of GUI or terminal... so it's just a really lazy way to make a server and not common at all)
[13:02:04 CET] <flux> let's say I had a proprietary ip-cam camera with only IE6.0 ActiveX plugin to view the contents, maybe a hack like that would be useful ;).
[13:02:12 CET] <chungy> aye
[13:02:45 CET] <chungy> You could probably use ffmpeg to visualize a log somehow, but I don't think that fits the question
[13:03:39 CET] <flux> ffmpeg doesn't yet have an ansi sequence renderer, does it?-)
[13:03:51 CET] <flux> could be useful for converting those nethack streams!
[13:05:25 CET] <chungy> Haha, I don't know. I imagine you'd probably have a python script to convert text to images, then ffmpeg to take those to video...
[13:06:26 CET] <DHE> well, in reality I use Xvnc when I need a GUI on a non-GUI system. essentially X server without a GPU but accepts VNC sessions
[13:08:28 CET] <zotherstupidguy> flux nethack videos was done by emacs i guess! not sure
[13:08:50 CET] <furq> probably done with ttyrec or termrec or something
[13:08:59 CET] <chungy> zotherstupidguy: do you mind explaining what you're trying to do?
[13:09:02 CET] <zotherstupidguy> oh true
[13:09:27 CET] <zotherstupidguy> i am trying to record a video of emacs file editing, but i dont want to open emacs
[13:09:30 CET] <flux> furq, ttyrec does bitmaps? video files? the description seems like it does only terminal.
[13:09:36 CET] <zotherstupidguy> emacs file editing via a script
[13:09:40 CET] <furq> there are forks which save to png
[13:09:56 CET] <chungy> ttyrec only records console output, including control codes
[13:10:54 CET] <furq> i just use asciinema these days
[13:11:13 CET] <zotherstupidguy> as a video recording of a deamon sounds too much of a leap, i am asking about getting an image then do the video later
[13:15:19 CET] <furq> zotherstupidguy: https://asciinema.org/a/637
[13:17:22 CET] <flux> I imagine asciinema should work just as well in screen, and you can start screen as a daemonized session
[13:18:30 CET] <furq> it works just fine in tmux
[13:20:09 CET] <zotherstupidguy> help me get this straight, asciinema can just record a video of somthing without it actually being displayed anywhere?
[13:22:29 CET] <furq> yes
[13:22:52 CET] <furq> it'll continue recording inside a detached tmux session
[13:23:12 CET] <furq> i assume screen will work just as well
[13:23:52 CET] <furq> you might have to be able to reattach to it to get it to finish recording, but there's probably some way around that
[13:24:02 CET] <zotherstupidguy> i think thats good enough :)
[13:24:23 CET] <zotherstupidguy> asciinema is opensource, correct?
[13:24:48 CET] <furq> https://github.com/asciinema/asciinema
[13:24:51 CET] <zotherstupidguy> i think the trick is they are creting each new recording inside its own shell instance
[13:25:18 CET] <furq> https://github.com/asciinema/asciinema-player
[13:41:32 CET] <Soni> how can I use ffmpeg to smooth a pixelart(?) mp4 into a lossless format (e.g. split PNGs, APNG, etc)?
[13:44:09 CET] <Soni> (and yes that includes artifact removal)
[14:08:16 CET] <DHE> what do you mean artifact removal? if it were that easy every video decoder in  the world would implement it
[14:12:33 CET] <Soni> turn similar colors into the same color given a threshold
[14:13:10 CET] <Soni> e.g. anything between 0xF8F8F8 and 0xFFFFFF into 0xFFFFFF
[14:13:33 CET] <Soni> this is pixel art after all
[14:54:14 CET] <JoshX_> Hello! can i make ffmpeg continue and retry streaming a rtsp stream? now if something fails, ffmpeg exits
[14:59:14 CET] <livtyler> how do I have filter-complex, I'm using ubuntu
[14:59:14 CET] <Ulrar> Hi, I am looking for a way to make the "alpha" option work in drawtext, any idea ? I get that error : [Parsed_drawtext_0 @ 0x19b0a80] Option 'alpha' not found
[15:56:59 CET] <bencc1> anYc: I've posted two issues with link to videos
[16:00:22 CET] <yongyung> I'm having trouble understanding the .wav format ffmpeg is using. Is it not just the "windows" .wav format? This is the start of a .wav file generated by ffmpeg: http://i.imgur.com/sDdFM0F.png - Now, according to this: https://msdn.microsoft.com/en-us/library/windows/desktop/dd757929(v=vs.85).aspx and this https://msdn.microsoft.com/en-us/library/windows/desktop/dd390970(v=vs.85).aspx the WAVEFORMATEX structure (marked by me in blue)
[16:00:22 CET] <yongyung> should be 18 byte long, but it's only 16 byte long (missing the "cbsize" header), can someone here explain wtf is going on there? Am I just looking at the wrong specification?
[17:44:57 CET] <Yutz> there is a video i want to download from the net.... its subdivied into mini video files.....  ive never really messed with this stuff.. so I assum e ffmpeg is what I want to use?
[17:46:39 CET] <c_14> hmm? something like hls?
[17:46:58 CET] <Yutz> hls?
[17:47:17 CET] <c_14> HTTP Live Streaming
[17:47:22 CET] <Yutz> http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/ is what i want to download
[17:49:20 CET] <furq> Yutz: apparently youtube-dl supports that site
[17:50:45 CET] <Yutz> ahh.. i see...
[17:50:59 CET] <Yutz> so these video downloaders are usually site specific?
[17:51:15 CET] <furq> https://rg3.github.io/youtube-dl/supportedsites.html
[17:51:26 CET] <Yutz> ya.. im there now...
[17:51:36 CET] <Yutz> im just a complete newb to how video works on the internet...
[17:54:34 CET] <furq> there isn't one solution which works for all VOD
[17:55:10 CET] <furq> there are a bunch of different ways of delivering it, including fun stuff like RTMPE
[17:56:19 CET] <Yutz> i seen that acronym...  thats like live video.. correct?
[17:56:28 CET] <furq> video on demand
[17:56:32 CET] <furq> i.e. not live
[17:56:53 CET] <Yutz> ahh... sorry
[17:57:07 CET] <Yutz> ive got python... lemme check my version
[17:57:19 CET] <furq> most distros have it in repos
[17:58:01 CET] <Yutz> i need to update my python.. one min
[18:32:43 CET] <pepee> anyone knows how to reduce the quality of ogg audios?
[18:33:58 CET] <c_14> Lower the bitrate?
[18:34:12 CET] <c_14> Write a worse encoder?
[18:34:42 CET] <pepee> lemme explaing what I'm trying to do
[18:35:07 CET] <JEEB> drag the bit rate so low you aren't capable of hearing what the hell is going on
[18:36:12 CET] <pepee> I have this website that encodes WAV files generated by TTS engines. the idea is to offer "good" and "bad" qualities, and mp3/ogg audios. the problem being, the quality of mp3 and ogg files differs too much
[18:37:02 CET] <furq> wouldn't speex or opus be a better choice for TTS
[18:37:24 CET] <pepee> voices generated by TTS don't need too much quality anyway. atm, I'm doing something that is really stupid, but it works: encode wav to mp3, then mp3 to ogg
[18:38:00 CET] <furq> well they say knowing is half the battle
[18:38:24 CET] <pepee> but it works... furq, honestly, I don't know much about audio encoding
[18:38:52 CET] <pepee> anyway, the idea, as I said, is to offer two qualities. the bad quality should mimic that of a wired phone
[18:38:56 CET] <furq> if "the quality of mp3 and ogg files differs too much" means that the ogg is audibly worse than the mp3, then it's probably because you're doing that
[18:39:08 CET] <furq> especially if they're low bitrate
[18:39:10 CET] <pepee> no, the ogg is much better than the mp3
[18:39:29 CET] <furq> that's what i'd expect at low bitrate
[18:40:07 CET] <c_14> You're encoding wave -> mp3 -> ogg, and the ogg is better than the mp3?
[18:40:32 CET] <pepee> no, it is if I encode directly, wave -> ogg
[18:41:15 CET] <pepee> for wave -> mp3 -> ogg, ogg don't sound too different to the mp3s, but the files are larger
[18:41:29 CET] <pepee> ... depending on the input
[18:42:05 CET] <JEEB> just set the same bit rate for both?
[18:42:15 CET] <JEEB> that way both files should be within a very few % of each other
[18:42:33 CET] <furq> just don't do wav > mp3 > ogg if you can encode them both from the wav
[18:42:54 CET] <JEEB> and if vorbis is still better than mp3, then just lower vorbis's rate even further
[18:42:55 CET] <furq> if you want it to sound like a wired phone then try speex or opus at very low bitrate
[18:42:59 CET] <furq> ~16kbps or thereabouts
[18:44:13 CET] <furq> maybe even lower than that actually
[18:44:54 CET] <furq> or you could try 8khz mp3
[18:46:05 CET] <pepee> for mp3, I'm using 32k for "good" and 16k for "bad"
[18:46:46 CET] <furq> the sample rate is probably more important if you want a characteristic sound
[18:48:55 CET] <furq> also you can encode mp3 and ogg from the same source with the same command
[18:50:05 CET] <pepee> at the same time, or you just mean with the same parameters?
[18:50:31 CET] <pepee> because I was doing that, same params, and still get different qualities
[18:50:47 CET] <furq> ffmpeg -i src.wav -c:a libmp3lame out.mp3 -c:a libvorbis out.ogg
[18:51:18 CET] <pepee> ah, nice!
[18:51:27 CET] <pepee> I had no idea
[18:51:49 CET] <furq> anyway yeah try -ar 8000
[18:51:53 CET] <furq> i'm not sure if mp3 goes lower than that
[18:54:48 CET] <pepee> I ran "ffmpeg -i src.wav -ar 8000 -c:a libmp3lame out.mp3 -c:a libvorbis out.ogg" and I get different qualities
[18:55:08 CET] <furq> you need to specify the bitrate for both
[18:55:39 CET] <furq> and the sample rate for that matter
[18:57:42 CET] <pepee> ah, I'm dumb. thanks, it works great now
[19:03:12 CET] <OrthATG_> Im still having a PTS / AV Sync issue on live broadcasts using the api. I def appreciate the help thus far, but my company is putting out a $2000 bounty to help solve the issue
[19:07:05 CET] <OrthATG_> The core issue is that if the upload internet connection drops or doesn't have enough bandwidth at a given time, then av_write_frame blocks until it can send packets. after that, then sync problems occur
[19:10:17 CET] <jasom> pepee: convert it to 8kHz u-law wav and add a bit of echo to make it sound like phone
[19:11:49 CET] <pepee> jasom, how do you add echo?
[19:12:25 CET] <jasom> pepee: I would use sox for this (it's a command-line audio tool)
[19:12:33 CET] <pepee> ah, yeah
[19:12:42 CET] <jasom> pepee: if you prefer a gui, something like audacity can do it too (as well as convert to u-law and back)
[19:16:39 CET] <furq> pepee: https://ffmpeg.org/ffmpeg-filters.html#aecho
[19:16:56 CET] <pepee> yeah, was reading exactly that
[19:45:42 CET] <pepee> if I try applying "-filter_complex aecho=0.8:0.9:500:0.1" to both outputs, I get: "Cannot find a matching stream for unlabeled input pad 0 on filter Parsed_aecho_0"
[19:48:35 CET] <durandal_1707> you dont need filter_complex
[19:49:02 CET] <durandal_1707> paste full command
[19:49:50 CET] <pepee> BITRATE="16k"; RATE="8000" FILE="out-bad-echo"; ffmpeg -i asd.wav -ab "$BITRATE" -ar "$RATE" -c:a libmp3lame -filter_complex aecho=0.8:0.9:500:0.1 "$FILE".mp3 -ab "$BITRATE" -ar "$RATE" -c:a libspeex -filter_complex aecho=0.8:0.9:500:0.1 "$FILE".ogg
[19:50:25 CET] <DHE> filter_complex is only needed once. sounds like you want -af instead
[19:50:59 CET] <pepee> ah, thanks
[19:52:38 CET] <pepee> yeah, s/-filter_complex/-af/ works, thanks
[19:57:07 CET] <furq> do browsers support ogg speex
[20:00:10 CET] <pepee> that's a good question, and it doesn't look like firefox/chromium do, heh.
[20:03:51 CET] <llogan> furq: why use speex instead of opus? (although I'm ignorant of any browser support)
[20:04:52 CET] <jasom> furq: no, but ffox, chrome and opera support opus
[20:05:29 CET] <furq> i was asking on pepee's behalf
[20:05:37 CET] <furq> i was going to suggest using opus instead
[20:05:45 CET] <furq> or just using vorbis
[20:06:43 CET] <pepee> hmm, the version of ffmpeg I'm using in the server doesn't support opus
[20:06:50 CET] <furq> to be honest you might as well just use mp3
[20:06:54 CET] <furq> it's supported everywhere
[20:06:54 CET] <pepee> yeah, I'll just use vorbis
[20:07:12 CET] <pepee> I was just using mp3, but some people complained
[20:07:15 CET] <furq> vorbis should be better at lower bitrates but i'm not sure it'll make a difference at such a low samplerate
[20:07:29 CET] <furq> especially if you want it to sound low quality
[20:07:46 CET] <pepee> also, the browsers seem to fetch ogg first, and it seems to have better quality
[20:07:48 CET] <furq> better than mp3, i mean
[20:08:05 CET] <pepee> well, what I wanted the most is to have similar or equal quality
[20:08:21 CET] <llogan> pepee: just download a static build of ffmpeg. most seem to support libopus
[20:08:48 CET] <pepee> tuning bit and sample rates worked
[20:09:16 CET] <pepee> llogan, huh, I didn't even think about searching for a static build
[20:09:26 CET] <furq> apparently opus has the same browser support as vorbis
[20:09:34 CET] <furq> no IE/safari for either
[20:09:46 CET] <llogan> http://ffmpeg.org/download.html
[20:09:59 CET] <furq> pepee: what OS
[20:10:11 CET] <pepee> yeah, I used binaries from some random repos :P
[20:10:22 CET] <llogan> probably ancient
[20:10:31 CET] <pepee> furq, I'm using openshift, they use redhat 6.7
[20:10:37 CET] <furq> oh fun
[20:10:55 CET] <furq> isn't there some way to get fedora repos on RHEL
[20:11:03 CET] <pepee> I actually had to download, unpack(?) the RPMs and do some LD_PRELOAD tricks
[20:11:07 CET] <furq> EPEL
[20:11:20 CET] <pepee> because they don't give complete access to the OS
[20:11:31 CET] <furq> that sounds just great
[20:11:38 CET] <pepee> it's free, so I don't complain much :P
[20:11:43 CET] <furq> well yeah the static builds will probably be the least hassle
[20:12:22 CET] <pepee> I'm still adding stuff and deciding what/how to do things. I'll move to ubuntu in the future
[20:12:39 CET] <furq> you mean debian, right
[20:13:09 CET] <pepee> well, whatever distro makes it easier to do things
[20:13:26 CET] <pepee> could be debian, yeah. no need for any specific distro tbh
[20:33:10 CET] <pepee> btw, this is the site: http://www.nask.co/
[21:33:08 CET] <Soni> why do so many things run so much faster when your input and output are both a bunch of split .png files?
[21:34:36 CET] <llogan> compared to ...?
[21:35:44 CET] <Soni> llogan, mp4 to png: 400-500 FPS, png to png: 1000+ FPS
[21:35:57 CET] <Soni> and the filters run much faster on the latter
[21:36:34 CET] <Soni> (0.5 FPS vs 70 FPS depending on the filter)
[21:38:16 CET] <BtbN> the filters don't care at all, they never see encoded data.
[21:39:15 CET] <Soni> yes but some filters run much faster if you split the video first
[21:39:21 CET] <Soni> and you won't spend hours waiting for it to do stuff
[21:39:39 CET] <c_14> yuv rgb conversion?
[21:39:42 CET] <BtbN> No, they allways do the exact same thing.
[21:40:11 CET] <BtbN> Decoding a bunch of png images propably is just lighter on the CPU than a h264 video
[21:40:46 CET] <DHE> decoding h264 is actually pretty CPU intensive, relatively speaking
[21:41:21 CET] <llogan> you can compare with "ffmpeg -i input -f null -". prepend with "time" command if preferred.
[21:42:08 CET] <Soni> I can understand png to png being twice as fast but 140x faster?!
[21:42:49 CET] <Soni> also isn't h264 hardware accelerated anyway?
[21:43:00 CET] <BtbN> no.
[22:17:53 CET] <prelude2004c> hey guys.... question.. i am using ffmpeg to encode using the GPU.. ffmpeg seems to always use about 130% of a CPU .. i doubled the speed of the CPU cores on a new system and still 130%.. then i put it on a weaker system. .and still 130% ( give or take ).. why does ffmpeg use so much CPU resources and is there some way to minimize this process ?
[22:19:12 CET] <prelude2004c> also i have a 24 core system running pciexpress 2.0 and the card is the m4000.. for some reason i start to loose fps soon as i go over 2 3 streams.. the weird part is.. the pci card is only pushing like 500MB/s on a 16x 2.0 which is nowhere near capacity
[22:19:13 CET] <fritsch> sounds like you are polling on one core
[22:19:23 CET] <prelude2004c> no i have threads set to 0
[22:19:33 CET] <fritsch> sounds like you are polling on one core
[22:19:34 CET] <fritsch> :p
[22:19:44 CET] <fritsch> one loop to fill the gpu?
[22:20:01 CET] <prelude2004c> one loop ? what do you mean ?
[22:20:43 CET] <prelude2004c> i do ffmpeg -i and then i have 2 ooutputs for HLS ( 2Mbit/s & 1Mbit/s ) output... so not very large.. 720p and like 480p
[22:21:04 CET] <prelude2004c> i am soooo lost.. been at this for days banging my head on this stuff
[22:21:05 CET] <prelude2004c> i don't get it
[22:22:08 CET] <prelude2004c> here is what blows me away... using the old nvidia cards with cuda encoding .. i can do like 8 channels without a problem and use up the cpus.. using the new m4000 card.. i can't do more than 2 channels before it dies off :(
[22:22:26 CET] <prelude2004c> same hardware
[22:22:41 CET] <prelude2004c> can anyone point me in the right direction.. i want to find a solution but don't know where to look
[22:29:17 CET] <JoshX> I'm trying to combine 4 rtsp streams into a 2x2 matrix but i'm having a hard time getting it to work properly
[22:29:25 CET] <JoshX> anyone have an idea?
[22:32:30 CET] <c_14> JoshX: https://trac.ffmpeg.org/wiki/FilteringGuide#multipleinputoverlayin2x2grid
[22:39:02 CET] <JoshX> c_14: i did see that but i dont want any 'effects'
[22:39:16 CET] <JoshX> is just want 4 inputs without any fancy filters in a 2x2 matrix
[22:39:18 CET] <c_14> >multipleinputoverlayin2x2grid
[22:39:39 CET] <c_14> It literally describes exactly what you want there, just leave out all the non-overlay filters
[22:39:47 CET] <JoshX> yes but how to 'not' use effects like negate and
[22:39:55 CET] <c_14> Just don't use them
[22:40:05 CET] <JoshX> i dont know how.. the filter syntax is very confusing to me
[22:41:06 CET] <c_14> Take the first input, pad it until it's twice as long and twice as wide, then successively add 3 overlays offsetting one by the width, one by the height and one by both
[22:42:46 CET] <c_14> Look at the documentation for the pad filter, and the overlay filters.
[22:43:46 CET] <JoshX> [AVFilterGraph @ 0x24e3760] No such filter: ''
[22:44:37 CET] <JoshX> http://pastebin.com/vJBvYE99
[22:45:23 CET] <JoshX> c_14: what should I change??
[22:45:55 CET] <c_14> replace [b] [c] and [d] with [1:v] [2:v] and [3:v] respectively and then delete lines 2-4
[22:48:14 CET] <JoshX> Invalid file index 2 in filtergraph description [0:v]pad=iw*2:ih*2[a];  [a][1:v]overlay=w[x];  [x][2:v]overlay=0:h[y];  [y][3:v]overlay=w:h[out].
[22:48:25 CET] <JoshX> i must be a complete idiot
[22:50:06 CET] <JoshX> can't you just change the pastebin?
[22:50:11 CET] <JoshX> please?
[23:01:36 CET] <llogan> or use hstack and vstack
[23:12:04 CET] <JoshX> llogan: any examples on how?
[23:13:36 CET] <llogan> ffmpeg -i in0 -i in1 -i in2 -i in3 -filter_complex "[0:v][1:v]hstack[top];[2:v][3:v]hstack[bot];[top][bot]vstack"
[23:13:50 CET] <llogan> or something like that. see http://ffmpeg.org/ffmpeg-filters.html#hstack & http://ffmpeg.org/ffmpeg-filters.html#vstack
[23:21:32 CET] <darktoad> anyone knows why ''ffmpeg -i test.mkv -c:a libvo_aacenc -b:a 96k -ac 2 -ar 44100 -c:v copy -f flv'' works until the fps keeps dropping?
[23:21:54 CET] <darktoad> my CPU is not even at 20%
[23:24:12 CET] <JoshX> llogan: when i output to a file it works great
[23:24:19 CET] <JoshX> but when i stream it to ffserver i get
[23:24:52 CET] <JoshX> Mon Nov 16 23:24:38 2015 Feed '/tmp/input1.ffm' stream number does not match registered feed
[23:25:11 CET] <JoshX> should i put it through one more filter to make it 1 stream again?
[23:25:18 CET] <JoshX> or what could this be?
[23:30:48 CET] <JoshX> does anyone have an idea what
[23:30:49 CET] <JoshX> Mon Nov 16 23:30:23 2015 Feed '/tmp/input1.ffm' stream number does not match registered feed
[23:30:53 CET] <JoshX> can mean?
[23:36:39 CET] <llogan> JoshX: i don't know. i've never used ffserver
[23:43:18 CET] <JoshX> I'm almost lost.. i need a simple solution to stream a rtsp stream to a webm/vpx stream
[23:43:34 CET] <JoshX> and it seems that it is just not available or workable
[00:00:00 CET] --- Tue Nov 17 2015