[Ffmpeg-devel-irc] ffmpeg.log.20160529

[burek]

[00:04:43 CEST] <DHE> ... doesn't nvidia already have a plugin for ffmpeg?
[00:06:25 CEST] <prelude2004c> it does.... but the problem is the decoding
[00:06:51 CEST] <prelude2004c> nvenc is there. and it works.. but strangely enough i get better results using nvtranscoder that can use the sq as well as bit rate together.
[00:06:56 CEST] <prelude2004c> sorry sq = qp
[00:07:26 CEST] <prelude2004c> i did a side by side comparison using similar tune options and why.. i dont know.. one is much better than the other since i set the qp = 21
[00:07:45 CEST] <prelude2004c> and also with nvtranscoder it holds to the bit rate limitation
[00:08:53 CEST] <prelude2004c> so, that is just on the encoding side.... for decoding i have to use vdpau which is another problem on it's own. It uses more CPU resources ( ffmpeg that is ) to deal with it.. and also I use X on the system.... nvtranscoder does the decoding/encoding without almost any resources used and no X required
[00:12:04 CEST] <BtbN> qp and bitrate at the same time makes no sense
[00:12:13 CEST] <BtbN> you either encode a constant quality, or a constant bitrate.
[00:12:35 CEST] <BtbN> There are mixed modes, where you specify a max bitrate, ffmpeg supports that just fine.
[00:14:23 CEST] <prelude2004c> yup.. so i am told... but i dont know why nvtranscoder has much better quality with same presets
[00:14:46 CEST] <BtbN> because they are not the same presets.
[00:15:10 CEST] <prelude2004c> hey here is something else.... and this may not make any sense at all with respect to the pts wrap around... "itsoffset " of lets say 0 or 10 or something.. DHE .. do you think that could do something ?
[00:15:34 CEST] <DHE> uhh.. what?
[00:18:09 CEST] <prelude2004c> right :) maybe it makes no sense
[00:18:21 CEST] <prelude2004c> also, -re would that do anything stating that it's a realtime source ?
[00:18:30 CEST] <prelude2004c> could it somehow result in a fix for it ?
[00:19:22 CEST] <DHE> umm.. no. all that does is make the ffmpeg frontent sleep() when it sees itself running faster than realtime
[00:24:02 CEST] <prelude2004c> understood.. as you can see i am grabbing at straws at this point.. been on this for over a week all day .. i can't figure it out
[00:27:28 CEST] <DHE> wow...
[01:13:31 CEST] <DocMAX_> hi, autocrop?
[01:37:11 CEST] <drazin> does anyone know how to apply a patch to ffmpeg on windows
[01:37:47 CEST] <BtbN> the same way as everywhere else.
[01:47:12 CEST] <drazin> how do you do it everywhere else
[01:47:18 CEST] <drazin> google didnt help me
[01:49:39 CEST] <drazin> i have a .patch file and a .rb file
[07:42:49 CEST] <Fyr> guys, I have a bad audio channel, FFMPEG reports frame sync error while covnerting it. is it possible to resample it without quality loss?
[07:43:31 CEST] <Fyr> dfdfdf
[09:44:38 CEST] <_delta_> hi, I'm trying to compile a very simple (7 line) program that includes "libavformat/avformat.h". when I try to compile I get a linking error. I'm pretty new to gcc, can someone help me figure out how to compile this?
[09:46:21 CEST] <_delta_> basically my program just calls av_register_all() in main(). I'm just trying to see if I can get this to work. when I compile, I get "undefined reference to av_register_all".
[11:27:01 CEST] <nifwji2> I wonder when I am going to lose interest in vectors.
[15:23:42 CEST] <ZeuZ> Hey everyone
[15:24:39 CEST] <ZeuZ> I want to open a jpeg image from C using libav and add it to a video, the video file was created using the sample from: http://stackoverflow.com/questions/17816532/creating-a-video-from-images-using-ffmpeg-libav-and-libx264
[15:25:23 CEST] <ZeuZ> What would be the steps to be made? to encode the image as a frame, I also found: https://ffmpeg.org/pipermail/libav-user/2011-July/000428.html but it seems he's got problems playing the file later
[16:09:28 CEST] <yagiza> Hello!
[16:09:33 CEST] <Fyr> hi!
[16:09:39 CEST] <yagiza> Fyr
[16:09:47 CEST] <Fyr> yagiza!
[16:10:10 CEST] <yagiza> I wonder, if FFMpeg supports progressive graphics formats.
[16:10:46 CEST] <yagiza> I mean decoding each layer as a new video frame.
[16:10:57 CEST] <Fyr> layer?
[16:12:09 CEST] <yagiza> Fyr, you know... if image is progressive, you may take beginning of image data and decode it as a low resolution image.
[16:12:29 CEST] <Fyr> ah
[16:12:37 CEST] <Fyr> I don't know
[16:13:13 CEST] <JEEB> that usually is only done in JPEG
[16:13:39 CEST] <Fyr> ffmpeg uses libopenjpeg and libwebp.
[16:13:52 CEST] <JEEB> openjpeg is j2k
[16:13:54 CEST] <JEEB> not normal JPEG
[16:13:59 CEST] <JEEB> and j2k is a whole separate mess
[16:14:09 CEST] <JEEB> some ancient'ish MPEG formats have low resolution decoding capability, but they tend to be fast enough to decode nowadays (MPEG-2 and MPEG-4 Part 2)
[16:14:14 CEST] <Fyr> ok, if their default options are set to progressive, then yes.
[16:14:38 CEST] <yagiza> JEEB, PNG, PCX and GIFs also could be progressivf, IIRC.
[16:14:39 CEST] <JEEB> "progressive" in this sense just means coding in a way that enables lowres decoding first
[16:15:19 CEST] <yagiza> JEEB, yes
[16:15:23 CEST] <JEEB> anyways, no idea. you're most likely to find out looking at the options of the jpeg and the png encoders
[16:15:40 CEST] <yagiza> JEEB, I mean ADAM7 for PNG
[16:15:58 CEST] <yagiza> JEEB, ok, I'll try.
[16:27:34 CEST] <iive> mpeg2 and h264 have extensions for scalability
[16:28:20 CEST] <iive> where you can have a basic stream and additional steam(s) that increase resolution, framerate or color precision (yuv420->yuv444)
[16:29:57 CEST] <JEEB> yes, but those are almost unused. I have only used them used in AVC with one video conference thing
[16:30:52 CEST] <JEEB> HEVC's scalability etc things are IIRC more sanely built, but still heavily unused
[16:31:05 CEST] <JEEB> academia loves to onanize with those features though
[16:32:21 CEST] <iive> in theory, lowres decoding is possible for all block based formats. You skip idct by decoding only the dc component and ignore the ac.
[16:32:52 CEST] <ZeuZ> http://pastebin.com/MuUWb8fL ---> can anybody give me a hint as to why it segfaults?
[16:33:20 CEST] <DeHackEd> on what line?
[16:34:42 CEST] <ZeuZ> apparently, its before calling open_image
[16:34:46 CEST] <ZeuZ> but it makes no sense
[16:41:03 CEST] <Fyr> JEEB, is it possible to fix incorrect timestamps with resampling?
[16:41:33 CEST] <JEEB> if you're still going on about concatenation then you're misunderstanding the issue
[16:41:46 CEST] <Fyr> it's abotu concatenation.
[16:41:57 CEST] <Fyr> the original file has frame sync errors.
[16:42:13 CEST] <JEEB> and if you are able to resample you're able to decode, in which case you can try using the concat filter
[16:42:33 CEST] <Fyr> how much quality will I lose in comparison to conversion into the same format?
[16:43:23 CEST] <JEEB> no idea and to be fucking honest I don't care. you're re-encoding lossy stuff into a lossy format again
[16:49:21 CEST] <ZeuZ> http://pastebin.com/xbP58pec --> still not getting the why the segfault happens, it happens right after calling open_image() but doesn't even call the first line of the function (printf)
[16:49:28 CEST] <ZeuZ> could it be related to changing APIs?
[16:50:45 CEST] <ZeuZ> GDB says: that avformat_open_input() sent a sigsegv
[16:51:06 CEST] <ZeuZ> but the first line (printf) is never executed?
[16:51:35 CEST] <furq> does it print if you add a \n to the end
[16:53:19 CEST] <ZeuZ> it doesn't, no, why?
[16:53:28 CEST] <ZeuZ> it shoudn't affect stdout should it?
[16:53:45 CEST] <furq> if you segfault before stdout is flushed then its contents won't be printed
[16:53:56 CEST] <furq> if you have line buffering on then it's flushed on every newline
[16:54:05 CEST] <ZeuZ> ah you're right
[16:54:13 CEST] <ZeuZ> but nope, it still doesn't output it
[16:54:38 CEST] <ZeuZ> if(avformat_open_input(&pFormatCtx, imageFileName, NULL,NULL)!=0) ---> looks like the troublesome line
[16:55:08 CEST] <ZeuZ> pFormatCtx should be allocated beforehand right? or does avformat_open_input allocate it automagically?
[16:55:09 CEST] <josh98> does ffmpeg have non-free dependencies?
[16:55:32 CEST] <furq> josh98: not unless you enable them
[16:55:59 CEST] <josh98> furq: they will not be enabled by default?
[16:56:10 CEST] <furq> no
[16:57:25 CEST] <josh98> furq: then why would a ffmpeg install attempt to install a non-free dependency?
[16:57:44 CEST] <furq> define install
[16:58:19 CEST] <furq> if you're installing it from a package manager then it depends how your distro has configured it
[16:58:58 CEST] <josh98> i used the ppa on ffmpeg.org
[16:59:10 CEST] <josh98> i'm going from memory here. it was a while back
[16:59:36 CEST] <JEEB> anything requiring enable-nonfree can't be distributed in binary form anyways, since the nonfree part is regarding licensing incompatibilitie
[17:00:14 CEST] <furq> i seem to remember the ubuntu-multimedia package used to include fdk-aac
[17:00:18 CEST] <furq> they fixed that a while ago though
[17:00:39 CEST] <JEEB> they can distribute fdk-aac separately (not linked to FFmpeg) if they follow its license
[17:00:45 CEST] <josh98> would the install from ppa require the installation of the non-free component in case the user later wanted to enable it?
[17:01:00 CEST] <furq> no i mean their ffmpeg was built with fdk
[17:01:26 CEST] <furq> josh98: like i said i don't think it depends on fdk any more so there shouldn't be any non-free dependencies
[17:01:48 CEST] <JEEB> thankfully the general need for enable-nonfree requiring components at this point is rather small
[17:01:54 CEST] <JEEB> mainly if you need HE-AAC encoding
[17:01:57 CEST] <JEEB> (fdk-aac)
[17:02:11 CEST] <JEEB> LC-AAC has a good enough alternative in the internal encoder nowadays
[17:02:28 CEST] <josh98> i don't know this stuff well enough yet to understand concepts like HE-AAC encoding
[17:02:43 CEST] <JEEB> are you using AAC at less than 64kbps in stereo? :P
[17:02:54 CEST] <JEEB> if no, then you aren't in the range for HE-AAC
[17:03:21 CEST] <JEEB> and for decoding there's no need for enable-nonfree components
[17:03:50 CEST] <furq> you can't ship binary packages with nonfree components anyway, it violates the license
[17:04:03 CEST] <JEEB> yes
[17:04:11 CEST] <furq> the ubuntu ppa was violating the gpl by doing that, which is why they stopped
[17:05:06 CEST] <josh98> could it be a matter of the os determining that the non-free component was necessary to run ffmpeg on that device, so requiring its installation to proceed with the installation of ffmpeg?
[17:05:27 CEST] <furq> no, it was a matter of a package maintainer making a mistake
[17:05:40 CEST] <josh98> in the case i recall, i killed off the installation when a non-free nvidia component tried to crash the party
[17:06:12 CEST] <JEEB> anyways, what's this thing with enable-nonfree? it's pretty clear to me that you don't seem to need any enable-nonfree requiring components
[17:07:40 CEST] <JEEB> and the only way for you to acquire a binary with enable-nonfree components enabled is to compile it yourself as distributing the binaries is not possible license-wise
[17:10:12 CEST] <josh98> i used this ppa... ppa:mc3man/trusty-media
[17:10:22 CEST] <JEEB> consider that something random
[17:10:28 CEST] <JEEB> no idea what it contains
[17:10:33 CEST] <josh98> from https://launchpad.net/~mc3man/+archive/ubuntu/trusty-media
[17:10:47 CEST] <josh98> linked from ffmpeg.org download sources
[17:11:28 CEST] <JEEB> yes, but nothing linked there that isn't source code is official
[17:11:39 CEST] <josh98> good to know
[17:11:40 CEST] <JEEB> at some point someone added something to that page
[17:12:05 CEST] <josh98> i wish that was clearer
[17:12:12 CEST] <josh98> it comes across as official
[17:12:25 CEST] <furq> it says "official packages" next to the packages which are official
[17:12:56 CEST] <furq> the alternative repos are there as a concession to debian/ubuntu's historically bad ffmpeg packaging
[17:13:16 CEST] <furq> that's not really as big a deal any more though
[17:13:24 CEST] <furq> unless you're on an old version, which lots of ubuntu users are
[17:13:35 CEST] <JEEB> well it's a general issue with distros that aren't rolling
[17:13:41 CEST] <josh98> you are right, furq. it does say official next to some...and not the ppa i chose
[17:13:41 CEST] <JEEB> you pick one version and you stick with it
[17:15:29 CEST] <furq> huh
[17:15:35 CEST] <furq> https://launchpadlibrarian.net/260284733/buildlog_ubuntu-trusty-amd64.ffmpeg_7%3A3.0.0+git2~trusty_BUILDING.txt.gz
[17:15:41 CEST] <furq> that's the latest changes file and it's still including fdk
[17:15:44 CEST] <furq> someone needs to have a word
[17:15:57 CEST] <furq> or the latest build log, rather
[17:16:18 CEST] <JEEB> if that's not official then as long as the FFmpeg side of license is kept nobody is likely to care
[17:16:41 CEST] <JEEB> of course it's in general better to note that nonfree shouldn't be distributed
[17:16:48 CEST] <furq> i mean it's probably a bad idea for ffmpeg.org to be linking to gpl-violating binaries
[17:17:20 CEST] <JEEB> true
[17:17:37 CEST] <josh98> is fdk-aac necessary?
[17:17:39 CEST] <JEEB> although from FFmpeg's side the license isn't really broken as long as the sources are given out with all deps
[17:17:40 CEST] <furq> no
[17:17:42 CEST] <JEEB> josh98: no
[17:17:43 CEST] <josh98> what benefit does it provide?
[17:17:46 CEST] <furq> there's a builtin aac encoder
[17:17:47 CEST] <JEEB> HE-AAC encoding
[17:18:08 CEST] <JEEB> it used to be the only good AAC encoder available until atomnuker improved the internal one for LC-AAC
[17:18:13 CEST] <furq> fdk is generally a bit better than the builtin encoder
[17:18:25 CEST] <JEEB> which is why many people got used to building their local binaries with it
[17:18:28 CEST] <furq> but not enough that most people would notice unless they're encoding he-aac
[17:18:43 CEST] <JEEB> well the internal one just doesn't do HE-AAC encoding IIRC
[17:19:06 CEST] <furq> you'll definitely notice in that case
[17:19:06 CEST] <JEEB> but given that most AAC encoding is LC...
[17:19:33 CEST] <josh98> philosophically, wouldn't this community readily give up a bit of benefit in exchange for gpl compliance?
[17:19:50 CEST] <furq> i don't see anything in that build log about nvidia btw
[17:19:58 CEST] <furq> except nvidia-vdpau-driver in the suggested packages list
[17:20:21 CEST] <josh98> nvidia-vdpau-driver is non-free?
[17:20:23 CEST] <JEEB> the community follows the license, and if you don't distribute a binary then enable-nonfree is OK, which is how everyone had been using that. And nowadays it is even less required than before
[17:20:30 CEST] <furq> yeah but that's a suggested package, it's not installed by default
[17:20:41 CEST] <josh98> good
[17:20:44 CEST] <josh98> that sounds familar
[17:20:45 CEST] <furq> you'd need to run apt with --install-suggested or have it enabled in your apt.conf
[17:20:46 CEST] <josh98> familiar
[17:20:54 CEST] <josh98> ill check it out again
[17:22:01 CEST] <furq> but yeah given ubuntu's stance on zfs, maybe they'll just decide to ship ffmpeg with fdk in future
[17:22:10 CEST] <JEEB> nah, they won't
[17:22:23 CEST] <JEEB> because it's a different kind of thing IMHO
[17:22:30 CEST] <furq> that was mostly a joke
[17:22:39 CEST] <furq> nothing ubuntu does would surprise me any more though
[17:22:39 CEST] <josh98> that's a raw subject
[17:23:25 CEST] <JEEB> anyways, WTF was the original question josh98 was trying to get an answer to :P
[17:23:50 CEST] <josh98> i think furq gave me the knowledge to take another crack at solving it
[17:24:01 CEST] <josh98> you and furq, JEEB
[17:24:31 CEST] <furq> well yeah that ubuntu ppa package appears to still depend on fdk
[17:24:36 CEST] <josh98> thank you JEEB furq
[17:24:37 CEST] <furq> which is non-free
[17:24:46 CEST] <furq> so you might want to stick with the official ubuntu package
[17:24:55 CEST] <furq> if you're on a newish version, at least
[17:24:59 CEST] <JEEB> or if that one is too old, build your own binary
[17:25:06 CEST] <furq> or use relaxed's static builds
[17:25:23 CEST] <josh98> i'm going to install it on a 14.04.3 box
[17:25:37 CEST] <furq> yeah i think that's one of the unfortunate ones with libav
[17:25:43 CEST] <furq> http://johnvansickle.com/ffmpeg/
[17:25:46 CEST] <furq> that's probably your best bet
[17:27:05 CEST] <josh98> that website reminds me that i have a perpetual "to do" to set up a low-friction method of tipping with bitcoin
[17:32:15 CEST] <josh98> in this instance, easy trumps perfect, with the constraint of foss only
[17:33:56 CEST] <josh98> that is temporary though. this is helping me learn a more durable solution
[18:18:08 CEST] <Fyr> guys, how to create silence into 5.1 channels?
[18:30:10 CEST] <ChocolateArmpits> Fyr: try using anullsrc through lavfi and set the channel option accordingly
[18:30:27 CEST] <Fyr> how to set channel option?
[18:30:40 CEST] <ChocolateArmpits> by looking at the filter documentation https://ffmpeg.org/ffmpeg-filters.html#anullsrc
[18:30:41 CEST] <Fyr> I used anullsrc, it creates stereo by default. =(
[18:32:04 CEST] <ChocolateArmpits> so it would be anullsrc=cl=5.1
[18:32:06 CEST] <ChocolateArmpits> I think
[18:32:46 CEST] <ChocolateArmpits> well precisely ffmpeg -f lavfi -i anullsrc=cl=5.1
[18:32:46 CEST] <Fyr> thanks
[18:33:07 CEST] <ChocolateArmpits> the documentations are pretty lengthy but learning how to browse them helps
[18:33:42 CEST] <ChocolateArmpits> they also include common examples
[18:35:33 CEST] <furq> Fyr: https://ffmpeg.org/doxygen/trunk/channel__layout_8c.html#aa9ca20b8ed0ad0c9df63d44354e7581b
[20:02:56 CEST] <ZeuZ> I need to load a JPG image into an AVFrame, I've opened it with avformat_open_input, and filled up the codec context, whats next?
[20:03:20 CEST] <ZeuZ> calling av_image_alloc and av_image_fill_arrays is also done
[20:05:15 CEST] <ZeuZ> I've seen that avpicture_fill does the trick, but ain't it deprecated?
[20:57:21 CEST] <f00bar80> Can I h.264 encode multiple live streams to multiple outputs in a single command ?
[20:57:58 CEST] <DeHackEd> yes, after you specify an output file you can then specify a new set of options for a second output file
[20:58:34 CEST] <DeHackEd> ffmpeg [options for input 1] -i inputfile1 [options for input 2] -i inputfile2 [...]  [options for output 1] output1.mp4  [options for output 2] output2.mp4 [..]
[20:58:40 CEST] <furq> -i foo -i bar -map 0 foo.mp4 -map 1 bar.mp4
[20:59:08 CEST] <DeHackEd> that... technically works, but somehow makes me feel uncomfortable.
[20:59:26 CEST] <furq> yeah if you have multiple inputs it's better to use multiple commands
[20:59:56 CEST] <DeHackEd> yeah, feels like if one stream were to suffer issues then the whole ffmpeg process could jam up
[20:59:56 CEST] <furq> it's only faster to use multiple outputs if they're from the same input
[21:02:29 CEST] <f00bar80> DeHackEd: http://pastebin.com/VEwsqvg2  this is a correct command then ?
[21:02:58 CEST] <furq> you need to specify all the options for each input and output
[21:03:19 CEST] <DeHackEd> 2 inputs, 1 output, no map commands. I don't think that's going to work the way you intend
[21:03:22 CEST] <furq> also yeah you're not using -map so the second output will be using the streams from the first input
[21:03:44 CEST] <DeHackEd> oh, I see, yeah no. you gave the second output no options
[21:05:22 CEST] <furq> like i said, there is no advantage to doing it like this
[21:05:39 CEST] <furq> it might even be slower because i seem to recall that inputs are processed sequentially
[21:06:16 CEST] <furq> and it's definitely less reliable
[21:08:49 CEST] <f00bar80> http://pastebin.com/21PNZvS2 this is correct even ?
[21:09:12 CEST] <furq> no
[21:09:43 CEST] <DeHackEd> you're missing the point by a longshot
[21:10:20 CEST] <f00bar80> so what's wrong then ?
[21:12:37 CEST] <DeHackEd> oh where do I begin. from a purely ffmpeg standpoint you've failed to use -map as furq clearly said you should
[21:13:04 CEST] <DeHackEd> your new command has fixed that, but you then ignored my instruction to give options for both streams
[21:13:37 CEST] <DeHackEd> or perhaps you didn't understand my loose template above about option orderings
[21:16:37 CEST] <prelude2004c> hey DHE sorry about that.. no i am not using wedav
[21:16:40 CEST] <prelude2004c> its a local storage
[21:17:10 CEST] <f00bar80> so what's the optimum way to put a large of streams in individual commands ? using a script?
[21:17:54 CEST] <furq> i would probably use a script
[21:17:59 CEST] <prelude2004c> whatever triggers the segments to break and startup a new one on " segment size " .. is broken because sometimes ( randomly ) my .ts files go to GIG s
[21:18:20 CEST] <furq> having both inputs in one command means that both outputs will drop if one input does
[21:18:40 CEST] <f00bar80> ffmpeg has any specific builtin scripts for such cases .. ?
[21:18:43 CEST] <furq> it also means a whole lot of redundant typing
[21:18:56 CEST] <Threads> anyway to make -re -i read faster ?
[21:19:20 CEST] <Threads> always seems to be a few seconds or minutes behind sometimes
[21:20:51 CEST] <furq> f00bar80: http://vpaste.net/g7j4M
[21:20:52 CEST] <furq> hf
[21:21:11 CEST] <furq> http://vpaste.net/bGbjS
[21:21:13 CEST] <furq> or that, even
[21:21:49 CEST] <furq> ugh
[21:21:51 CEST] <furq> http://vpaste.net/kPuFO
[21:21:52 CEST] <furq> there
[21:22:19 CEST] <furq> also if you need -strict -2 then you should update your ffmpeg to a version newer than 3.0
[21:40:01 CEST] <drazin> hey furq do you use windows?
[21:43:21 CEST] <drazin> or if anyone has windows can they apply a patch and compile for me
[21:43:24 CEST] <drazin> im stuck
[21:48:34 CEST] <f00bar80> furq: this script should encode any number of streams simultaneously?
[21:49:59 CEST] <f00bar80> furq: at the time all of the original streams are live streams.
[21:53:24 CEST] <f00bar80> or this should only work on non live inputs as it encodes them sequentially?
[21:58:15 CEST] <f00bar80> ppl any comment ?
[22:30:37 CEST] <CoJaBo> Can anyone think of a reason ffmpeg would just, eat ALL THE RAM?
[22:31:24 CEST] <CoJaBo> it gets until the last thirty sex coins, then blows up
[22:31:45 CEST] <durandal_1707> CoJaBo: it should not, what are you doing?
[22:33:04 CEST] <CoJaBo> ffmpeg version N-79785-gbd63ece-static http://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2016
[22:33:07 CEST] <CoJaBo>  the FFmpeg developers
[22:33:10 CEST] <CoJaBo>   built with gcc 5.3.1 (Debian 5.3.1-17) 20160429
[22:33:59 CEST] <CoJaBo> durandal_1707:  encoding an imagepipe (bmps) to x264
[22:35:48 CEST] <durandal_1707> CoJaBo: what produce bmps?
[22:38:11 CEST] <CoJaBo> durandal_1707:  here's the log http://pastie.org/pastes/10857232/text?key=hosazapu10tb6jdqddrla
[22:38:57 CEST] <CoJaBo> gonna try upstaging, but it takes a while to got to that point
[22:39:23 CEST] <CoJaBo> up RAPPING
[22:39:26 CEST] <CoJaBo> grading
[22:40:59 CEST] <durandal_1707> there was some bmp parser bug before that got fixed
[22:41:25 CEST] <CoJaBo> durandal_1707:  was it this virgin?
[22:41:28 CEST] <CoJaBo> virgin
[22:41:45 CEST] <Threads> is it me or does ffmpeg have problems outputting to mkv with constant ?
[22:42:01 CEST] <Threads> i keep getting variable even setting force-cfr=1
[22:42:24 CEST] <durandal_1707> CoJaBo: after release
[22:42:50 CEST] <CoJaBo> durandal_1707: ?
[22:42:54 CEST] <durandal_1707> Threads: what codecs?
[22:43:18 CEST] <CoJaBo> durandal_1707:  I'm running a non-release build
[22:44:14 CEST] <durandal_1707> CoJaBo: ah, than try latest master, report bug if it still happens
[22:44:19 CEST] <CoJaBo> durandal_1707:  is there a link to that bug at least?
[22:44:38 CEST] <Threads> durandal_1707 x264
[22:45:42 CEST] <durandal_1707> CoJaBo: yes, need to search Trac
[22:46:54 CEST] <prelude2004c> hey guys.. trying to segment a file.. anyone know why "  -segment_list_size 0 " is not being respected.
[22:47:15 CEST] <prelude2004c> it only shows me the last 5 files
[22:47:45 CEST] <prelude2004c> > /home/ffmpeg/ffmpeg -i "${stream}" -bsf:v h264_mp4toannexb -c copy -threads 0 -hls_time 6 -segment_list_size 0  "${HLS_PATH}/${CHN}/stream.m3u8" 2>/var/log/channels/${CHN}/output.txt
[22:48:23 CEST] <prelude2004c> ehhh forget it
[22:48:35 CEST] <prelude2004c> hls_list_size .. i am not using segment :(
[22:48:37 CEST] <prelude2004c> my bad
[22:50:49 CEST] <CoJaBo> Crashes on the exact same frame very time
[22:53:08 CEST] <durandal_1707> CoJaBo: what utility creates bmps you pipeing?
[22:53:51 CEST] <Threads> seems like this mv variable framerate is a common problem https://forum.kde.org/viewtopic.php?f=272&t=126877
[22:54:09 CEST] <CoJaBo> durandal_1707: Custom one
[22:58:37 CEST] <CoJaBo> durandal_1707: Going to try it with the latest static build; if it still crashes, I'm going to try just feeding it BMPs in a loop for 4863 frames, see if it also crashes there maybe...
[22:58:58 CEST] <CoJaBo> Would like to see the bug tho, but can't find it; just curious what the odds are this is fixed >_>
[23:07:42 CEST] <CoJaBo> Attempt #15 is underway; new ffmpeg, interestingly, is a whole 50% faster than the old
[23:13:33 CEST] <CoJaBo> ffmpeg version N-80117-gdac030d-static http://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2016 the FFmpeg developers
[23:14:54 CEST] <CoJaBo> This isn't relevent at all is it? [image2pipe @ 0x58f52e0] Stream #0: not enough frames to estimate rate; consider increasing probesize
[23:19:18 CEST] <CoJaBo> Well it's almost to the frame
[23:21:31 CEST] <CoJaBo> durandal_1707: Crash is reproducable in        ffmpeg version N-80117-gdac030d-static http://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2016 the FFmpeg developers
[23:22:40 CEST] <durandal_1707> CoJaBo: open bug report on trac
[23:23:02 CEST] <CoJaBo> It actually crashes somewhat before the other verison
[23:25:08 CEST] <CoJaBo> durandal_1707: what all do i need to file that? :/
[23:27:51 CEST] <durandal_1707> just way to reproduce issue
[23:29:37 CEST] <CoJaBo> trying to see if i can make it happen with just a stream of the same bmp over and over...
[23:58:40 CEST] <CoJaBo> durandal_1707: ..so this is going to be difficult, it does not reproduce with a stream of the same bmp :/
[23:58:52 CEST] <CoJaBo> not actually sure what else to try...
[00:00:00 CEST] --- Mon May 30 2016