[Ffmpeg-devel-irc] ffmpeg.log.20170428

[burek]

[00:22:46 CEST] <schelleursli> still trying to get h264_cuvid to deinterlace a 1080i file to 720p 50 fps, still failing, outputs only heavily stuttering file ffmpeg cmd:
[00:22:58 CEST] <schelleursli> ffmpeg -hwaccel cuvid -c:v h264_cuvid -deint adaptive -resize 1280x720 -i H:\input.ts -c:v h264_nvenc cuvid.mp4
[00:23:39 CEST] <schelleursli> if I add -r 50 after the resize it doesn't stutter as much but still not properly working, really don't know whats broken, any ideas?
[00:25:41 CEST] <schelleursli> https://1drv.ms/f/s!At6cEyJzE4-RjG1eMMLRqISoo2UL I've uploaded the source.ts plus the outputs and ffmpeg cmd.txt there
[00:27:06 CEST] <schelleursli> https://pastebin.com/1C1sQ6js also on pastebin
[00:28:16 CEST] <ClearSky> Recorded a game with FRAPS. When converting it to Huffyuv for editing, FFmpeg shows the following warning: Stream #0:0: Video: fraps (FPS1 / 0x31535046), yuvj420p(pc, bt709/unknown/unknown), 960x720, 79693 kb/s, 30 fps, 30 tbr, 30 tbn, 30 tbc; deprecated pixel format used, make sure you did set range correctly. Also, the recorded Fraps video looks darker when compared to in-game video. How...
[00:28:17 CEST] <ClearSky> ...do I fix this?
[00:28:36 CEST] <BtbN> you don't.
[00:28:50 CEST] <BtbN> The decoder needs to be fixed
[00:32:12 CEST] <ClearSky> BtbN: What decoder?
[00:32:19 CEST] <iH2O> sorry, I lost the line. my question was that I often get mp4 videos from youtube with only a fixed image. Id rather convert them to mp3 to save space. here's an example of a mp3 video with a fixed image https://skeptoid.com/audio/skeptoid-4554.mp3
[00:33:13 CEST] <BtbN> ClearSky, the one it's using.
[00:33:19 CEST] <iH2O> that's what I'd like to get in general by converting a mp4 video with a fixed image to mp3
[00:35:20 CEST] <BtbN> just extract like the first frame as png image or something, and pass it when creating the mp3.
[00:35:25 CEST] <BtbN> The mp3 "muxer" should do the right thing
[00:36:25 CEST] <ClearSky> Ok, I just checked the huffyuv output, and the game doesn't look dark anymore, it looks like it should. Can anyone explain what happened between converting dark-looking input and normal-looking output video?
[00:38:23 CEST] <ClearSky> There is a following message in FFmpeg: Stream #0:0: Video: huffyuv (HFYU / 0x55594648), yuv422p, 960x720, q=2-31
[00:39:13 CEST] <kerio> iH2O: please don't download music from youtube
[00:39:32 CEST] <furq> why would you convert aac to mp3 to save space
[00:39:35 CEST] <kerio> and if you really must, don't reencode it at least
[00:40:28 CEST] <thebombzen> schelleursli: I find this interesting
[00:40:31 CEST] <thebombzen> dup=451 drop=468
[00:41:09 CEST] <kerio> ClearSky: ffvhuff is better than huffyuv, fwiw
[00:41:16 CEST] <thebombzen> iH2O: if the audio is aac, you can just do ffmpeg -i input.mp4 -map a -c copy out.m4a
[00:41:19 CEST] <schelleursli> I can get rid of some errors by adding -r 50 like so https://pastebin.com/0vBdDNy4 at thebombzen, but it still isn't properly deinterlacing it
[00:41:39 CEST] <furq> thebombzen: if he's converting it from fraps to something else then whatever editing program he's using isn't using ffmpeg
[00:41:42 CEST] <furq> so ffvhuff probably won't work
[00:41:48 CEST] <thebombzen> hm?
[00:42:09 CEST] <thebombzen> furq: I'm confused now
[00:42:09 CEST] <kerio> is huffyuv supported in non-ffmpeg-y editors?
[00:42:11 CEST] <furq> er
[00:42:15 CEST] <furq> s/thebombzen/kerio
[00:42:18 CEST] <thebombzen> ah okay
[00:42:22 CEST] <furq> huffyuv is widely supported
[00:42:24 CEST] <furq> ffvhuff not so much
[00:42:42 CEST] <thebombzen> schelleursli: what do you mean by "not properly deinterlacing it"
[00:42:42 CEST] <kerio> i thought they were both kind of made up by ffmpeg
[00:42:47 CEST] <furq> no huffyuv is ancient
[00:42:50 CEST] <kerio> the moar i know
[00:42:58 CEST] <furq> circa 2000 or so
[00:43:00 CEST] <thebombzen> huffyuv is an old thing and ffvhuff is "ffmpeg's huffyuv variant"
[00:43:06 CEST] <thebombzen> but why not use ffv1
[00:43:13 CEST] <kerio> thebombzen: because ffv1 is SLOW AS BALLZ
[00:43:16 CEST] <furq> ffvhuff is really fast
[00:43:20 CEST] <furq> and also compresses better than huffyuv
[00:43:22 CEST] <schelleursli> thebombzen, hard to explain, the picture does slightly stutter every couple of frames, you can check the output mp4 files on the onedrive link
[00:43:39 CEST] <thebombzen> did you follow it up with -vf decimate?
[00:43:55 CEST] <thebombzen> also do you have 50i interlaced from 25p original?
[00:43:59 CEST] <thebombzen> what was the source before interlacing?
[00:44:18 CEST] <schelleursli> source is a 1080i recording from my dvb-c tv signal
[00:44:39 CEST] <schelleursli> I'm trying to deinterlace it to 720p 50 fps with cuvid
[00:44:47 CEST] <thebombzen> there's like five files there
[00:44:51 CEST] <thebombzen> can you upload to 0x0.st or something
[00:44:54 CEST] <schelleursli> which should work in theory if I read the helpfile correctly
[00:45:00 CEST] <thebombzen> and you didn't answer my question
[00:45:08 CEST] <schelleursli> -vf decimate?
[00:45:11 CEST] <thebombzen> what was the source? well it's 1080i sure
[00:45:14 CEST] <thebombzen> but framerate
[00:45:15 CEST] <furq> it's 1080i50
[00:45:15 CEST] <schelleursli> ah
[00:45:22 CEST] <furq> dvb-c is 50fps
[00:45:29 CEST] <schelleursli> actually
[00:45:38 CEST] <schelleursli> dvb-c is 1080i 25 fps usually
[00:45:41 CEST] <thebombzen> then you probably don't need to decimate
[00:45:41 CEST] <schelleursli> or 720p 50 fps
[00:45:45 CEST] <furq> 1080i50 is 50 fields per second
[00:45:45 CEST] <schelleursli> as far as I know
[00:46:02 CEST] <thebombzen> btw your onedrive link has like 5 videos in it
[00:46:28 CEST] <schelleursli> yes sorry i've just uploaded all of the outputs with different commandlines :x
[00:46:38 CEST] <schelleursli> sec
[00:46:52 CEST] <thebombzen> is there any particular reason you need to decode and deinterlace it with cuvid rather than with a software deinterlacer?
[00:47:04 CEST] <furq> nobody needs to do that, but i assume it's much faster
[00:47:40 CEST] <schelleursli> yes, I was trying to save some cpu time so I can do x264 encoding
[00:47:47 CEST] <schelleursli> cpu time is rather limited
[00:48:13 CEST] <furq> tbh i'd rather spend the time on a better deinterlacer and use a faster x264 preset
[00:48:22 CEST] <furq> but maybe cuvid's deinterlacer is good
[00:48:36 CEST] <thebombzen> I don't know, I've never used it
[00:48:44 CEST] <BtbN> the cuvid deinterlacer is more than decent
[00:48:47 CEST] <furq> fair enough
[00:48:47 CEST] <BtbN> but... problematic to use
[00:48:54 CEST] <thebombzen> problematic, as above?
[00:48:57 CEST] <schelleursli> re onedrive, looks like it goes into thumbnail mode by default, you can switch to list view, I've named the files properly
[00:49:11 CEST] <BtbN> you basically need to pass its output framerate as input option, as ffmpeg.c can't handle a decoder dictating the framerate
[00:49:29 CEST] <BtbN> so for i50 content, you need to pass -r 50 as input option.
[00:49:37 CEST] <schelleursli> thats what I've tried BtbN
[00:49:38 CEST] <BtbN> otherwise you end up with every second field dropped
[00:50:00 CEST] <schelleursli> didn't help much either, it's better then with no -r option at all but still stutters
[00:50:13 CEST] <BtbN> ... stutters?
[00:50:17 CEST] <BtbN> did you use bob or something?
[00:50:44 CEST] <schelleursli> I've tried all 3 deint options which are weave (25 fps horrible) bob and adaptive
[00:50:46 CEST] <dystopia_>  [23:44:40] <schelleursli> I'm trying to deinterlace it to 720p 50 fps with cuvid
[00:50:50 CEST] <dystopia_> use yadif
[00:50:50 CEST] <schelleursli> bob and adaptive produce the same problems
[00:50:53 CEST] <BtbN> weave is the off-option.
[00:50:57 CEST] <schelleursli> ah
[00:50:58 CEST] <dystopia_> yadif=1:0
[00:51:01 CEST] <schelleursli> that explains it :D
[00:51:01 CEST] <dystopia_> bob
[00:51:08 CEST] <dystopia_> yadif=0:0
[00:51:08 CEST] <BtbN> yadif is a software deinterlacer
[00:51:11 CEST] <BtbN> cuvid uses the GPU
[00:51:12 CEST] <dystopia_> weave
[00:51:27 CEST] <schelleursli> dystopia_ I am trying to use the gpu, yadif deinterlace works nicely but I'm on the limit with cpu time
[00:51:38 CEST] <furq> isn't bwdif recommended over yadif now
[00:51:38 CEST] <thebombzen> it appears you have a weird frame inserted for no reason
[00:51:45 CEST] <thebombzen> furq: it's more cpu taxing
[00:51:51 CEST] <BtbN> ./ffmpeg.exe -deint adaptive -r 50 -c:v h264_cuvid -i something.mkv ....
[00:51:56 CEST] <BtbN> works flawlessly for me
[00:52:01 CEST] <furq> sure, but yadif sucks so that's to be expected
[00:52:28 CEST] <furq> good things are often slow
[00:52:29 CEST] <furq> except libvpx
[00:52:58 CEST] <iH2O> kerio, why do you say not to download music from youtube
[00:52:59 CEST] <schelleursli> mh, that is strange I get this https://pastebin.com/0vBdDNy4 with -r50 at BtbN, looks fine on ffmpeg side but when I check the file it isn't
[00:53:04 CEST] <furq> iH2O: because the quality sucks
[00:53:06 CEST] <schelleursli> sec I'll upload the samble
[00:53:39 CEST] <iH2O> hmm
[00:54:04 CEST] <furq> it's not like it's difficult to illegally acquire decent quality music on the internet
[00:54:12 CEST] <furq> at least, so i've heard
[00:54:15 CEST] <furq> i'm very law-abiding
[00:54:16 CEST] <schelleursli> https://1drv.ms/v/s!At6cEyJzE4-RjG52Hs6hLkcmU_nz at BtbN
[00:54:30 CEST] <kerio> you're taking the time to store music, spend a little more effort and grab a flac
[00:54:35 CEST] <furq> i have never broken a law
[00:55:19 CEST] <iH2O> the folks in the original 13 colonies never broke the law
[00:55:27 CEST] <furq> i'm disappointed that nobody has said "but have you broken ¼ law"
[00:55:35 CEST] <BtbN> looks like a completely broken video to me, not a deinterlacing artifact
[00:55:38 CEST] <furq> what kind of ffmpeg irc channel is this
[00:55:51 CEST] <schelleursli> the video deinterlaces fine with yadif though at BtbN
[00:56:12 CEST] <BtbN> I have deinterlaced tons of stuff with cuvid, and never had any issues
[00:56:13 CEST] <schelleursli> source file is also up https://1drv.ms/v/s!At6cEyJzE4-RjHKfgHbvaShNN9uy
[00:56:44 CEST] <schelleursli> I'll test it on another one then
[00:57:06 CEST] <kerio> furq: the kind of channel where people use onedrive unironically, apparently
[00:57:10 CEST] <schelleursli> but the source is really nothing special, just a dvb-c recording created by vdr and then I cut it
[00:57:18 CEST] <furq> why would you ironically use onedrive
[00:57:24 CEST] <thebombzen> schelleursli: I think it's just a problem with with the hardware deinterlacer
[00:57:33 CEST] <kerio> memes need no explanation
[00:57:36 CEST] <thebombzen> I think the cuda hardware deinterlacer is producing buggy output
[00:59:02 CEST] <furq> out of interest, what card is it
[00:59:07 CEST] <james999> https://en.wikipedia.org/wiki/%CE%9C-law_algorithm
[00:59:25 CEST] <BtbN> yeah, it chokes on that video
[00:59:28 CEST] <schelleursli> still strange that it works nicely for BtbN, maybe he is deinterlacing ntsc or something else?
[00:59:33 CEST] <james999> lol does that look like a mu or a %CE% thingie like it does in this client?
[00:59:42 CEST] <BtbN> If I deinterlace my sample video with cuvid, it works fine
[00:59:47 CEST] <BtbN> deinterlacing that source.ts is broken
[01:00:01 CEST] <schelleursli> furq, it's some cheapo usb dvb-c card, terratec h5 I think
[01:00:08 CEST] <furq> i meant gfx card, but nvm
[01:00:10 CEST] <schelleursli> oh
[01:00:13 CEST] <furq> apparently it's not that
[01:00:14 CEST] <schelleursli> sorry :D
[01:00:35 CEST] <schelleursli> gtx 770, but ye looks like it's at least not user error, that makes me feel better already
[01:00:59 CEST] <james999> speaking of usb, what's the difference in data rate between a hdmi cable and a usb cable?
[01:01:08 CEST] <furq> which usb
[01:01:14 CEST] <furq> and also which hdmi
[01:01:17 CEST] <dystopia_> depends on hdmi version and usb version
[01:01:19 CEST] <james999> I've always wondered why people don't just carry around universal PnP usb monitors all the time
[01:01:21 CEST] <dystopia_> what furq said
[01:01:34 CEST] <thebombzen> schelleursli: I just tried playing the video using "mpv source.ts --hwdec=cuda --vd-lavc-o=deint=adaptive" and got the same stuttering
[01:01:37 CEST] <james999> the latest of both hdmi 2 and usb 3.1
[01:01:39 CEST] <dystopia_> usb-c > hdmi *.*
[01:01:40 CEST] <furq> you'd never get a 1080p60 signal over usb2
[01:01:43 CEST] <thebombzen> I think cuda is just outputting bad frames
[01:01:48 CEST] <furq> i don't think even 1080p30 would work
[01:01:55 CEST] <thebombzen> I'd just use yadif here tbh
[01:01:59 CEST] <james999> that would be a fun calculation to do
[01:02:08 CEST] <furq> it's not really that fun
[01:02:23 CEST] <thebombzen> schelleursli: make sure if you're using yadif that you software scale it. don't scale the video before deinterlacing or you will be sad
[01:02:30 CEST] <schelleursli> ye, I guess I'll go back to that, just sounded like a nice clean solution to get the deinterlace and scaling done on gpu
[01:02:44 CEST] <furq> (1920 * 1080 * 60 * 12) / 1048576
[01:02:50 CEST] <kerio> 12?
[01:02:51 CEST] <furq> is 1423Mbps
[01:02:55 CEST] <james999> you would just have to determine the bitrate of a 1080p at 30hz video and compare it to the bitrate of usb2
[01:02:56 CEST] <furq> 12bpp for yuv420p
[01:03:09 CEST] <schelleursli> that would've been my next question, I guess there is no way to do gpu scaling after I do software deinterlace?
[01:03:13 CEST] <BtbN> schelleursli, I feel like -r 50 is not correct for that video
[01:03:14 CEST] <kerio> why would a computer monitor use yuv
[01:03:22 CEST] <thebombzen> schelleursli: not with ffmpeg.c
[01:03:25 CEST] <furq> i was assuming some kind of best-case scenario
[01:03:26 CEST] <BtbN> is it something like 24000/1001?
[01:03:36 CEST] <thebombzen> you'd have to use opengl to do that
[01:03:38 CEST] <furq> if it is rgb then obviously that's 2850mbps
[01:03:49 CEST] <furq> which is a bit more than the 480mbps theoretical maximum of usb2
[01:03:56 CEST] <furq> which is of course a number that no usb2 connection has ever achieved in reality
[01:03:59 CEST] <thebombzen> but software scaling is much faster than preset slow+ on x264
[01:04:02 CEST] <thebombzen> so I wouldn't worry too much
[01:04:03 CEST] <schelleursli> it's normal pal 1080i tv standard
[01:04:14 CEST] <schelleursli> 1080i 25 fps interlaced
[01:04:20 CEST] <thebombzen> also using zscale is more accurate but slower scaling. given that your input is somewhat compressed I wouldnt' worry about that
[01:04:31 CEST] <dystopia_> not all pal .ts can be bobbed schelleursli
[01:04:37 CEST] <kerio> how does yuv420p work again
[01:04:38 CEST] <dystopia_> you should do a bob test with -t 10
[01:04:46 CEST] <dystopia_> to see if it's true 50 or dupe frames
[01:04:47 CEST] <furq> kerio: half vertical and horizontal resolution for u and v
[01:04:49 CEST] <thebombzen> kerio: a computer monitor wouldn't use yuv
[01:05:02 CEST] <thebombzen> YUV is a linear change-of-basis from RGB
[01:05:08 CEST] <james999> according to wikipedia the max bitrate of usb1, 2, and 3 is 1.5, 480, and 5,000 Mbit/s respectively
[01:05:20 CEST] <kerio> furq: how many bits per channel?
[01:05:26 CEST] <furq> 8
[01:05:28 CEST] <thebombzen> kerio: the difference is the YUV basis is subsampled
[01:05:33 CEST] <BtbN> schelleursli, the file is less
[01:05:36 CEST] <thebombzen> U and V are only recorded for a 2x2 square
[01:05:44 CEST] <furq> there is obviously such thing as yuv420p10le etc
[01:05:48 CEST] <thebombzen> so the U and V vertical and horizontal resolutions are cut in half
[01:05:56 CEST] <furq> didn't i just say that
[01:05:56 CEST] <schelleursli> I can do 50 fps yadif deinterlace with yadif=1 at dystopia_, how would I check with -t 10?
[01:06:01 CEST] <thebombzen> furq: you did just say that
[01:06:13 CEST] <thebombzen> Y contains the Luma component and U and V contain the chroma component
[01:06:16 CEST] <schelleursli> mh BtbN, not sure why that would be the case, maybe something went wrong when I cut it?
[01:06:33 CEST] <dystopia_> basically stick -t 10 infront of -i and do the rest of your command as normal
[01:06:34 CEST] <james999> furq: so now we should see a plethora of usb hd monitors... maybe
[01:06:34 CEST] <furq> if you start explaining what luma is i'm going to start crying
[01:06:39 CEST] <dystopia_> it will only encode 10 seconds
[01:06:40 CEST] <kerio> why did we get stuck with rgb and ycbcr
[01:06:46 CEST] <BtbN> no, it's just not exactly 25, and that's the hickups you see in the video
[01:06:50 CEST] <furq> rgb because it makes sense
[01:06:56 CEST] <thebombzen> kerio: it's all human eye psychovisual stuff
[01:06:56 CEST] <furq> yuv because of black and white tv compatibility
[01:07:10 CEST] <thebombzen> the human eye has 3 color receptors that are approximated well by red/green/blue
[01:07:12 CEST] <kerio> furq: sRGB doesn't make a lick of sense
[01:07:13 CEST] <schelleursli> BtbN, maybe I had a signal hickup, I'll use another source file to recheck
[01:07:17 CEST] <dystopia_> advance through the video frame by frame, if each frame is unique, it can be bobbed, if every other frame is a dupe it cannot be bobbed and yo should weave
[01:07:22 CEST] <slalom> hm
[01:07:25 CEST] <furq> what does srgb have to do with it
[01:07:28 CEST] <dystopia_> you should check all pal content before encoded
[01:07:34 CEST] <kerio> we're emulating CRT phosphor responses
[01:07:36 CEST] <thebombzen> kerio: the human eye sucks ass at seeing color but we're pretty good at seeing differences in lightness. so by reducing the chroma information we save space without really being able to tell a difference
[01:07:36 CEST] <dystopia_> cuz lots can't be bobbed and lots can
[01:07:39 CEST] <dystopia_> it's kinda random
[01:07:46 CEST] <james999> what's a good introductory textbook on YUV, luma, rgb colorspaces?
[01:07:55 CEST] <thebombzen> en.wikipedia.org
[01:07:56 CEST] <kerio> instead of using a decent linear color space
[01:07:56 CEST] <furq> also subsampled yuv is also because of the 1960s
[01:08:03 CEST] <furq> but reducing bandwidth still makes sense today
[01:08:09 CEST] <schelleursli> bloody interlacing standards, why are we still using these anyway, nothing but trouble
[01:08:14 CEST] <thebombzen> read the article on the the 1931 colorspace for an intro
[01:08:21 CEST] <kerio> like ciecam02
[01:08:32 CEST] <furq> schelleursli: funny you should ask, it's also because of reducing bitrate
[01:08:33 CEST] <james999> thebombzen: If you're gonna put it like that, may as well say The Book of Knowlege: Wikipedia.org
[01:08:49 CEST] <thebombzen> "read the article on the 1931 colorspace for an intro"
[01:08:53 CEST] <schelleursli> ye, my local tv sends it's signal at 720p 50 fps
[01:08:57 CEST] <furq> at least it's not telecine
[01:09:02 CEST] <schelleursli> not more deinterlacing on my end
[01:09:05 CEST] <schelleursli> -t
[01:09:23 CEST] <schelleursli> not perfect but still more convenient for me :D
[01:09:25 CEST] <thebombzen> kerio: so RGB is based on human eyes. the full range of color for the human vision isn't a polygon so the best we can do on a computer is approximate.
[01:09:43 CEST] <thebombzen> we have "primaries" which are red green and blue values that are some colors we can see
[01:09:51 CEST] <thebombzen> every color is a convex combination of those primary colors
[01:09:56 CEST] <kerio> it's still nonlinear
[01:10:05 CEST] <james999> fun fact: bees target certain flowers because they have ultraviolet designs on them that are invisible to humans
[01:10:07 CEST] <thebombzen> you are correct because the human eye is nonlinear
[01:10:07 CEST] <kerio> for "backwards compatibility" with CRT screens
[01:10:17 CEST] <thebombzen> no, because humans are nonlinear.
[01:10:18 CEST] <kerio> no, it's not linear in perception either
[01:10:26 CEST] <dystopia_> i've only ever seen 50fps been broadcast on hevc 4k stuff schelleursli
[01:10:26 CEST] <thebombzen> correct, because the human eye is nonlinear.
[01:10:40 CEST] <dystopia_> what country are you in to see 50fps broadcast via cable?
[01:10:46 CEST] <schelleursli> german and swiss television are doing 720p/50 broadcasts
[01:10:52 CEST] <dystopia_> ahh
[01:10:56 CEST] <dystopia_> cool
[01:10:59 CEST] <thebombzen> that sounded like a swiss video
[01:11:05 CEST] <thebombzen> from the announcer
[01:11:07 CEST] <schelleursli> it's german :)
[01:11:14 CEST] <thebombzen> well they speak german in switzerland
[01:11:21 CEST] <thebombzen> so how am I supposed to know lol
[01:11:34 CEST] <thebombzen> kerio: It's not possible to create a linear colorspace because the human eye doesn't perceive color linearly
[01:11:34 CEST] <schelleursli> you would know! it's one weird dialect we speak here
[01:11:34 CEST] <schelleursli> :D
[01:11:34 CEST] <furq> they don't speak real german
[01:11:44 CEST] <thebombzen> schelleursli: I wouldn't know unless I speak german
[01:11:45 CEST] <thebombzen> which I don't
[01:11:49 CEST] <schelleursli> eheh
[01:11:52 CEST] <james999> like the amish in america?
[01:11:56 CEST] <furq> swiss german speakers need subtitles on german tv
[01:12:09 CEST] <dystopia_> it's all french to me
[01:12:11 CEST] <thebombzen> idk what the Amish sound like
[01:12:18 CEST] <thebombzen> I have never interacted with amish people
[01:12:32 CEST] <furq> i am an expert on the amish because i have seen that episode of the simpsons where they accidentally build a barn instead of a pool
[01:12:40 CEST] <furq> also the film "kingpin"
[01:12:50 CEST] <james999> They speak a weird bastardized german: https://www.youtube.com/watch?v=oMbvC_siQyc
[01:12:53 CEST] <thebombzen> my best guess is they sound fairly similar to an older version of American english
[01:13:02 CEST] <thebombzen> I mean I know they speak a weird german but I don't know what it means to be a "weird german"
[01:13:02 CEST] <james999> Helps them keep separate from those "english" (everyone else)
[01:13:05 CEST] <thebombzen> given that I'm nota german speaker
[01:13:22 CEST] <thebombzen> also Amish don't speak german?
[01:13:30 CEST] <james999> idk maybe it's like someon speaking heavily accented english/pidgin
[01:14:24 CEST] <thebombzen> >googles it
[01:14:29 CEST] <thebombzen> >amish do speak german
[01:14:35 CEST] <thebombzen> [19:12:17] <thebombzen> I have never interacted with amish people
[01:14:59 CEST] <james999> This channel "Get Germanized" has a video comparing them: https://www.youtube.com/watch?v=PDChsbf_0kY
[01:15:27 CEST] <thebombzen> haha, unfortunately I don't speak german so this wont' really mean anything to me
[01:15:59 CEST] <thebombzen> is it mutally intelligible with german?
[01:16:03 CEST] <schelleursli> just tried with another channel and new sample, also checked for dvb stream errors first this time BtbN, same problem, guess it's just not meant to be
[01:16:20 CEST] <james999> he's speaking english, and listing pairs of words from each and comparing them
[01:16:24 CEST] <james999> they sound very similar to me
[01:19:09 CEST] <thebombzen> james999: btw w/r/t the bees thing
[01:19:13 CEST] <schelleursli> maybe I should open a bug report for this then, looks like it's not on my end and the input file should be standard pal dvb signal
[01:19:19 CEST] <thebombzen> probably not
[01:19:22 CEST] <thebombzen> it's a hardware issue
[01:19:25 CEST] <thebombzen> (most likely)
[01:19:30 CEST] <thebombzen> if it's a hardware decoder
[01:19:40 CEST] <schelleursli> ah
[01:19:44 CEST] <schelleursli> crap
[01:19:45 CEST] <schelleursli> ^^
[01:19:53 CEST] <james999> bee hardware decoders? rofl
[01:20:01 CEST] <thebombzen> jab416171:  fun fact: humans can see ultraviolet light, but our lens filters it out
[01:20:22 CEST] <thebombzen> an old version of cataract surgery was to remove the lens and give you enormous glasses
[01:20:42 CEST] <thebombzen> but patients who did that were able to see ultraviolet that is otherwise invisible. light my great grandmother would always talk about my dad's purple tie
[01:20:49 CEST] <thebombzen> but it was just brown to us
[01:21:42 CEST] <james999> that is... far out
[01:21:59 CEST] <james999> also why the hell doesn't google glass do that?
[01:22:07 CEST] <james999> would have been an actually useful feature of the things
[01:22:48 CEST] <furq> because google glass doesn't require you to remove the lenses in your eyes
[01:23:17 CEST] <james999> idk someting like making uv light appear as dark purple
[01:23:22 CEST] <james999> and IR as blood red
[01:23:41 CEST] <furq> how would that help you use google+
[01:24:44 CEST] <james999> haha, good point
[01:25:22 CEST] <thebombzen> james999: https://en.wikipedia.org/wiki/CIE_1931_color_space
[01:25:24 CEST] <thebombzen> read this
[01:25:42 CEST] <thebombzen> essentially it's impossible to fully capture human vision without dead space using 3 primaries
[01:26:17 CEST] <james999> is taht why monitors can never fully simulate actual vision as they are now?
[01:26:31 CEST] <thebombzen> yes
[01:26:41 CEST] <thebombzen> althougth you can get close
[01:26:47 CEST] <thebombzen> bt.2020 is pretty close
[01:27:12 CEST] <james999>  kind of makes it feel like a scam buying a $6000 tv set if it doesn't fully simulate real life
[01:29:21 CEST] <thebombzen> can I have my money back on goat simulator it doesn't simulate being a goat like real life
[01:29:50 CEST] <james999> XD
[01:40:45 CEST] <schelleursli> it's getting late here, thanks for all the infos and help I'm going to go back to software deinterlacing :)
[02:17:40 CEST] <cunhaw> test.
[02:26:01 CEST] <Threads> no homo
[02:29:15 CEST] <thebombzen> ?
[02:30:23 CEST] <dsc_> its not working
[02:33:02 CEST] <thebombzen> this looks like a job for batman
[03:44:46 CEST] <Threads> when extracting teletext subtitles how can i keep the colors ?
[05:30:47 CEST] <diverdude> hi, i am trying to compile a format detection program but i am getting undefined symbols, but i cannot understand why because i have pretty much included all libraries in ffmpeg in the compile command.....Can anybody help me understand what I am missing? Here is comile command and error: http://paste.ubuntu.com/24470717/
[05:39:32 CEST] <james999> i'm very confuse
[05:39:43 CEST] <james999> i tried streaming to the xbox again and I noticed the picture from a youtube vid is horrible
[05:39:49 CEST] <james999> but when I added -vcodec libx264 it got better
[05:40:01 CEST] <james999> h264 is not simlar or identical to x264?
[05:40:37 CEST] <james999> the ytube vid is 1280x720 with h264 video i thought
[05:50:36 CEST] <thebombzen> what codec had you been using before
[05:54:19 CEST] <furq> if you were just doing -f mpegts then it'll have been using mpeg2video
[05:54:21 CEST] <furq> which is indeed horrible
[05:54:31 CEST] <furq> you need to add -c copy if you don't want to reencode
[06:05:24 CEST] <Tatsh> hitting this bug :( https://trac.ffmpeg.org/ticket/5086
[06:14:35 CEST] <james999> oh ok furq
[06:14:38 CEST] <james999> that makes more sense
[06:15:06 CEST] <james999> it was a 1920x1080 screen and i suppose mpeg2video wasn't built with that res in mind or something
[06:15:34 CEST] <furq> well it was probably also using the default options
[06:16:12 CEST] <furq> but you should just copy the streams or use x264 anyway
[06:17:04 CEST] <Tatsh> hmm, getting this with concat
[06:17:09 CEST] <Tatsh> concat: protocol and not as file
[06:17:10 CEST] <Tatsh> Too many packets buffered for output stream 0:1.
[06:17:14 CEST] <Tatsh> 0:1 is my audio stream
[06:17:29 CEST] <Tatsh> and it's just PCM, in several MKVs i'm trying to concatenate
[06:20:37 CEST] <james999> hmm i still haven't ifgured out a way to test the speed of my wifi network locally
[06:20:54 CEST] <james999> I tried udp streaming of 1920x1080 to the tv and it was choppy
[06:21:05 CEST] <james999> but 1280x720 seemed to work fine
[06:24:35 CEST] <james999> hmm using that 12bpp number from earlier, that comes out to 316Mbps
[06:24:43 CEST] <diverdude> anybody?
[06:24:44 CEST] <james999> but i thought my wireless was only 54Mbps. o_0
[06:27:54 CEST] <james999> well it says lib not found for x86_64
[06:27:59 CEST] <james999> maybe you have 32 bit libraries installed?
[06:28:10 CEST] <james999> or mb you need a flag to gcc to search them or something
[06:43:00 CEST] <Tatsh> it appears i can use the concat protocol
[06:43:10 CEST] <Tatsh> the audio streams i'm muxing back in are too long
[06:43:23 CEST] <Tatsh> that's why i get this 'Too many packets buffered for output stream 0:1.' message
[06:57:19 CEST] <diverdude> bah...when i output compiler commands from the make examples it seems that each example require 8 different non-trivial compile commands....why o why must it be so hard :/
[08:28:49 CEST] <JC_Yang> how to demux raw stream from a mpeg program stream file?
[08:44:55 CEST] <JC_Yang> well, done
[10:18:38 CEST] <JC_Yang> how to customize passed to compiler and linker options in the configure script?  seems no way to enable or disable some options
[10:44:33 CEST] <JC_Yang> I'm trying to specify clang as cross compiler and also config --sysroot, but the c compiler test pass  -isysroot= option to the ld(I also specify the cross-ld), and it is a unrecognized option, cause the link failed. Shouldn't this be fixed? ld does not accept -isysroot, and then I realize that specify the cross-ld directly is not the correct path, what I need is specify the -B/path/to/cross/bin t
[10:44:35 CEST] <JC_Yang> o clang, but how to do that with the current configure script? any options?
[11:41:00 CEST] <JC_Yang> hmm, switch to gcc, but --sysinclude are not respected. errno.h not found
[11:44:37 CEST] <JC_Yang> fixed with --extra-cflags...
[11:48:44 CEST] <sinanksu> Hi guys
[11:49:00 CEST] <sinanksu> I have a problem.
[11:49:11 CEST] <sinanksu> I want help
[11:49:28 CEST] <sinanksu> I am getting a live stream in raw format with rtmpdump. Then I broadcast this live broadcast via rtmp with ffmpeg. But the live broadcast continues to freeze for 5 seconds. 5 seconds playing 1 second freezing, 5 seconds playing 1 second freezing, 5 seconds playing 1 second freezing. This is how the live broadcast progresses.
[11:51:07 CEST] <sinanksu> my code : rtmpdump -r rtmpe://xx.xx.xx.xx/live -a xlive -f WIN 23,0,0,162 -s http://link/VideoPlayer.swf -w dsa4d64as9d4as89d498as4d98a4sd894a89d4894s98d49asd a -x 585534 -p "http://link" -C S:client -C S:3.1.0.10 -C S:en --live -y raw:999999 | ( ffmpeg -re -i - -sn -vcodec copy -acodec copy -f flv rtmp://127.0.0.1:1935/restream/999999 )
[11:56:02 CEST] <matej_k> is h264 parser adding prefix (14) nal units, or am I seeing things?
[12:00:10 CEST] <sinanksu> ?
[12:01:50 CEST] <matej_k> i have mpeg ts file with h.264 video, demuxing it with avcodec, and im getting nal prefix units 14, which i dont think are in the file itself.
[12:02:52 CEST] <cryptodechange> Hm
[12:02:54 CEST] <cryptodechange> https://pastebin.com/cTLDCc6g
[12:03:16 CEST] <cryptodechange> Too many packets buffered, for an audio stream
[12:03:52 CEST] <cryptodechange> I could just convert video only, then remux audio and subtitles
[12:10:16 CEST] <sinanksu> H0
[12:10:18 CEST] <sinanksu> I registered here before. I do not want a password reset email for 3 days. There is also no spam in the box .. I can not log in.
[12:10:27 CEST] <sinanksu> Help me..
[12:11:19 CEST] <sinanksu> http://ffmpeg.gusari.org
[12:11:24 CEST] <sinanksu> mail does not come
[12:15:58 CEST] <BtbN> That's some inofficial forum, ffmpeg is in no way affiliated with it.
[12:37:22 CEST] <matej_k> well, it was missing AVFMT_FLAG_KEEP_SIDE_DATA
[12:41:59 CEST] <matej_k> its strange to have flag that basically causes broken output and is not set by default
[12:43:09 CEST] <matej_k> okay, i see this now https://www.mail-archive.com/ffmpeg-cvslog@ffmpeg.org/msg26105.html
[13:07:24 CEST] <Phantom_Hoover> hi, i've got some videos which are high-quality but seem to have really sparse keyframes or something, so seeking takes forever
[13:07:48 CEST] <Phantom_Hoover> is there anything i can do to reencode to fix that?
[13:27:26 CEST] <james999> Phantom_Hoover: according to this ask you may not see much increase in quality, but here ya go: http://stackoverflow.com/questions/30979714/how-to-change-keyframe-interval-in-ffmpeg
[13:27:40 CEST] <james999> you have to give a paramter to -vcodec libx264
[13:27:47 CEST] <james999> -x264-params
[13:30:44 CEST] <Phantom_Hoover> quality's not the issue, it's usability
[13:31:33 CEST] <Phantom_Hoover> if i click on the timebar i get a frozen picture with running audio for a few seconds, and often there are artefacts on the video for a few seconds more once it starts playing
[13:33:13 CEST] <Phantom_Hoover> just reencoding through vlc with its stock x264/mp4 profile fixes that up pretty well though
[13:34:44 CEST] <Phantom_Hoover> fucked up the subtitles though
[13:59:59 CEST] <james999> i didn't even know vlc let you encode anything
[14:00:20 CEST] <james999> the options are so poorly documented it may as well go to the moon with --launch-moon-rocket and i'd never know either lol
[14:01:17 CEST] <BtbN> that's a bug in your player. Seeking in files with sparse keyframes should not be an issue
[14:03:52 CEST] <bwe> I want to make a slidecast from several images. Yesterday teratorn was so kind to give me a boilerplate (a complex filter). I don't get the syntax right: https://bpaste.net/show/f5a1905bc9f0 Where can I find out in the documentation how to do it correctly?
[14:11:21 CEST] <Phantom_Hoover> BtbN, well i mean it's just vlc, but if you say so
[14:11:27 CEST] <Phantom_Hoover> any fixes for it though?
[14:14:33 CEST] <nyuszika7h> is there a way to scan a file for H.264/AAC decoding errors without playing it and waiting for the whole duration?
[14:14:39 CEST] <bwe> What does   Undefined constant or missing '(' in 'start'   mean (I am new to ffmpeg but already converted some videos in the past successfully)?
[14:15:11 CEST] <nyuszika7h> bwe: that should probably be start=0 not start:0
[14:15:19 CEST] <nyuszika7h> I'm not sure though
[14:16:15 CEST] <nyuszika7h> actually trim=start=0 might be what it's complaining about
[14:17:10 CEST] <nyuszika7h> or maybe not, according to the documentation that should be fine
[14:17:49 CEST] <bwe> nyuszika7h: Oh, could you point me to the location in the documentation?
[14:18:59 CEST] <nyuszika7h> change loop=-1:start:0:size=1 to loop=loop=-1:start=0:size=1
[14:19:03 CEST] <nyuszika7h> https://ffmpeg.org/ffmpeg-filters.html#loop
[14:19:38 CEST] <nyuszika7h> https://ffmpeg.org/ffmpeg-filters.html#Filtergraph-syntax-1
[14:20:00 CEST] <nyuszika7h> though loop=-1:start=0:size=1 will probably work too
[14:20:07 CEST] <nyuszika7h> so I was right the first time, start:0 -> start=0
[14:20:24 CEST] <nyuszika7h> but it's more logical to either put them in order or use option names everywhere
[14:20:35 CEST] <nyuszika7h> loop=-1:1:0 is another option (loop, start, size)
[14:25:56 CEST] <BtbN> Phantom_Hoover, try another player, MPC-HC or something
[14:27:09 CEST] <Phantom_Hoover> BtbN, on linux
[14:27:19 CEST] <BtbN> well, then something else
[14:27:58 CEST] <nyuszika7h> https://mpv.io/
[14:29:31 CEST] <Phantom_Hoover> mpv seems p good
[14:30:47 CEST] <BtbN> Or a VLC nightly, if you are using some 2.2 release
[14:39:13 CEST] <bwe> nyuszika7h: Thanks. I have got the syntax right. How can I scale the pictures to a sensible output size proportionally and fill the left over space with white colour? https://bpaste.net/show/e755cdc9a4a1
[14:39:47 CEST] <vlt> nyuszika7h: You could write the raw video you get from `ffmpeg -i testfile` to /dev/null and see if decoding errors appear.
[14:40:02 CEST] <nyuszika7h> vlt: yeah I just found that, thanks
[14:40:17 CEST] <nyuszika7h> (actual command I used: ffmpeg -v error -i INPUT -f null -)
[14:40:45 CEST] <nyuszika7h> bwe: hmm, you could take take a completely white PNG file as input and overlay the scaled video on top of that
[14:40:49 CEST] <nyuszika7h> one sec
[14:41:52 CEST] <nyuszika7h> though it seems there are filters to do that without making a PNG image
[14:44:51 CEST] <bwe> nyuszika7h: drawbox? Draw a colored box on the input image.
[14:45:01 CEST] <nyuszika7h> yes
[14:45:04 CEST] <nyuszika7h> I'm testing that atm
[14:52:17 CEST] <cryptodechange> Finished encoding, mediainfo gives some extra params I didn't notice
[14:52:19 CEST] <cryptodechange> crf_max=0
[14:54:05 CEST] <nyuszika7h> that sounds like lossless encoding
[14:54:15 CEST] <nyuszika7h> which likely produces a very big file
[14:54:52 CEST] <cryptodechange> cabac=1 / ref=5 / deblock=1:-3:-3 / analyse=0x3:0x133 / me=tesa / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=64 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=18 / lookahead_threads=4 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=16 / b_pyramid=2 / b_adapt=2
[14:54:52 CEST] <cryptodechange> / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=150 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / vbv_maxrate=62500 / vbv_bufsize=78125 / crf_max=0.0 / nal_hrd=none / filler=0 / ip_ratio=1.40 / aq=1:0.80
[14:55:08 CEST] <cryptodechange> 10.7gb from 32gb
[14:55:49 CEST] <nyuszika7h> hmm strange that it says crf_max=0.0
[14:55:55 CEST] <nyuszika7h> that definitely doesn't look like lossless encoding
[14:56:08 CEST] <nyuszika7h> oh you need to use 'pad' not 'drawbox' bwe
[14:57:10 CEST] <nyuszika7h> bwe: do you want to center the image?
[14:58:05 CEST] <nyuszika7h> hmm
[14:58:07 CEST] <nyuszika7h> "If x or y evaluate to a negative number, theyll be changed so the input image is centered on the padded area."
[14:58:16 CEST] <nyuszika7h> [Parsed_pad_1 @ 00000000025afac0] Negative values are not acceptable.
[14:58:42 CEST] <bwe> Yes, I want to center it.
[14:59:07 CEST] <bwe> Which format are most YouTube videos?
[14:59:36 CEST] <nyuszika7h> okay this works
[14:59:53 CEST] <nyuszika7h> -filter_complex "[0:v:0]scale=640:360,pad=1280:720:in_w/2:in_h/2:color=white[vid]" -map "[vid]"
[15:00:00 CEST] <nyuszika7h> bwe: define "format"
[15:00:03 CEST] <nyuszika7h> container? video/audio codec?
[15:00:36 CEST] <nyuszika7h> YouTube can take pretty much anything, they will transcode it anyway
[15:00:55 CEST] <nyuszika7h> you could use H.264 and AAC for audio (if there's audio), or VP9 and Opus for audio (though VP9 encoding is much slower I think)
[15:00:59 CEST] <bwe> nyuszika7h: (with ffmpeg =). Oh, I regret of being completely imprecise. I mean resolution, aspect ratio.
[15:01:08 CEST] <nyuszika7h> oh
[15:01:13 CEST] <nyuszika7h> well nowadays they're mostly 16:9
[15:01:44 CEST] <bwe> which resolution do you recommend?
[15:01:55 CEST] <cryptodechange> Should I not be concerned over crf_max=0.0, or did I set my encode wrong?
[15:02:13 CEST] <nyuszika7h> cryptodechange: since the output is not a huge file but much smaller than the input, nothing to worry about I think
[15:02:19 CEST] <bwe> Well, that would be slides with an audio narration.
[15:02:29 CEST] <nyuszika7h> 1080p at most is probably fine
[15:02:34 CEST] <nyuszika7h> could do 720p too if you want smaller
[15:02:46 CEST] <nyuszika7h> it's just slides but at least 720p is recommended for any content really
[15:02:49 CEST] <cryptodechange> I used the veryslow preset but used some placebo settings
[15:03:49 CEST] <bwe> I can't simply do `pad=1080p:x=-1:y=-1:color=white`?
[15:03:58 CEST] <nyuszika7h> -1 didn't work for me
[15:05:35 CEST] <nyuszika7h> pad=1920:1080:x=in_w/2:y=in_h/2:color=white
[15:06:07 CEST] <nyuszika7h> the x=/y=/color= are not needed if you put them in that order but it doesn't hurt either
[15:06:44 CEST] <nyuszika7h> anyway this is assuming you scale to a size *smaller* than 1080p and then pad it with white to 1080p
[15:07:10 CEST] <nyuszika7h> though you could also just upload as-is if the input images are the same resolution
[15:07:27 CEST] <nyuszika7h> unless it doesn't look good without any padding, I don't know, I would have to see
[15:08:16 CEST] <nyuszika7h> there is probably some way to just pad everything to the biggest input image's resolution
[15:08:37 CEST] <nyuszika7h> the player is 16:9 so it will add black bars if your aspect ratio is different (full screen will have black bars on non-16:9 displays)
[15:11:00 CEST] <bwe> nyuszika7h: I've added pad to each image's filter: https://bpaste.net/show/354c1972d018 It raises `Cannot find a matching stream for unlabeled input pad 0 on filter Parsed_pad_9
[15:11:50 CEST] <nyuszika7h> you did "trim=start=0:end=10[0],pad=width=1920:height=1080:x=-999:color=white"
[15:11:58 CEST] <nyuszika7h> it should be "trim=start=0:end=10,pad=width=1920:height=1080:x=-999:color=white[0]"
[15:12:38 CEST] <nyuszika7h> the [0] and such are the outputs of each filter chain and go at the end, then you use a semicolon to start the next input/output pair
[15:13:03 CEST] <nyuszika7h> the pad goes BEFORE the [0], [1] etc. which were before the semicolons
[15:13:16 CEST] <bwe> The [0] refers to which stream?
[15:13:27 CEST] <nyuszika7h> none, [0] is the output of the filter chain
[15:13:34 CEST] <nyuszika7h> though I don't know if it's valid to call your outputs that
[15:13:45 CEST] <nyuszika7h> it may be confused with streams
[15:13:57 CEST] <nyuszika7h> I'd call them [a], [b], etc. or something
[15:14:37 CEST] <nyuszika7h> -filter_complex '[0:v]setsar=sar=1/1,fps=30,loop=loop=-1:start=0:size=1,trim=start=0:end=10,pad=width=1920:height=1080:x=-999:color=white[a];[1:v]setsar=sar=1/1,fps=30,loop=loop=-1:start=0:size=1,trim=start=0:end=10,pad=width=1920:height=1080:x=-999:color=white[b];[a][b]concat=n=2[out]'
[15:16:32 CEST] <nyuszika7h> though instead of a set size, you could also do pad=aspect:16/9:x=in_w/2:y=in_h/2:color=white
[15:16:45 CEST] <nyuszika7h> hmm in_w/2 and in_h/2 might only work in my example because it's exactly double, right
[15:17:57 CEST] <bwe> nyuszika7h: I want to script it for a text file wherein each line maps a filename to a timestamp. In turn I desire to iterate over any number of images. While the alphabet ends with 26 characters, the numbers don't. What's the best case to handle it? https://bpaste.net/show/f3c1b6d97fca
[15:18:05 CEST] <sinanksu> MY CHANGE
[15:18:09 CEST] <sinanksu>  ffmpeg -re -i - -sn -vcodec libx264 -x264-params keyint=120:no-scenecut -acodec copy -f flv rtmp://127.0.0.1:1935/restream/
[15:18:28 CEST] <sinanksu> add  -vcodec libx264 -x264-params keyint=120:no-scenecut
[15:18:58 CEST] <sinanksu> The problem continues
[15:19:13 CEST] <sinanksu> Continues to freeze
[15:19:24 CEST] <bwe> nyuszika7h: Negative values are not acceptable.`
[15:19:30 CEST] <nyuszika7h> yeah I got that error too
[15:20:08 CEST] <sinanksu> I've also been on the forum but nobody has written yet : http://ffmpeg.gusari.org/viewtopic.php?f=11&t=3611
[15:21:36 CEST] <sinanksu> Could you write me the right values?
[15:23:31 CEST] <nyuszika7h> bwe: okay I got it to center for any resolution
[15:24:08 CEST] <nyuszika7h> pad=width=1920:height=1080:x=(out_w-in_w)/2:y=(out_h-in_h)/2:color=white
[15:25:14 CEST] <nyuszika7h> I'll think about the other thing
[15:27:23 CEST] <bwe> It works with the numbers.
[15:31:30 CEST] <nyuszika7h> also, the default CRF for x264 (28) might not have such a good quality, if you want to improve the output quality, try values from "-crf 23" to "-crf 18" (before the output file name)
[15:31:51 CEST] <nyuszika7h> also if the input has audio, make sure to add "-c:a copy" to avoid transcoding the audio
[15:32:28 CEST] <nyuszika7h> also hmm awk, interesting, never really used it for such complex scripting, but I understand the code :P
[15:40:08 CEST] <bwe> nyuszika7h: Yup, audio is the next thing. Before, can I increase `speed= 0.6x` ?
[15:41:05 CEST] <nyuszika7h> bwe: use a faster preset
[15:41:08 CEST] <nyuszika7h> the default is medium
[15:41:25 CEST] <nyuszika7h> '-preset fast', '-preset faster', '-preset veryfast', '-preset superfast', '-preset ultrafast'
[15:41:37 CEST] <nyuszika7h> this will increase the file size
[15:42:04 CEST] <nyuszika7h> also probably a good idea to explicitly specify "-c:v libx264" though it's usually the default
[15:42:24 CEST] <nyuszika7h> https://trac.ffmpeg.org/wiki/Encode/H.264
[15:42:36 CEST] <nyuszika7h> oh 23 is default not 28
[15:43:14 CEST] <nyuszika7h> that's not so bad for slides but you can still reduce it if the output quality doesn't look good (18 is generally visually lossless, especially for slides no need to go below that most likely)
[15:43:18 CEST] <sinanksu> add -c:a copy
[15:43:34 CEST] <nyuszika7h> yes, I mentioned that aleady
[15:43:35 CEST] <nyuszika7h> *already
[15:43:43 CEST] <nyuszika7h> 2017-04-28 15:31:51+0200 <nyuszika7h> also if the input has audio, make sure to add "-c:a copy" to avoid transcoding the audio
[15:43:55 CEST] <sinanksu> There has not been a change. There is still a problem.
[15:44:06 CEST] <nyuszika7h> oh you're not talking about bwe's problem
[15:44:15 CEST] <nyuszika7h> I was talking to bwe, sinanksu
[15:44:18 CEST] <sinanksu> sory
[15:44:30 CEST] <nyuszika7h> I'm not sure about your issue sorry, maybe someone else will be able to help
[15:45:40 CEST] <sinanksu> Thank you. I will wait.
[15:55:49 CEST] <bwe> nyuszika7h: The approach works very robust: https://bpaste.net/show/c71d1ba797e4 Yet I have to find out why there is no audio in the output video.
[15:56:46 CEST] <arcamargo> g'morning guys. I have a question: I'm trying to figure out what that 0x6040 means in -vf  "[in] scale=1280:720:flags=0x6040 [out]"
[15:57:03 CEST] <bwe> nyuszika7h: I do not include it in concat, I'll try.
[15:58:29 CEST] <nyuszika7h> bwe: -map '[out]' -map 0:a
[15:58:49 CEST] <nyuszika7h> oh wait 1:a
[15:58:53 CEST] <nyuszika7h> actually 2:a
[15:58:56 CEST] <nyuszika7h> because you have 2 image inputs
[15:59:12 CEST] <nyuszika7h> if you put the audio first it can be always 0:a
[15:59:26 CEST] <nyuszika7h> but then you have to reindex the video tracks to start from 1:v
[16:02:41 CEST] <bwe> Done. Works seemlessly!
[16:02:58 CEST] <bwe> This is terriffic =). Thanks a bunch.
[16:07:51 CEST] <bwe> Can ffmpeg take pdf as input?
[16:09:03 CEST] <Mandevil> This is driving me nuts... I reencode a full range video from a DSLR. The ffmpeg output shows it's treated as full range on both input and output... yet the video is clearly lighter, less contrasty on the output (yes, in ffplay too). What am I doing wrong? https://pastebin.com/ZDgzrmVN
[16:20:23 CEST] <cdidd> the bitrate that ffmpeg shows during transcoding, is it an instant bitrate or an average over whole length of encoding process, or some other average?
[16:20:45 CEST] <Mandevil> cdidd: I'd like to know too.
[16:20:56 CEST] <Mandevil> But it definitely doesn't look like current bitrate.
[16:22:10 CEST] <cdidd> I searched github for "Lsize=", couldn't find anything
[16:22:33 CEST] <cdidd> any printf I mean
[16:30:34 CEST] <cdidd> bitrate = pts && total_size >= 0 ? total_size * 8 / (pts / 1000.0) : -1;
[16:32:06 CEST] <cdidd> so they calculate bitrate encoded size and time
[16:34:50 CEST] <cdidd> from*
[16:36:14 CEST] <Mandevil> So average.
[16:40:07 CEST] <Mandevil> swscaler src_range/dst_range accept what values?
[16:40:19 CEST] <Mandevil> 'pc' doesn't parse.
[16:44:30 CEST] <Mandevil> So it looks like -vf scale=scr_range=1:dst_range=1 solves my problem.
[16:44:38 CEST] <kepstin> Mandevil: looks like it's a boolean, where true is full range (pc), false is limited range?
[16:44:47 CEST] <Mandevil> Yeah, looks like that.
[16:45:07 CEST] <Mandevil> It looks like the scaler was scaling the range despite the input and output were both full range.
[16:46:30 CEST] <kepstin> hmm, the src_range and dst_range parameters on the scale filter seem redundant with the in_range and out_range parameters
[16:46:40 CEST] <kepstin> (the latter are more expressive, and default to "auto")
[16:47:40 CEST] <Mandevil> Hm, I don't see in_range/out_range in the documentation?
[16:48:03 CEST] <Mandevil> Provided this is the documentation: https://ffmpeg.org/ffmpeg-scaler.html
[16:50:21 CEST] <kepstin> if you're using the scale *filter* please read the docs here: https://www.ffmpeg.org/ffmpeg-filters.html#scale-1
[16:51:13 CEST] <Mandevil> Hm.
[16:52:04 CEST] <Mandevil> so "-vf scale=in_range=full:out_range=full"?
[16:53:27 CEST] <kepstin> that should be equivalent, yeah
[16:56:19 CEST] <Mandevil> Still, this shouldn't be happening... it's quite a nasty gotcha.
[16:56:34 CEST] <Mandevil> Everything looks good on the surface, but the video is getting mangled.
[16:57:10 CEST] <kepstin> it shouldn't be, unless the input file has the color range incorrectly flagged in metadata (or not flagged at all, so ffmpeg is guessing)
[16:57:24 CEST] <Mandevil> No, the tagging is ok.
[16:57:29 CEST] <Mandevil> That's the confusing part.
[16:57:39 CEST] <Mandevil> ffmpeg clearly sees everything fine.
[16:57:51 CEST] <Mandevil> But then swscaler comes and rescales the range anyway.
[16:58:05 CEST] <Mandevil> Without saying anything, of course.
[17:06:33 CEST] <erick3k> Good day everyone, having a problem. After 60 some hours the streams throw this https://0bin.net/paste/2gFUV2JHj4JPbPKk#NDZ55droKt4SdRr6UWuuO3wlSqrtuJxX132KySuKaW0
[17:06:42 CEST] <erick3k> any ideas what is goin on?
[17:10:44 CEST] <Mandevil> "Connection reset by peer"
[17:37:02 CEST] <erick3k> right, what does that mean?
[17:37:08 CEST] <erick3k> my server failed or youtube failed?
[17:37:37 CEST] <erick3k> i see youtube streamings 24/7 so why would youtube be reseting the connection and randomly?
[17:37:48 CEST] <erick3k> why does it not happens with other streams i see on youtube?
[17:42:44 CEST] <erick3k> Mandevil i've been trying to solve this for a week, noone seems to know. So just blame Youtube and give up?
[17:42:51 CEST] <erick3k> its that the ultimate result?
[17:53:21 CEST] <bwe> Negative values are not acceptable. I get this error only if I add the third trim for the third image: http://bpaste.net/show/1b6a45ddc697 Which negative value is meant?
[17:58:32 CEST] <erick3k> anyone hello?
[17:58:36 CEST] <kepstin> bwe: your 4th video is taller than 1080, and "1240 - 1080" is a negative number
[17:59:26 CEST] <durandal_1707> you meant other way around?
[17:59:50 CEST] <bwe> kepstin: True. May I scale it automatically down?
[18:00:35 CEST] <kepstin> bwe: sure, you can use the scale filter to do that, it can use expressions to e.g. only scale down the video if it's too big.
[18:02:32 CEST] <bwe> kepstin: eval Specify when to evaluate width and height expression. It accepts the following values:
[18:02:40 CEST] <bwe> Is this what I want?
[18:02:58 CEST] <Mandevil> erick3k: You can try to find out how the connection gets dropped by capturing the stream with tcpdump.
[18:03:22 CEST] <kepstin> no, you can just put expressions into the height and width fields directly, it works a lot like the pad filter...
[18:07:30 CEST] <alex88> hi there, I'm trying to stream a video to ffserver, I've this command and output https://gist.github.com/alex88/02f7dfc386181519dc9412cb3466f6bc why do I get `Stream #0:0 -> #0:0 (rawvideo (native) -> mpeg4 (native))` even if I select libx264?
[18:07:51 CEST] <erick3k> Mandevil what command should i add to ffmpeg to capture that?
[18:07:57 CEST] <alex88> if I add -f mp4, which doesn't work because the output is not seekable I correctly get "Stream #0:0 -> #0:0 (rawvideo (native) -> h264 (libx264))"
[18:09:15 CEST] <Mandevil> erick3k: Are you in Windows or on Linux?
[18:09:32 CEST] <erick3k> linux
[18:09:37 CEST] <erick3k> this is the command hold on
[18:09:51 CEST] <Mandevil> tcpdump is completely unrelated to ffmpeg.
[18:09:57 CEST] <Mandevil> It's a packet capture utility.
[18:10:22 CEST] <erick3k> https://0bin.net/paste/no3ATdIMvRYJc11D#nT6WuZqmE8QJigtJpcUicY-p4viK2/0tzHX0iokEciG
[18:10:51 CEST] <erick3k> so what would that tell me?
[18:11:14 CEST] <erick3k> i tried two different server on really high end datacenters same thing happen
[18:11:18 CEST] <Mandevil> It will tell you who is dropping the connection.
[18:11:29 CEST] <erick3k> oh so its not youtube then
[18:11:31 CEST] <erick3k> it can be me
[18:11:34 CEST] <Mandevil> erick3k: Are you sure YT doesn't impose any limits on stream duration?
[18:11:41 CEST] <erick3k> is there a command to reconnect ffmpeg if its dropped?
[18:11:50 CEST] <Mandevil> erick3k: It says "peer closed the connection"
[18:11:50 CEST] <erick3k> Yes, youtube have no limits
[18:12:02 CEST] <Mandevil> erick3k: Just put the whole thing into a loop?
[18:12:04 CEST] <erick3k> i can last 24 hours 3 days or 6 days, same thing happens
[18:12:08 CEST] <erick3k> so no, no limit
[18:12:20 CEST] <erick3k> it is on a loop already
[18:12:28 CEST] <erick3k> but doesn't reconnect when dropped
[18:12:36 CEST] <erick3k> ffmpeg just stops
[18:12:48 CEST] <Mandevil> It still is running, but isn't streaming?
[18:13:17 CEST] <erick3k> looks like yea
[18:13:17 CEST] <erick3k> https://i.imgur.com/wcuZGZs.png
[18:13:55 CEST] <Mandevil> I don't see ffmpeg there...
[18:13:56 CEST] <Mandevil> ?
[18:14:23 CEST] <erick3k> is on stream_to_youtube sh
[18:14:38 CEST] <erick3k> which has this https://0bin.net/paste/no3ATdIMvRYJc11D#nT6WuZqmE8QJigtJpcUicY-p4viK2/0tzHX0iokEciG
[18:14:44 CEST] <Mandevil> But no ffmpeg. Looks like ffmpeg actually quit.
[18:14:56 CEST] <erick3k> i guess yep
[18:15:08 CEST] <Mandevil> So just start it again and be done with it.
[18:15:22 CEST] <erick3k> lol its gonna fail again
[18:15:28 CEST] <erick3k> 1000% sure
[18:15:29 CEST] <Mandevil> Immediately?
[18:15:43 CEST] <erick3k> no, it sometimes does it after 24h maybe 3 days maybe 6 days
[18:15:48 CEST] <erick3k> is unpredictable
[18:16:04 CEST] <Mandevil> Do you realize that network outages and short hiccups etc are a thing?
[18:16:20 CEST] <Mandevil> You can't count on having perfect connection.
[18:16:25 CEST] <erick3k> Yep, i wonder how some people manage to stream 24/7 no dc
[18:16:45 CEST] <erick3k> ok so is there a way to reconnect if it disconnects automatically?
[18:16:47 CEST] <Mandevil> They definitely have to deal with networking going down.
[18:16:59 CEST] <Mandevil> Just run the ffmpeg again.
[18:17:19 CEST] <erick3k> i want to stream 24/7 no maintenance
[18:17:40 CEST] <erick3k> it will stop randomly so is there anything i can add to the scripts to reconnect automatically
[18:17:48 CEST] <erick3k> instead of me going back to the server and rerun it?
[18:17:50 CEST] <Mandevil> Just loop the ffmpeg execution so it will just try again and again until it connects.
[18:18:14 CEST] <erick3k> i think its looped?
[18:18:18 CEST] <erick3k> already
[18:18:48 CEST] <Mandevil> Do you understand what that script is doing?
[18:18:55 CEST] <erick3k> Yes,
[18:19:06 CEST] <erick3k> re playing the ts
[18:19:13 CEST] <erick3k> but nothing to do with "reconnect"
[18:19:14 CEST] <Mandevil> So you see that the only looped thing is that cat command, right?
[18:19:19 CEST] <erick3k> correct
[18:19:23 CEST] <erick3k> perhabs why am asking lol
[18:19:31 CEST] <Mandevil> When the ffmpeg quits, it simply isn't there.
[18:19:39 CEST] <Mandevil> The script doesn't handle ffmpeg failing.
[18:19:47 CEST] <erick3k> sorry am no linux expert
[18:19:56 CEST] <alex88> is vcodec supported only when there is a format set?
[18:19:58 CEST] <erick3k> how can i make it handle ffmpeg failing?
[18:20:34 CEST] <erick3k> i am not entirely sure, furq did help me with those lines
[18:20:50 CEST] <Mandevil> I think that the loop goes on even with ffmpeg exiting.
[18:21:04 CEST] <erick3k> i agree
[18:21:21 CEST] <erick3k> any idea how to make it restart if it exits or something?
[18:23:34 CEST] <bwe> kepstin: scale=h='min(1080, ih):w=-1'  would be the filter. How do I escape it correctly?
[18:26:00 CEST] <Mandevil> erick3k: Why is the loop even there? What is that source.ts?
[18:26:33 CEST] <erick3k> is a rain chill video and repeats so it streams 24/7
[18:26:46 CEST] <erick3k> https://www.youtube.com/watch?v=JGd4LumCPpg
[18:26:48 CEST] <erick3k> that one
[18:27:16 CEST] <Mandevil> I see.
[18:28:56 CEST] <Mandevil> erick3k: Use -stream_loop -1 in ffmpeg.
[18:29:03 CEST] <Mandevil> erick3k: And do away with that shell loop completely.
[18:29:18 CEST] <Mandevil> And then loop the ffmpeg in while true; do ffmpeg ...; done
[18:30:00 CEST] <Mandevil> https://ffmpeg.org/ffmpeg-all.html#Main-options
[18:31:59 CEST] <erick3k> Mandevil thanks, so something like this https://0bin.net/paste/4DHjW+0ynjSW8a+d#YT+f8kHpF9rn6uOvnGN28FjG7HyhhpVTTDepkAovUOL
[18:32:48 CEST] <Mandevil> erick3k: The "-i -" becomes "-i source.ts" obviously.
[18:33:24 CEST] <Mandevil> Also, the "done" is missing.
[18:33:54 CEST] <erick3k> cool so just run it in ssh like while true; do ffmpeg -f -stream_loop -1 mpegts -i source.ts - -flags +cgop -vcodec libx264 -pix_fmt yuv420p -preset $QUAL -r $FPS -g $(($FPS * 2)) -b:v $VBR -c:a copy -bsf:a aac_adtstoasc -f flv "$YOUTUBE_URL/$KEY" 2> /root/ffmpeglog.txt
[18:34:10 CEST] <Mandevil> Done is still missing there.
[18:34:26 CEST] <erick3k> what's done
[18:34:28 CEST] <erick3k> where
[18:34:37 CEST] <Mandevil> while true; do <something>; done
[18:34:45 CEST] <Mandevil> That's the syntax of the while construct.
[18:34:51 CEST] <Mandevil> Also you still have extra '-' there.
[18:34:55 CEST] <erick3k> oh so change do to done?
[18:35:07 CEST] <Mandevil> Append "; done"
[18:35:22 CEST] <Mandevil> And remove the lone '-'
[18:36:04 CEST] <erick3k> like this while true; do ffmpeg; done -f -stream_loop -1
[18:36:10 CEST] <Mandevil> No.
[18:36:21 CEST] <erick3k> hehe sorry man am linux stupid
[18:36:40 CEST] <Mandevil> Like this: https://pastebin.com/tJP54Cmj\
[18:36:46 CEST] <Mandevil> Sans the backslash.
[18:36:48 CEST] <alex88> why `/opt/ffmpeg/ffmpeg -f v4l2 -framerate 25 -video_size 640x480 -i /dev/video0 -vcodec libx264 -pix_fmt yuv420p -profile:v main -level 3 http://127.0.0.1:8090/feed1.ffm` says `No pixel format specified, yuv422p for H.264 encoding chosen.`
[18:36:58 CEST] <alex88> if I've specified -pix_fmt yuv420p ?
[18:37:18 CEST] <erick3k> ok gonna try that
[18:37:45 CEST] <Mandevil> alex88: It looks good to me...
[18:37:54 CEST] <alex88> don't know why it's saying that
[18:37:57 CEST] <Mandevil> alex88: Maybe try to move it towards the end?
[18:38:14 CEST] <Mandevil> Before the feed url.
[18:39:01 CEST] <alex88> Mandevil: same :/
[18:39:34 CEST] <Mandevil> I don't know, I am using it and no problems.
[18:39:51 CEST] <Mandevil> How about using -v:c libx264 instead of -vcodec?
[18:39:54 CEST] <alex88> Mandevil: found it, it was ffserver :/
[18:40:02 CEST] <alex88> had to specify PixelFormat there
[18:40:12 CEST] <Mandevil> Alright.
[18:41:27 CEST] <erick3k> Mandevil, having trouble transforming this $FPS -g $(($FPS * 2))
[18:41:45 CEST] <erick3k> should it be 30 -g ((30 * 2))?
[18:41:56 CEST] <erick3k> or just 30 and delete the things in ()
[18:44:21 CEST] <Mandevil> Transforming?
[18:46:36 CEST] <erick3k> i mean instead of using $FPS to set it
[18:46:46 CEST] <erick3k> i tried like this https://pastebin.com/9g587AfS
[18:46:53 CEST] <erick3k> seems like it runs but nothing on the log
[18:46:59 CEST] <erick3k> and live stream doesn't start
[18:48:07 CEST] <Mandevil> I have no idea with the -g thing does.
[18:48:36 CEST] <erick3k> says Unrecognized option '1'.
[18:49:16 CEST] <erick3k> the stream loop -1
[18:49:25 CEST] <erick3k> is that a number 1 or an L letter
[18:50:14 CEST] <Mandevil> erick3k: -stream_loop -1 mpegts -i
[18:50:17 CEST] <Mandevil> This part is wrong.
[18:51:03 CEST] <Mandevil> The -i parameter specifies input.
[18:51:06 CEST] <Mandevil> And you have none.
[18:51:21 CEST] <erick3k> oh lol
[18:51:22 CEST] <erick3k> got you
[18:51:46 CEST] <erick3k> so source.ts in there
[18:51:56 CEST] <erick3k> but still says -1 not recognized
[18:54:34 CEST] <erick3k> maybe that stream loop doesn't work with ts?
[18:55:15 CEST] <Mandevil> It should.
[18:55:36 CEST] <Mandevil> Documentations says -1 is infinite loop.
[18:55:43 CEST] <Mandevil> I don't know why it doesn't work.
[18:59:33 CEST] <erick3k> am using 2.8.11 maybe is too old?
[19:00:49 CEST] <erick3k> well Mandevil
[19:01:03 CEST] <erick3k> tried the old script, just added the while true think to ffmpeg
[19:01:03 CEST] <erick3k> https://pastebin.com/eQ2tAskA
[19:01:18 CEST] <erick3k> works, but is that correct? would the ffmpeg restart if fails?
[19:01:32 CEST] <Mandevil> No.
[19:01:52 CEST] <Mandevil> The problem is that the inner loop apparent never exits even if the ffmpeg fails.
[19:04:31 CEST] <erick3k> sorry mate half of that sentence is Chinese to me lol
[19:04:55 CEST] <erick3k> inner loop apparent
[19:07:02 CEST] <Mandevil> apparently
[19:13:02 CEST] <thebombzen> have you tried while true; do cat source.ts; done | ffmpeg -f mpegts -i -
[19:13:15 CEST] <thebombzen> mpegts is concatenation-friendly
[19:13:18 CEST] <Mandevil> thebombzen: That is what does not work.
[19:13:31 CEST] <Mandevil> thebombzen: Apparently when ffmpeg fails, the loop just goes on.
[19:14:17 CEST] <thebombzen> then you could try using set -e
[19:14:35 CEST] <thebombzen> or you could try
[19:14:50 CEST] <thebombzen> while cat source.ts; do true; done | ffmpeg -f mpegts -i -
[19:14:55 CEST] <thebombzen> I think that might work
[19:15:18 CEST] <thebombzen> ffmpeg crashing will cause a broken pipe which will cause "cat source.ts" to be killed by a broken pipe signal
[19:15:27 CEST] <thebombzen> which will cause the while loop to stop
[19:15:32 CEST] <Mandevil> thebombzen: That's exactly what I would have assumed to.
[19:15:44 CEST] <thebombzen> well did you try it?
[19:15:47 CEST] <Mandevil> thebombzen: But if erick3k is to be believed, it's not the case.
[19:15:49 CEST] <thebombzen> this, I mean: while cat source.ts; do true; done | ffmpeg -f mpegts -i -
[19:15:59 CEST] <thebombzen> did you try this ^
[19:16:02 CEST] <Mandevil> No.
[19:16:10 CEST] <Mandevil> I am not the one who wants it working :)
[19:16:14 CEST] <thebombzen> haha
[19:16:17 CEST] <thebombzen> erick3k: try this
[19:16:22 CEST] <thebombzen> while cat source.ts; do true; done | ffmpeg -f mpegts -i -
[19:16:24 CEST] <Mandevil> I have my own ffmpeg fights going :)
[19:21:09 CEST] <erick3k> thebombzen that is what i had while cat source.ts; do true; done
[19:21:26 CEST] <erick3k> but when ffmpeg disconnects or failes it doesn't start again
[19:21:32 CEST] <thebombzen> well yes
[19:21:40 CEST] <thebombzen> wasn't that the goal?
[19:21:52 CEST] <erick3k> nop, i want it to start again
[19:22:03 CEST] <erick3k> if for some reason fails
[19:22:24 CEST] <thebombzen> you want to loop source.ts to a network stream?
[19:22:39 CEST] <erick3k> i am looping it to youtube
[19:22:46 CEST] <thebombzen> youtube streaming?
[19:22:56 CEST] <erick3k> want that thing runing 24/7 but randomly disconnects after days
[19:23:03 CEST] <thebombzen> that's not surprising at all
[19:23:08 CEST] <erick3k> so the script keeps runing (the sh) but ffmpeg closes
[19:23:09 CEST] <thebombzen> having a days-long youtube live stream
[19:23:21 CEST] <thebombzen> well have you considred just restarting it in the script >_>
[19:23:30 CEST] <thebombzen> while true; do
[19:23:36 CEST] <thebombzen> <the script thing you mentioned>
[19:23:37 CEST] <thebombzen> done
[19:24:05 CEST] <erick3k> you mean the sh?
[19:24:12 CEST] <erick3k> script.sh file?
[19:24:33 CEST] <thebombzen> yes that's what I meant by the script, I meant the shell script
[19:24:34 CEST] <erick3k>  while true; do script.sh; done?
[19:24:53 CEST] <thebombzen> something like that. or just put the while loop *in the script*
[19:30:24 CEST] <erick3k> like this https://pastebin.com/bNNQdA4b
[19:30:27 CEST] <erick3k> thebombzen
[19:30:51 CEST] <thebombzen> ......................
[19:31:01 CEST] <thebombzen> when I said put the while loop in the script
[19:31:11 CEST] <thebombzen> I didn't mean put the words "while loop" in the script
[19:31:17 CEST] <thebombzen> have you ever written a shell script before
[19:32:19 CEST] <thebombzen> oh you're the dunce from before who is uploading other people's IP to youtube, right?
[19:32:37 CEST] <thebombzen> and is completely incapable of following instructions
[19:32:41 CEST] <thebombzen> I'm out of here
[19:33:16 CEST] <erick3k> no never lol
[19:33:37 CEST] <erick3k> other peoples ip?
[19:33:40 CEST] <erick3k> no am not that stupid
[19:33:56 CEST] <thebombzen> oh you're someone else then
[19:34:10 CEST] <erick3k> you can "follow instruction" when you know how to follow them.....
[19:34:12 CEST] <thebombzen> but still, read a shell script tutorial and then come back
[19:35:28 CEST] <erick3k> na doing this gig for a customer for $10 just gonna tell him i give up, not worth any more time
[19:36:03 CEST] <erick3k> how is it so hard for mmpeg to reconnect once it loses connection is beyond stupid
[19:36:42 CEST] <erick3k> free is free i guess
[20:24:40 CEST] <faLUCE> Hello. I published this C++ library, based on ffmpeg+ x264. It's for h264/aac http-live streaming (Linux)   https://github.com/paolo-pr/laav . If anyone wants to try/test it, he is welcome
[20:56:45 CEST] <SolarAquarion> libavcodec.so.57: undefined reference to `x265_api_get_110'
[20:56:53 CEST] <SolarAquarion> i'm getting this when trying to build ffmpeg
[20:57:42 CEST] <SolarAquarion> oh found the solution
[20:57:47 CEST] <SolarAquarion> needed to clean the srcdir
[21:07:10 CEST] <zeryx_> stupid linux question: If I'm using the static binaries, and I have some really obnoxiously written python package that links to the binaries, what would be the process of letting the system know where ffmpeg is located
[21:07:42 CEST] <zeryx_> I thought something simple in python like os.environ["PATH"] += os.pathsep + ffmpeg_file would work, but it doesn't seem to show up; plus there's probably some other environment variables the installers set
[21:07:55 CEST] <kepstin> "links to the binaries"? that doesn't really make sense
[21:07:58 CEST] <DHE> the word "link" is misleading
[21:08:10 CEST] <DHE> you don't link the binaries. you run/invoke them, or you link to the libraries
[21:08:29 CEST] <kepstin> but in general, putting the directory containing the binaries (not the binary itself) early in the path should work
[21:08:54 CEST] <zeryx_> yea sorry, "let this really obnoxious python package find the ffmpeg binary"
[21:09:26 CEST] <zeryx_> do you need to do anything else besides setting the path variable?
[21:09:34 CEST] <furq> depends how it invokes ffmpeg
[21:09:52 CEST] <zeryx_> I've been looking, really terrible spaghetti code
[21:09:53 CEST] <furq> if it's doing something stupid like calling /usr/bin/ffmpeg then you either need to make a symlink or use a better package
[21:10:33 CEST] <zeryx_> trying to get youtube_dl working btw* https://github.com/rg3/youtube-dl/
[21:10:45 CEST] <furq> oh
[21:10:48 CEST] <zeryx_> I've gotten it to work fine with raw ffmpeg usage, just there's probably some config stuff that's weird
[21:10:51 CEST] <furq> youtube-dl has always been fine for me
[21:11:05 CEST] <furq> i assume it just uses the first ffmpeg on the path
[21:11:07 CEST] <zeryx_> I'm installing from a static build in my prod environment
[21:11:21 CEST] <zeryx_> so my problem right now is figuring out why my little code snippet isn't working properly
[21:11:41 CEST] <furq> there's an -ffmpeg-location switch
[21:11:47 CEST] <furq> --
[21:12:12 CEST] <zeryx_> https://gist.github.com/zeryx/00173fcd8a15ccee7f640e468b08038e
[21:12:26 CEST] <zeryx_> I may have made the wrong decision and tried to use youtube_dl as a python package
[21:12:34 CEST] <zeryx_> I assumed it wouldn't be this difficult to do :D
[21:12:51 CEST] <csucsu99> I'm extracting opus audio from webm containers like this: ffmpeg -i 1.webm -vn -acodec copy a.opus.
[21:13:06 CEST] <furq> zeryx_: what, as a library?
[21:13:10 CEST] <zeryx_> yea
[21:13:14 CEST] <csucsu99> If I do it again, ffmpeg -i 1.webm -vn -acodec copy b.opus, and "compare" a.opus with b.opus the output differs.
[21:13:15 CEST] <furq> i didn't even know you could do that
[21:13:15 CEST] <zeryx_> it's been a tough fight
[21:13:19 CEST] <csucsu99> Any idea why?
[21:13:32 CEST] <furq> csucsu99: compare?
[21:13:41 CEST] <csucsu99> Yep, with cmp, binary comparison.
[21:13:42 CEST] <furq> oh
[21:13:44 CEST] <zeryx_> furq: it's just a regular python project, but yea it hasn't been an easy process, the source itself is a patch work of patch works
[21:13:44 CEST] <kepstin> csucsu99: when writing ogg containers (opus uses ogg container), a random stream id is chosen each time, so things like concat work properly
[21:13:50 CEST] <furq> yeah
[21:14:03 CEST] <zeryx_> it's nearly impossible to actually understand the process flow IMO
[21:14:04 CEST] <furq> if you want to compare the streams then use the hash muxer
[21:14:06 CEST] <furq> !muxer hash
[21:14:07 CEST] <nfobot> furq: http://ffmpeg.org/ffmpeg-formats.html#hash-1
[21:14:20 CEST] <csucsu99> Oh, cool, thanks very much!
[21:14:55 CEST] <furq> you should use that for most containers
[21:15:04 CEST] <furq> a lot of things will write timestamps as metadata and so forth
[21:15:18 CEST] <zeryx_> furq: you sure -ffmpeg-location is a config option? I can't find any reference to it in the repository
[21:15:27 CEST] <furq> https://github.com/rg3/youtube-dl/blob/master/README.md#readme
[21:15:29 CEST] <furq> it's listed in there
[21:16:12 CEST] <zeryx_> rofl
[21:16:13 CEST] <csucsu99> Any way to avoid the "random" stream id? For my use case that would be easier, but the hash-muxer isn't half bad.
[21:16:28 CEST] <zeryx_> yea furq it doesn't actually exist in the current source, unless they do more funky stuff
[21:16:33 CEST] <kepstin> csucsu99: iirc, there's an ffmpeg flag that does that, but I can't remember it offhand
[21:17:05 CEST] <furq> zeryx_: i'm on the latest and it works for me
[21:17:14 CEST] <furq> i just set it to a nonexistent location and it broke
[21:17:29 CEST] <furq> csucsu99: maybe -fflags bitexact
[21:17:54 CEST] <furq> not sure if that does anything with ogg though
[21:19:11 CEST] <csucsu99> Ah, works like a charm man!
[21:19:18 CEST] <csucsu99> That did the trick!
[21:19:31 CEST] <furq> i should never have doubted myself
[21:19:47 CEST] <csucsu99> :D Thanks! Now I can get back to work. Cheers!
[00:00:00 CEST] --- Sat Apr 29 2017