[Ffmpeg-devel-irc] ffmpeg.log.20190519

[burek]

[00:17:43 CEST] <Soni> no, just ffmpeg
[00:19:00 CEST] <Soni> look, ppl have achieved recursion with PowerPoint. and punchcards. if PowerPoint can do it, why can't ffmpeg?
[00:19:15 CEST] <Soni> isn't lavf turing-complete anyway?
[00:23:00 CEST] <kepstin> i think the tblend filter is turing complete, fwiw
[00:23:24 CEST] <kepstin> er, maybe not
[00:23:40 CEST] <kepstin> since it uses two frames from the source video, not the previous filtered frame and a new frame
[00:24:39 CEST] <kepstin> if you just want to programmatically generate video within ffmpeg, geq can do it
[00:26:10 CEST] <kepstin> huh, actually, ffmpeg expressions do support persistent variables now with ld() and st() so I suppose they can do arbitrary calculations over frames.
[00:26:22 CEST] <kepstin> only 10 variables tho :)
[00:32:16 CEST] <Soni> can it seek?
[00:32:40 CEST] <Soni> (i.e. 10 registers, but infinite video tape)
[00:32:57 CEST] <Soni> ("infinite")
[00:34:21 CEST] <kepstin> nah, purely linear. you'd be feeding it a blank video source, then geq generates values for each pixels in one frame, then the next frame, and so on.
[00:35:57 CEST] <snap1> hi i am doing a survey:  what is the top 10 third party application do you install  after doing a clean installation of windows?
[00:36:47 CEST] <kepstin> it's hilarious that with current versions of windows 10, "Linux" is actually a legitimate possible answer to that.
[01:21:56 CEST] <Soni> linux and x11
[01:22:08 CEST] <Soni> wait, linux is first party
[01:23:28 CEST] <snap1> i am getting crappy answers
[01:23:32 CEST] <snap1> in this channel
[01:23:50 CEST] <klaxa> not many people install windows in this channel i think
[01:24:50 CEST] <snap1> but ffmpeg is multimedia related
[01:25:42 CEST] <klaxa> but most windows software uses gui wrappers
[01:26:23 CEST] <klaxa> and most users in here interact with ffmpeg directly through the libraries or through the cli program
[01:27:05 CEST] <klaxa> and that's more akin to people comfortable in the command line which is usually not windows users
[01:27:20 CEST] <Ariyasu> i use ffmpeg from windows cli
[01:27:29 CEST] <snap1> i like  command line because it's scriptable
[01:27:31 CEST] <Ariyasu> but made my own scritps to handle common tasks with it
[01:27:37 CEST] <iive> i think he meant that you can install linux kernel inside win10
[01:27:42 CEST] <snap1> hard to script a  GUI application
[01:27:57 CEST] <iive> and use it to run linux programs.
[01:29:15 CEST] <klaxa> (because apparently translating is more expensive than emulating)
[01:29:46 CEST] <klaxa> kind of makes sense though, i guess the overall call-stack will shrink
[02:07:47 CEST] <snap1> do you think we will ever reach a video resolution and people will say,  okay this is good enough,  let's stop now?
[02:11:24 CEST] <DHE> hopefully 8k will be the end of it...
[02:11:30 CEST] <DHE> but maybe 3d will make a comeback?
[02:11:32 CEST] <DHE> :)
[02:12:25 CEST] <snap1> 8k is    4320p ?
[02:13:25 CEST] <snap1> DHE  so you don't after 8k is standard,  people won't want 16K ?
[02:13:40 CEST] <snap1> DHE  so you don't think after 8k becomes standardized,  people won't want 16K ?
[02:16:16 CEST] <DHE> besides the practicality of human eye resolution, it's not a matter of what people want. these things are hard to make (failure rates, etc) and expensive. the data transfer standards need to exist - some TV still don't do 4k at 60hz yet or are limited to certain HDMI ports, etc.
[02:17:01 CEST] <snap1> DHE okay let me make a modification of my question
[02:17:11 CEST] <snap1> DHE  so you don't think after 8k becomes standardized,  people won't want 16K  in year 2200
[02:18:31 CEST] <tdr> DHE, it will make a comeback once a few chip prices come down and the active glasses massively shrink
[02:20:12 CEST] <DHE> and yes, 8k is a measurement of width. 1080p would be 2k
[02:20:20 CEST] <DHE> where 1920 is close enough about 2k
[02:20:53 CEST] <snap1> DHE  so you don't think after 8k becomes standardized,  people won't want 16K  in year 2200 ?
[02:22:47 CEST] <DHE> I don't want to speculate about what would happen that far in the future. personally I'm hoping for some kind of revolution where tvs don't have discrete pixels anymore or something like that
[02:25:04 CEST] <snap1> DHE hmm, then what would it depend on
[02:25:12 CEST] <snap1> if it doesn't descrete pixels
[02:28:57 CEST] <kepstin> possibly have an analogue electron beam tracing across a continuous field of phosphers?
[02:29:01 CEST] <DHE> no idea. science does crazy things.
[02:29:05 CEST] <snap1> DHE audio has seems to hit the max  with  24bit/192khz
[02:29:22 CEST] <DHE> I recall a camera that took a picture using only a single photosensor. so reverse that concept
[02:29:51 CEST] <snap1> DHE  thanks for the reply to my question
[02:30:04 CEST] Action: kepstin got to use some short throw laser projectors at an event recently, those are pretty neat tech.
[02:30:28 CEST] <snap1> kepstin  are you saying laser projectors doesn't use pixels ?
[02:31:06 CEST] <kepstin> their input is a raster image, i don't know much about how their output actually works
[02:31:09 CEST] <DHE> not in the traditional sense. but they're still generating a picture from input in a grid
[02:40:20 CEST] <snap1> people/human/civilazation has  managed to  digitalize  video and audio but  do you think we will ever able to digitalize  odor ?
[02:41:27 CEST] <another> probably yes
[02:41:54 CEST] <snap1> is that even possible
[02:42:11 CEST] <snap1> i am download "banana odor" to file   banana.odor
[02:42:23 CEST] <another> suck in air, analyze the chemicals, store the result
[02:42:27 CEST] <snap1> and playback  banana.odor
[02:43:14 CEST] <snap1> then why hasn't been done already?
[02:43:35 CEST] <another> i guess the analyze step is hard
[02:44:37 CEST] <snap1> playingback  file.odor  seems  really hard
[02:46:04 CEST] <another> have a bunch of cartridges with different chemicals, synthesize odor from chemicals
[03:01:58 CEST] <DHE> there's gonna be a new wave of trolling...
[03:17:53 CEST] <hanetzer> DHE: lemon party 2.0
[03:17:57 CEST] <hanetzer> where you get to smell it
[03:29:00 CEST] <snap1> what is lemonparty 1.0 ?
[03:34:42 CEST] <furq> ask your granddad
[03:35:49 CEST] <steve___> :)
[03:54:22 CEST] <DHE> there are some things that should NOT be google'd
[07:10:08 CEST] <snap1> why is it so hard to record both mic-in and  desktop-audio at the same time in windows OS?  does macosx/linux have same issue?'
[07:11:00 CEST] Last message repeated 1 time(s).
[17:08:04 CEST] <snap1> i know how to convert "a file" to opus  ffmpeg -i file.flac  -c:a libopus -b:a 128k file.opus,  but is ffmpeg allow creating opus from my mic input?
[19:48:53 CEST] <DHE> snap1: Yes, look at https://ffmpeg.org/ffmpeg-devices.html and see what suits your needs based on OS and stuff
[19:49:13 CEST] <snap1> DHE okay thanks
[20:34:06 CEST] <snap1> cehoyos : who is this person: this person is not very nice
[22:21:38 CEST] <Ristovski> Is there a reason why kmsgrab produces darker video output vs x11grab?
[23:15:43 CEST] <AndrewR> anyone here tried re-muxing intel's SVT-AV1 (0.5.0 release) encoder output into webm/mp4 with recent ffmpeg? ffmpeg -i out-1.ivy -vcodec copy -v verbose  av1_svt_0.5_1000frames.webm works without errors, but resulting webm apparently not very seekable (unfortunately I only have patched by me video editor, so I can't say for sure if such non-seekability is ffmpeg's remuxing fault, or my own fault)
[23:33:51 CEST] <AndrewR> [matroska,webm @ 0xf0c004c0] Element at 0xb76f83 ending at 0xb76f94 exceeds containing master element ending at 0xb76f7e (with latest ffmpeg/ffplay)
[00:00:00 CEST] --- Mon May 20 2019