[Ffmpeg-devel-irc] ffmpeg.log.20181117

[burek]

[01:56:28 CET] <microcolonel> https://en.wikipedia.org/wiki/LDAC_(codec) This codec makes me angry
[02:10:04 CET] <furq> high-resolution lossy audio is very important
[03:26:49 CET] <microcolonel> furq: but choosing the "24-bit" depth and "96kHz" bandwidths is ridiculous
[03:27:00 CET] <microcolonel> either they just say that, and it doesn't actually mean anything
[03:27:13 CET] <microcolonel> or they've designed tremendous inefficiency into their codec
[03:27:15 CET] <microcolonel> for no good reason
[03:27:18 CET] <furq> you see, it's high resolution, and therefore [mumbles for 90 seconds]
[03:27:41 CET] <microcolonel> they actually brag about the data rate
[03:28:07 CET] <microcolonel> as though it's a good thing for it to be using twice as much duty cycle on the radio
[03:28:33 CET] <microcolonel> scratch that, more like 2.5x competing standards
[03:29:38 CET] <furq> don't sony make a gold plated walkman-branded pmp that plays dsd512
[03:29:46 CET] <furq> that presumably costs one million dollars
[03:30:00 CET] <furq> this is very on brand for them
[03:30:49 CET] <furq> https://www.amazon.co.uk/Sony-NW-WM1Z-High-Resolution-Walkman-Multi-Touch/dp/B01LHGLALI
[03:30:53 CET] <furq> only £2500. sorry sony
[03:31:14 CET] <microcolonel> oof
[03:31:31 CET] <microcolonel> yeah DSD is... shocking
[03:31:47 CET] <microcolonel> like, that I can sorta allow
[03:31:56 CET] <microcolonel> since it generally only shows up in audiophool products
[03:32:07 CET] <microcolonel> they're subsidizing us with their idiocy
[03:32:18 CET] <microcolonel> whereas this is being integrated into a lot of devices
[03:32:36 CET] <furq> presumably everything still supports aptx or whatever
[03:32:44 CET] <furq> and knowing sony they'll be asking some crazy licensing fee
[03:32:56 CET] <microcolonel> nah
[03:33:00 CET] <microcolonel> at least they made this one royalty free
[03:33:05 CET] <microcolonel> because they make headphones which use it
[03:33:14 CET] <microcolonel> the encoder is here https://android.googlesource.com/platform/external/libldac
[03:33:14 CET] <furq> ah, of course
[03:33:28 CET] <microcolonel> gives me an idea though I guess
[03:33:34 CET] <microcolonel> Opus over A2DP
[03:33:35 CET] <microcolonel> I mean why not
[03:34:30 CET] <furq> i'll just stick with a cheaper phone and wired headphones
[03:34:32 CET] <microcolonel> market it as variable-bandwidth 32-bit floating point
[03:34:38 CET] <furq> lol
[03:34:39 CET] <microcolonel> :- )
[03:34:49 CET] <furq> that's what i tell everyone about my 128k fraunhofer mp3s
[03:35:13 CET] <microcolonel> True 32-bit audio :^ )
[03:36:08 CET] <microcolonel> I guess the one nice thing about ldac is complexity
[03:36:14 CET] <microcolonel> I'm sure it costs practically nothing to encode
[03:36:33 CET] <microcolonel> and the decoder is like... too small to manufacture as a separate chip
[03:36:43 CET] <furq> at least it's not atrac
[03:36:53 CET] <microcolonel> Unicode needs a cringe emoji
[03:36:59 CET] <microcolonel> for this exact moment
[03:37:13 CET] <furq> i microwaved a sonicstage cd once
[03:37:15 CET] <furq> it felt pretty good
[03:37:30 CET] <furq> i was half expecting it to tell me it wasn't compatible with the microwave
[03:38:12 CET] <microcolonel> :- )
[03:38:25 CET] <microcolonel> I liked that joke so much that I even giggled a little.
[03:39:16 CET] <microcolonel> furq: apparently LDAC actually has a 32-bit floating point mode
[03:39:18 CET] <microcolonel> kill me onw.
[03:39:19 CET] <microcolonel> *now
[03:56:35 CET] <microcolonel> Anyway, going to see how hard it is to make a bluetooth device that implements A2DP
[03:56:40 CET] <microcolonel> and test out embedding Opus
[04:05:50 CET] <microcolonel> https://www.sony.com/electronics/play-music-library-near-hi-res-audio-sound-quality furq this is somehow worse
[04:06:33 CET] <microcolonel> pages and pages of bullshit
[04:08:24 CET] <furq> you know it's advanced technology when the charts have no labels on the axes
[04:09:47 CET] <furq> i like the link to "Hi-Res Buzzwords" with no additional context
[04:12:09 CET] <microcolonel> yeah, it's amazing too
[04:12:18 CET] <microcolonel> I thought that was a broken alt text for an image
[04:12:21 CET] <microcolonel> some sort of word cloud
[04:12:37 CET] <microcolonel> the alt text being more honest than they otherwise are
[04:13:36 CET] <furq> Lossless compression allows the original data to be reconstructed almost perfectly from the compressed data achieved by a class of algorithms.
[04:13:39 CET] <furq> what
[04:15:17 CET] <microcolonel> You know how people of medium to high intelligence know that when things are written in a complicated fashion, they're probably either lies or something similar?
[04:15:28 CET] <microcolonel> I think that there's probably a special class of person
[04:15:50 CET] <microcolonel> who is just dumb enough to not notice they're being conned
[04:16:03 CET] <microcolonel> and just smart enough to be able to afford the equipment
[07:30:47 CET] <jin> is there a fool proof way of going from transparent png to alpha webm?
[07:30:55 CET] <jin> I been trying ffmpeg -framerate 30 -f image2 -i %03d.png -c:v libvpx -pix_fmt yuva420p -auto-alt-ref 0 out.webm
[07:31:23 CET] <jin> even though I made the green transparent, it's still looks green in my webm
[07:31:34 CET] <furq> what are you playing it back in
[07:31:51 CET] <jin> mpv
[07:32:06 CET] <jin> oh
[07:32:07 CET] <furq> you need --vd=libvpx for alpha in webm
[07:32:10 CET] <jin> my browser shows it transparent
[07:32:29 CET] <furq> yeah the ffmpeg vp8 decoder doesn't support alpha
[07:32:43 CET] <furq> --vd=libvpx forces the vpx decoder which does
[07:37:36 CET] <jin> furq: im doing -c:v libvpx isnt that redundant?
[07:37:47 CET] <furq> that's for encoding
[07:38:04 CET] <furq> --vd=libvpx is an mpv option
[07:38:12 CET] <jin> now i get it
[07:38:35 CET] <jin> weird, still showed green but
[07:38:38 CET] <jin> whatever, works in the browser!
[07:38:44 CET] <jin> thats what i was going for anyways
[07:39:07 CET] <jin> now I juts need an easier way instead of making every frame transparent from imagemagick
[07:39:47 CET] <furq> ffmpeg has a chroma key filter
[07:39:59 CET] <jin> yeah I tried ffmpeg -i in.mp4 -vf "colorkey=green:0.1:0.2" -c:v libvpx -b:v 1M -auto-alt-ref 0 out.webm
[07:40:17 CET] <jin> didn't quite work
[07:40:26 CET] <jin> or maybe it was the browser
[07:40:49 CET] <furq> you might just need to tweak the values
[07:41:03 CET] <furq> https://ffmpeg.org/ffmpeg-filters.html#colorkey
[07:41:05 CET] <jin> is 0.1 and 0.2 a fuzz?
[07:41:11 CET] <furq> yeah
[07:41:19 CET] <jin> with imagemagick im using 15%
[07:41:25 CET] <jin> so thats pretty much the same right
[07:41:45 CET] <furq> maybe, i've not used im in a while
[07:42:12 CET] <jin> got it!!!!
[07:42:15 CET] <jin> ffmpeg -i green_screen_video.mp4 -vf "colorkey=#12fd00:0.1:0.2" -c:v libvpx -b:v 1M -auto-alt-ref 0 alpha_webm.webm
[07:42:19 CET] <jin> works beautiful
[07:42:30 CET] <furq> nice
[14:14:14 CET] <livingbeef> Trying to use vaapi, I get this message: Failed to initialise VAAPI connection: -1 (unknown libva error).  Failed to set value '/dev/dri/renderD128' for option 'vaapi_device': Input/output error. Is that some sort of permission error?
[14:16:16 CET] <livingbeef> vainfo lists VAEntrypoingEncSlice for H264Main/High and others...
[23:25:33 CET] <AntumDeluge> I'm trying to figure out if it is possible to capture audio from a single application on Windows with dshow (or any other method). Haven't been able to find an answer on the internet yet. The problem is I don't want system notification sounds or sound from any other running app included in the output.
[23:25:37 CET] <AntumDeluge> I'm on Windows 10.
[23:27:22 CET] <AntumDeluge> If it's not possible with FFmpeg, I can try to figure out how to mute everything else.
[23:28:40 CET] <AntumDeluge> I tried muting "system sounds" in the Volume Mixer, but some app notifications still make noise. Perhaps just disabling all other background apps while recording is the answer.
[00:00:00 CET] --- Sun Nov 18 2018