What's the future of compression?

What's the future of compression?

get more specific, faggot

Are we at the limit of what losless compression can do already?

Whts t ftre of cmprssn?

kk'd

we may be able to see improvements using AI but it won't be as crazy as what's in that show

?ftr/cmprsn

i don't know, let's figure it out by discussing the speed at which we can jerk off everyone in a room!

fucking love this show

Outlawed. Compression is technically a form of encryption, and encryption is banned now.

I've already developed a superior compression algorithm to what's out there. Assign "1" to "11" and 0 to "00". Roughly cuts the file size by 66%. All I need to do is implement it. I'm learning C and ASM now.

yeah, kinda got stale after s3 though, if the t*rkish leaks are to be believed, s4 is gonna be s3 again so i might just pass on it

Fractal compression is the way forward; restoring terabytes of data from just a couple of lines of math that expand into the original data like a fractal.

lol baby's first ascii compression

>ASCII
Brainlet detected. I'm talking about working on the level of raw bits, something I'm sure you know nothing about. Leave this discussion to the larger minds.

I don't know but the fucking programmers who do it won't be using sublime text.

And why not?

Hypertheory goobs like high investment/high return tools like vi/vim/emacs that allow them to do insane operations after an insane amount of time learning them.

how are you gonna get 01 and 10 einstein?

This sounds interesting.

I'm going to ignore OP's post, because OP is a faggot.

Lossless compression? Probably future developments in things like fast finite-state entropy coders, ways to make optimal arithmetic/range coding faster... since lz4, Snappy and zstd, there's really been a boom in research into fast software compression again. It's possible there could also be some new techniques making context mixers more practical.

Lossy audio compression probably doesn't really have anywhere really useful left to go. The techniques combined together and used in Opus were quite spectacular and, especially with freq-bucket long/short block switching, most of the problem samples went away. It's low-delay, patent-free (if you ignore a couple of already-prior-arted trolls), yet reaches transparency sooner than everything else. Remarkable work.

There's still a very long way to go in pictures and video. HEVC was mainstream but solid; Thor and particularly Daala were interesting experiments and AV1 is something immediately useful out of their early stages; I really look forward to seeing where AV2 goes, and appreciate Xiph's willingness to look a little outside the mainstream. I think we're going to need a fair bit more processing power (and better entropy encoding) to realise some of the more blue-sky stuff.