was reading about asymmetric number systems and I thought I'd ask since I have some compression people following me: how do you use ANS when your probability distribution changes as you compress?

all I've seen so far are fixed probably distributions, but for range coding and similar the distribution can change over time. how does ANS handle that, like with the apparently Super Fast tabled variant?


I read through the paper again and I guess what I'm asking for is qANS

Sign in to participate in the conversation

cybrespace: the social hub of the information superhighway jack in to the mastodon fediverse today and surf the dataflow through our cybrepunk, slightly glitchy web portal support us on patreon or liberapay!