Content deleted Content added
Simon G Best (talk | contribs) →Range encoding and arithmetic encoding: On correcting and clarifying the relationship with arithmetic coding. |
The rewrite |
||
Line 30:
I've now edited the article to bring it into line with the [[arithmetic coding]] article (which I've also edited). This is to correct and clarify the relationship between arithmetic coding and range coding, since they're really just two, slightly different ways of understanding the same thing.
--[[User:Simon G Best|Simon G Best]] 15:54, 21 July 2006 (UTC)
:I don't really like the change. It confuses to the patent issue, which, as far as I can tell, is the only reason anyone cares about range coding. The reason range coding is believed to be without patent issues is because (again, as far as I can tell) it's an older form of arithmetic coding. Thus it was prior knowledge by the time any patents could have been filed that wouldn't have expired by now. The claim that arithmetic coding is a form of range coding seems wrong to me, since range coding claims to be a specific method, a fixed procedure to relate input to output. In contrast, arithmetic coding is a family of methods and, although it initially related specific input to specific output, it now has variants which trade off compression ratio (output) for performance (speed), and which consider how the probability models should adapt. The problem with arithmetic coding then is that there are lots of variants, so, while some forms of arithmetic coding are under patent, others aren't. So relabeling a specific type of arithmetic coding as "range coding" has the benefit that people know, "This is the type of arithmetic coding that can be done without worrying about patents." Of course, this is no small feat, as patent issues have prevented the adoption of arithmetic coding in applications where, technologically speaking, it would have significantly helped. [[User:Calbaer|Calbaer]] 22:10, 21 July 2006 (UTC)
| |||