From Wiki:

Quote:7z's LZMA algorithm reaches a higher compression ratio than RAR, except for "multimedia" files like .wav and .bmp files where RAR uses specialized routines that outperform LZMA

I guess those VNs have a lot of uncompressed media.

Quote:Version 3 of RAR is based on Lempel-Ziv and prediction by partial matching (PPM) compression, specifically the PPMd implementation of PPMII by Dmitry Shkarin.

Ah, so that explains why RAR outperforms LZMA on text, though 7z's PPMd still seems to outperform RAR.

I believe LZMA just uses a dictionary (modified version of LZ77) » entropy (adaptive arithmetic coding) coding chain.  IDK what RAR uses, but apparently it's based off deflate, which is also a dictionary (LZ77) » entropy (huffman) coding chain.
Huffman is old and only really was used in older formats because arithmetic coding was patented (expired in 1999), so I would expect RAR to use arithmetic coding too.
Funny that I heard LZMA2 was supposed to beat RAR consistently.  I guess that's not the case.

(This post was last modified: 31/08/2010 06:25 PM by ZiNgA BuRgA.)