- From: Levantovsky, Vladimir <Vladimir.Levantovsky@MonotypeImaging.com>
- Date: Tue, 30 Jun 2009 09:56:44 -0400
- To: "John Daggett" <jdaggett@mozilla.com>, "Joseph Felps" <jfelps@gmail.com>
- Cc: <www-font@w3.org>
On Tuesday, June 30, 2009 5:03 AM John Daggett wrote: > > As suggested by several folks, tried LZMA. Great stuff! > > For basic webfonts like Arial, Georgia, etc. MTX compression is 8-14% > better than LZMA but for Cleartype fonts LZMA is dramatically better > than MTX, the MTX compressed version is larger by 30-40% for some of > these fonts. Likewise, CJK fonts for the most part seem to compress > better: > > MS Gothic (Japanese) > raw: 8,272,028 > gzip: 4,441,566 > eot: 4,276,642 > lzma: 2,958,214 (36% of original!) > This is really impressive result, which means that there could be a significant room for MTX improvement if there is any interest. I am sure the MTX pre-processing and post-processing steps combined with LZMA compressor (instead of currently used LZCOMP) would produce even better results for fonts. Vladimir > Updated spreadsheet: > > http://spreadsheets.google.com/pub?key=rKT_wNzraVrkXQcKSWb- > jTA&output=html > > Cheers, > > John > > ----- Original Message ----- > From: "Joseph Felps" <jfelps@gmail.com> > To: www-font@w3.org > Sent: Tuesday, June 30, 2009 12:59:12 PM GMT +09:00 Japan > Subject: Re: gzip vs. mtx compression ratios > > What about also looking at LZMA for comparison. Average compression > ratio is 30% better than gzip: http://tukaani.org/lzma/ . > > There is also some benchmarks comparing gzip, bzip2, and lzma. > http://tukaani.org/lzma/benchmarks > >
Received on Tuesday, 30 June 2009 13:57:21 UTC