> Xft doesn't do anything weird with Freetype, but the core server does -- > it's got its own version of Freetype 1 embedded inside it. Xft uses the > system version of Freetype 2. I suspect the copy that Xft is using still > has the interpreter disabled. Use 'ldd' to see what copy Xft is using and > make sure that has the interpreter enabled. > > [EMAIL PROTECTED] XFree86 Core Team SuSE, Inc.
After hearing that Xft uses Freetype2 and X uses Freetype1, I went and downloaded the test programs from both. The ftstring program in each source tree bypasses the X font rendering, and should give a good idea of the quality of each library. For the sake of comparison, the disabled AA in the ftstring.c source file of FT2-demo. I was shocked to find that Freetype1 rendered the Verdana font much better than Freetype2. I went and looked to see if the bytecode interpreter in FT2 was on, and apparently there was a typo (an underscore was missing). I recompiled with the right #define, and the quality got even worse! Wheras there had previously been just some roughness around the edges of the font, now it looked like every other scanline had been offset by one pixel (sorry I don't have a screenshot, KSnapshot doesn't like png's and jpegs are useless. I'll try to post one if I can get it). Then I enabled AA for the FT2 demo, and wow, that's some aweful AA. According to the FT2 docs, FT2 only AA's parts that need it, like diagnol lines. However, every line in a 'T' is anti aliased, which makes the whole thing look terribly fuzzy. Is it a known problem that FT2 has worse quality that FT1, or am I doing something wrong. Could the fact that I'm using GCC-2.96 have something to do with it? _______________________________________________ Render mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/render
