The last two days I've rewritten a large part of the bytecode in ttfautohint to use twilight points for representing segments.
FreeType's autohinter treats segments as if they were a single point, located in the middle of the maximum and minimum coordinate value. The previous code in ttfautohint simplified this by using the first point of a segment (pars pro toto), and its location was saved in a storage area location. However, the deviation compared to the autohinter results was rather large. I've now fixed this by properly computing the middle positions for each segment, and I use two sets of twilight points to store these representations: The first set holds the original coordinates, the second one the hinted values. The reason for this is that all twilight points are originally located at (0,0), thus instructions like `MD[orig]' to access the original coordinates would always refer to (0,0) which is not really helpful. Unfortunately, I'm still behind my schedule, but there actually is a lot of progress! http://repo.or.cz/w/ttfautohint.git Werner _______________________________________________ Freetype-devel mailing list Freetype-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/freetype-devel