Andre Poenitz wrote:
On Wed, Jan 25, 2006 at 09:43:30PM +0100, Kevin Pfeiffer wrote:
I don't see how you arrive at that. If I assume perhaps twenty or so
ranges of 256 chars. for a very large font family (including greek
letters, punctuation and more -- and I assume that this would cover CM,
Knuth's font and the one I am using) I come up with roughly 5K
characters. And it seems to me that the kerning pairs would be 5K to
the power of 2 (not 2 to the power of 5K) -- though I admit that as a
non-mathematician I only arrived at this by putting five letters on a
page and adding up the possible pairs (AA, AB, AC, AD, AE, BA, BB,
etc.); this gave me 25, not 32.
So even this gives you 25 million combinations. I doubt anybody sits down
and writes down special kerning rules for all of them.
Writing kerning rules for each possible pair is silly. It is better to
take advantage of the fact that many letters kern the same way.
they can then be grouped to take advantage of this.
For example, various variants of "A" (ÂÄÀÁÅÃ) can all kern
the same way in all cases. They form an "A"-group.
Helge Hafting
