kerning in slanted/unslanted founts

[Lars Hellström]

Alan wrote:
>I'm not sure why the kern values should change when you slant a font,
>since horizontal displacement is invariant under slant transforms.  The
>italic correction I can understand (although trying to guess the italic
>correction for a glyph based on just bounding box infor is impossible).
>But I don't see why kern info should change.

You could argue that although the horizontal distance is invariant, the
shortest distance need not be. Unslanting two letters that are normally
slanted 1/4 and after unslanting become vertical along the side they face
each other in some given word would increase their optical distance by a
factor $\sqrt{\frac{17}{16}} \approx 1.0308$. Of course, these 3% aren't
visible if you just count the kern width, but it might well make a
noticable addition to the kern if the entire change of optical distance has
to be compensated by the kern.

Still, there is nothing one could possibly require fontinst to do about it,
since there is no way for fontinst to find out what the optical distance
may be.

Besides, there is no universally correct way of determining how large a
kern should be AFAIK, so the differences in kerning Vladimir sees may just
as well be caused by two font designers not agreeing on what is the size of
the best kern.

Lars Hellström