Re: [ft-devel] Supersampling of TrueType fonts, or: slight native hinting

[Werner LEMBERG]

> MIRP is commonly used to modify the width of a stem, using the
> previously grid-fitted point/edge of the other side of the stem as
> the reference.

The main issue is the `commonly', I guess...

> The results aren't perfect and DejaVu's hinting breaks, but it might
> be a first step.

I think that in parallel we had to maintain a database of fonts that
don't follow the `common' MIRP usage.  The main question is how large
such a database would become.  And what about fonts that we don't
know, and that are displayed in a completely broken way?


    Werner