[comp.fonts] bitmaps into splines

woody@rpp386.cactus.org (Woodrow Baker) (02/01/90)

In article <1027@umb.umb.edu>, karl@umb.umb.edu (Karl Berry) writes:
> 
> Besides the Plass & Stone article just mentioned, there is
> Philip Schneider's thesis at the University of Washington.
> (In Seattle.) He describes a system which allows interactive
> curve manipulation, and which is based in part of Plass & Stone's
> ideas.
> 
> When we were writing a bitmap-to-spline program for GNU,
> we found his thesis more helpful, in general, than the article.
> (It wasn't quite so condensed, for one thing, as one might expect.)
> Our experience so far, though, has been that actually
> getting the right curves out of the program (for letterforms)
> is much harder than getting curves that are ``almost right''.
> 
> Furthermore, a spline description does not have enough information
> for acceptable rasterization (except perhaps at typesetter
> resolutions) of letterforms.  More information is
> needed to ensure balanced stroke weights, curves, etc., etc.
> Letterforms are harder to make look right than general
> shapes.
> 
> Karl   karl@cs.umb.edu   ...!harvard!umb!karl

I have been working on this problem as well.  I'd like a copy of Scheiders
thesis, where can I get one?  Given the recursive subdivision algorthym for
drawing splines, which results in an array of point, how difficult is it
to work backwards from that array of points to the original 4 pairs of
coordinates.  You know the number of recursive calls, and the number of
coordinates (x and y pairs) are known.

Cheers
Woody