Re: medieval font design

[Juergen Reuter]

On Thu, 1 Feb 2007, Till Rettig wrote:

> Hei,
>
> I am slowly evolving in my ideas and trials on the "ancient" fonts for 
> lilypond. So far I have a couple of good scans from which I would like to 
> take the shapes. But how is the best to get them into metafont? I found 
> something about mfpic (probably not good in this case), fig2mf (sounds 
> already better) and ps2mf (this might be the best?). But are they good, can 
> they do the job? I am so far imaginating that I would take the scans as a 
> picture and then kind of draw around them inside of a vector programm. This 
> would give them at least a really "handwritten" lining. The other idea that 
> somehow sounded logical to me was the way suggested in the fontforge 
> tutorial: designing a glyph by setting points at the outline of the glyph 
> until there are enough points to describe the form. But how is this done for 
> instance in xfig? (As far as I understand fontforge doesn't export to fm).
> I acknowleged the fact that I will have to learn something about metafont 
> anyways, I thought I would go through the metafont tutorial by Christophe 
> Grandsire, this is old but since the program seems to be the same... At the 
> moment it just seems that I won't be able to draw the wanted forms 
> nonvisually, that is in the way that I would just write a mf file.

Hi, Till!

My personal experience is that you have to fine tune the glyphs anyway; 
hence, determining just a few coordinates should suffice (this can be 
easily done manually if you have a really big printout of the scan).

> Then about the single parts of the font (now for some white and black 
> mensural notation): I will create noteheads and stems extra, but since the 
> stems in some cases will have a form like the stem from the ! sign (bigger at 
> top), they won't fit together with the flags. So should there be a separate 
> flag + stem or rather a stem-flag combination?

For each glyph, a charbox must be specified, such that lily knows how 
wide/high the glyph is.  In practice, a glyph may stick out of the charbox 
(even though you probably should have a good reason to do so).  In this 
case, I think sticking out glyphs could be useful; see attached drawing 
(suppose that the blue curve represents the stem, the red curve the flag, 
and the dotted lines the corresponding (simplified) charboxes).

However, as far as I know, stems are currently drawn dynamically by 
just creating rectangles on the fly, rather than outputting a fixed 
glyph.  The reason for doing so is that the actual stem length in general 
depends on a lot of other things, and therefore there is no fixed stem 
glyph.  I guess the stem printing code in lily/stem.cc would need to be 
slightly revised to enable printing of stem glyphs.  Or, even better, if 
you could express the stem by a polygonal shape and, given a stem length, 
devise a method for determining this polygonal shape's coordinates, then 
polygonal stems could be drawn on the fly in lily/stem.cc, analogous to 
the rectangular stems that we currently have.

> And still the idea about introducing some variability to mimick the 
> handscribe, that is to have about four or five slightly different glyphs that 
> would be used in arbitrary order. Is something like this possible in 
> lilypond?

Should be basically possible, if you modify lily/stem.cc accordingly. 
However, I guess the result would not look beautiful, since the 
"arbitrariness" of hand-writings actually comprehends a very complex 
process of balancing unevenness; I fear that simulating this process based 
on just some random numbers will not work well.

Greetings,
Juergen

> Ok, so far
> Greetings
> Till

IMG_0951_small.JPG
Description: JPEG image