[Tfug] OT: font utility

[Bexley Hall]

Hi Rich,

> Yes, you have to own all those fonts and manually
> categorize them, but typography is an art, not a science.

Yes, I am only worried about fonts that I already "own"
(i.e., have a valid license for since you never really
*own* any of this stuff).
 
> Let's think about why it hasn't been done (or if it has,
> why it's not well known). Bad typographers will always use
> crappy fonts. Such a tool would turn the tables and end up
> being used by crappy foundries to recommend "alternatives"
> to classics such as Helvetica and Garamond. Line height is
> often changed, kerning is usually left unconfigured, or it's
> mangled. That is, unless it comes from a typographic
> virtuoso, in which case you'll know already, or be able to
> guess which fonts they're using.
> 
> In short, the payoff is small compared to the effort of
> writing an effective glyph-matching tool: it doesn't improve

I'm not sure *how* difficult the task would be.  I can
almost envision writing a *script* to do this "brute force".
My premis is that I will *have* the font in question, not
a "close alternative".  So, an exact match *will* be found.

The only real issue I see is capturing the font that
you are trying to match.  This would have to be done
from paper output.  So, you would need to scan at a fairly
high resolution (unless you are lucky enough to have very
large text samples -- or, a matching decorative fonts, etc.).

But, even scanning at a high resolution presupposes that the
images were created at a similarly high resolution (and not
just a crappy 300 dpi printer with lots of ink bleed!).

So, hoping for an *exact* match -- even if you have one -- is
a bit farfetched.  Instead, you'd have to score each match and
just manually inspect the suggested "best match".

> the situation, and it doesn't change the fact that you'll
> still have to rummage through a bunch of likely
> alternatives.
> 
> And then there's the simple fact that people don't look at
> how well the fonts kern -- they just read the words and hand
> over their credit card.

Yes, and there's nothing to prevent the typesetter from altering 
the spacing of the text, adjusting kerning between any two 
arbitrary glyph pairs, etc.

But, I don't think most folks *do* this.  Rather, they just
live with whatever the (software) typesetter generates and
maybe pad lines to cut down on whitespace, etc.

> So given all that, body type is easy: just pick a similar
> weight of sans or serif font, and substitute away. Even
> substitution of display fonts is simpler than it first
> appears: only a typographer will notice if you're using
> (expensive) Curlz instead of (cheap) Girls Are Weird. So use
> the cheap one: no tool required.

<frown>  I haven't found it that easy.  Just looking through
the thousands (?) of (genuine) Adobe fonts is enough to leave
your head spinning.  And, previewing fonts on the screen
whilst comparing to a printed sample never seems to work
for me.  (maybe I should try *scanning* the printed sample
and displaying it in a window adjacent to the font preview
window?)