Script fonts are all very well, including the one based on my handwriting which gets a fair amount of use. The thing which is missing from them though is a description of the density of the ink over the letter shapes.

So, very simply, my idea is for there to be a layer of description in truetype fonts which would encode the alpha or transparency of the ink over the letter shapes.
-- hippo, Dec 17 2002

SVG font http://www.w3.org/TR/SVG/fonts.html
Assuming that you can use the full arsenal of SVG in defining a glyph you would not only be able to add transparency by also texture and colour blending. [st3f, Oct 04 2004, last modified Oct 21 2004]

Colored Fonts http://www.halfbake...dea/Colored_20Fonts
Just add alpha. [bookworm, Oct 04 2004]

(gaping hole of uncertainty in my knowledge) But aren't type 3 fonts bitmaps?

SVG's another possibility (see link).
-- st3f, Dec 17 2002


I wonder how you'd go about hinting a handwriting font.

I suspect the notion of a "handwriting font" is less ideal. In real handwriting, each letter is unique, and the shape will depend on the letters before and after, the mood of the writer, and various other features.

A "handwriting simulator" would be an interesting project.
-- egnor, Dec 17 2002


I think they entire font technology needs redone, with tools to automatically convert and do proxy with legacy font systems.
-- ironfroggy, Dec 17 2002


This sounds like a special case of "Colored Fonts." Or maybe a superset.

(Also, I believe Coranta is allowed to use the letterforms, just not the 'program' which encodes them. Typeface is not copyrightable.)
-- bookworm, Dec 17 2002


I always thought you could mimic the effects of handwriting by incorporating certain random variables, some that changed with every letter, some over the course of a paragraph, and some through the course of the whole text. These might then be used to adjust height, slant, weight and other features. Been a very long time since I coded in PostScript, so I've forgotten what's possible.
-- DrCurry, Dec 17 2002


Speak for yourself.
-- DrCurry, Dec 18 2002


([bookworm] Absolutely right. Proving copyright of the shapes of the letters is next to impossible. It's easier to assert that the font file itself is my copyright.)
-- hippo, Dec 18 2002


Why stop at the transparency of the ink? I want to be able to control the matter used so we could have light ink and heavy ink - for light and heavy reading respectively. One day we could have ink made from neutron stars - apparently one thimbleful of this stuff is so dense that it weighs as much as a huge heavy thing.
-- DenholmRicshaw, Dec 18 2002


Handwriting font randomization, as [DrCurry] describes, has been around a while; a few years, anyway. The cheaper versions only provide randomized results when printed while handwritten font systems like vletter pro will use 4 different examples of each letter for variation. Some of the commercial faces available from one foundry in the Netherlands has randomization, carried out in postscript, as a standard feature.
-- bristolz, Dec 18 2002


Type 3 fonts can do just about anything, as noted. Unfortunately, fonts which incorporate grayscale or multi-color elements cannot be cached in Postscript, nor can fonts which are supposed to have random-ish variations in each character. On the other hand, type 3 fonts can use type 1 fonts within the character definitions, so one solution is to have a type 3 font that defines a collection of type 1 fonts and draws multiple overlaid glyphs to render each character. If the type 1 font contains e.g. six variations on the letter "e", the type 2 font would draw the letter "e" by selecting one of those six variations and then drawing it. If the font is supposed to appear as e.g. blue text with a black outline, the type 3 font would draw each character in black with an "outline" font and then in blue with a "solid" font. The type 3 glyph would not be cacheable, but the two type 1 glyphs would be.

Probably the biggest difficulty I can see trying to do a good variable-density font, though, is Postscript's lack of an alpha channel. This, unfortunately, is not something that can be faked or emulated. If the multi-density text is only drawn on a white background, results will be good, but if it's drawn on any other background results will be absurd.
-- supercat, Dec 19 2002



random, halfbakery