ClearType - a clarification

agree at carltonfields.com agree at carltonfields.com
Fri Oct 8 20:04:34 UTC 1999


By the way, the color bleed Andreas was concerned about in his posting is PRECISELY the visual effect of Hi-Res fonts on the good old Apple ][, which rendered relatively high-quality fonts (for that era) by exploiting the color-shifting on that CRT.  The Apple ][, you may remember, represented seven adjacent hi-res "bits" in two bytes as follows:

	(S) 1 1 2 2 3 3 4 | (S) 4 5 5 6 6 7 7

Where each number pair (1 1, 2 2, and so on) corresponded to one of four colors, depending upon the parity of the (S) bit.  I think, but don't recall

	(S) off

		0 0 black
		0 1 blue
		1 0 red
		1 1 white

	(S) on

		0 0 black
		0 1 purple
		1 0 green
		1 1 white

or something like that (I have some old manuals at home -- I'm actually curious how it went myself).  You could use the given resolution of each "bit", but you could also effectively double the resolution by ignoring the color issues, treating the 7 bits as 14, and get better looking font shapes.  The result led to funky color fringing, particularly for "thin letters," but you could adjust the fonts so they looked fine most of the time, and get neat effects.  (The diagonals of the Wizardry mazes and the graphics in the so-called Windo-Wizardry user interface (which in those days, believe it or not, was a technical feat on that 48K machine) depended, in no small part, upon this silly trick.  Yes, the colors were fringy, but given what we had to play with, it worked.

The reason we couldn't improve the fringy colors as described above on the Apple ][, is that each micro-pixel could only be on or off.  Most RGB today provides effectively at least 256 shades of intensity per micropixel.  If there is a difference between Microsoft's technology and Wozniak's that may be it.





More information about the Squeak-dev mailing list