From Mistake to Mascot
Have you ever wondered why your computer screen has jagged lines? Why do pyramids look like ziggurats? Why do uppercase E, L, and T look so good, yet S, W, and O look like badly made Christmas ornaments? Why do curved lines look like they've been drawn by someone with palsy? I've met people who think these staircase artifacts are intrinsic to computer displays - more or less a given with which they must live. After all, we've watched enough Westerns and seen stage- coach wheels go backwards, and we don't flame the movie studios.

Well, this month's column is my flame to almost every computer manufacturer and software developer on the planet. People are tired of your jaggies. It's time to correct your offensive fonts and graphics. And, as you know, it is not hard to do.

Here's an irony. Remember those funny fonts taken from magnetically sensitive characters on checks? One font was even given a name: MICR (my guess is that this is an acronym for Magnetic Ink Character Recognition).

During the 1960s and 1970s, graphic designers frequently used MICR to cast a look and feel to the electronic age. We are doing this all over again in the 1980s and 1990s with aliased fonts (so far nameless), frequently used in graphic design to signal "computer." Before this mascot does get a name, let's correct it, because today there is no need for lines and characters to be anything less than print quality and perfectly smooth.

I won't go into the added irritation we encounter in animation. As an image moves, the jagged little steps come and go, increase and decrease in number, and move in all sorts of counter-intuitive directions. The passenger beside me on the plane, as I wrote this, was playing a golf game on his laptop and did not seem to be fazed by the fact that the golf club went from being perfectly straight to being a staircase with moving steps. When I pointed this out, he suddenly found the game too annoying to play (sorry about that). He reacted with disbelief when he learned how unnecessary this condition is.

Why Can't New Dogs Learn Old Tricks?
The techniques of getting rid of the jaggies, called "anti-aliasing," were developed 25 years ago and can be credited to at least three research centers: Xerox PARC, the University of Utah, and MIT. Researchers (I include myself) observed, for example, that video cameras produce images of lines and letters that do not have staircases. Without going into technical detail, suffice to say that the graytone of those images allows for smoothness. By adding levels of gray or tonal depth (i.e. bits in the z-axis), one gains perceived spatial resolution (i.e. the plane of x and y). In other words, by putting your bits in z, you get better x-y resolution than if you put them directly into x and y. Time and time again, it has been proved that the human eye is better served by putting more memory (more levels of gray) into the z-axis than by adding more pixels per inch.

Part of the confusion and historical stubbornness in this area stems from a lack of awareness about the difference between half-tones in print media and the continuous tone of video. The pixels on your screen are not, repeat not, like half-tone dots (the use of the word "dot" here is misleading). The dots in newspaper and magazine pictures are, in fact, not dots at all. They are oddly shaped, amorphous blobs of ink that occupy a printed "cell" in accordance with the level of gray in the image being screened (what we call half-toning). The dots on your display screen are not amorphous, but they can have graytone, which is the whole point.

So, why aren't all computer displays anti-aliased? Here is the excuse: When a character or line is de-jaggied, it must be computed in accordance with the background. Just think of a black diagonal line passing over both a white and gray patch. The levels of gray that make the line look smooth go from black to white in the first case, but from black to gray in the second.

One must look at what is there before proceeding willy nilly, writing lines and characters. And the excuse goes on. If, suddenly, any part of the background changes, information must be at hand to recompute and re-anti-alias.

Ten years ago, one could stomach the argument that this was sufficiently difficult and time consuming to conclude that computer power was best spent elsewhere. Also, it was the case that graytone, let alone color, was not common in many displays (full color is just three sets of gray tone, one each for red, green, and blue). Tomorrow it will be almost unthinkable to work in anything but full color, and today even the least expensive computers have graytone.

Jaggies Should Be An OSHA Violation
What puzzles me the most is that we seem to have educated an entire generation of computer scientists who don't fully understand this simple phenomenon, and we seem to have trained the public to take it for granted. Perhaps it's time to make aliased graphics a violation of Occupational Safety and Hazards Administration minimum standards for display quality. Or, perhaps the Environmental Protection Agency can declare this condition to be visual pollution. The point is that it must stop.

I would have expected Japan to be a greater force in this area, because Kanji benefits even more than the Latin alphabet from the resolution added by graytone. I would have expected Europe to be more active, since there is much EC legislation concerning computer screen characteristics.

I would have expected the United States to implement anti-aliasing, if only because its theoretical and practical roots are in America. But, alas, the ambivalence is worldwide.

As we rush into a world of sophisticated games, electronic books, and multimedia everything, we will invariably see more and more jaggies and more and more people will assume they are intrinsic. They are not. If you don't believe me, ask a computer science friend. There really is no excuse any more.

So wake up: Apple, IBM, DEC, HP, Microsoft, and all you other companies. We're tired of the jaggies.

Next Issue: Talking to Computers

[Copyright 1994, WIRED Ventures Ltd. All Rights Reserved. Issue 2.01 January 1994.]