Render Judgement
Apple released their Safari internet browser for Windows this week. It caused quite a kurfuffle. Because of the browsers speed? No. (Well, actually, a bit.) Because of the feature set? No. Because of, it turns out, the font rendering. A bunch of big gun blogs took issue with the font rendering. First they were puzzled by why Safari was doing it “wrong”. Then they (and, incidentally, I) realized that neither was necessarily wrong, but were certainly different.
Fine. Cool.
For what it’s worth, I strongly prefer the Mac style of rendering. That means just about zilch, though, since I’ve never used Windows long enough to get accustomed to their style of aliasing.
Even so I take issue with Joel’s claim that anyone who prefers the Mac style does so because they are “untrained”. Though Joel claims it, I don’t see why it’s obvious that the Windows style is actually “pragmatically” better. He says it is easier to read, but offers no evidence. At the moment I must assume he’s coming to that conclusion through introspection.
I’m suspicious, because to me his claim sounds like an engineer’s argument: “It fits in the grid and is therefore sharper–obviously that makes it easier to read!” Really? For sure? Why? Our eyes do funny things, on a physical level. In particular they have a lot of hardwiring for certain kinds of image processing. They find edges really well. They take differentials across space (and time) really well. They don’t sit there passively, they actively modify the signal in a manner that turns out to be useful for our brain.
So it’s unclear to me if forcing fonts to render on the pixel grid is actually better. The only thing that appears clear to me is that it would be increasing the high frequency signal in the text’s rendered image. Is that helpful? I’d guess probably not. If nothing else, the more high frequency information there is, the higher the contrast required to see it:
For ease of reading, it is essential that text have a reasonable luminance difference from its background. The International Standards Organization (ISO 9241, part 3) recommends a minimum 3:1 luminance ratio of text and background; 10:1 is preferred. This recommendation can be generalized to the display of any kind of information where fine-detail resolution is desirable. In fact, as the spatial modulation sensitivity function shows, the finer the detail, the greater the contrast required.
(From Information Visualization: Perception for Design, 2nd edition, by Colin Ware.)
I’m not saying I know whether Microsoft’s approach is better or not. It may be neither is better, or they’re better for different people, depending on your goals and your biology. But whatever the claim, I’d like to see some evidence for it.