|
|
Posted by Ben C on 08/02/07 19:53
On 2007-08-02, Toby A Inkster <usenet200707@tobyinkster.co.uk> wrote:
[...]
> The fact that most fonts are vector-based and anti-aliased has got nothing
> to do with it. It comes from the fact that a pixel can only ever be one
> colour[1]. Say you're trying to draw a serif capital letter E at 10 pixels
> high and 6 pixels wide, how do you do it?
>
> *XXXXXX
> XXXXXX
> XX *
> XX
> XXXXX
> XXXXX
> XX
> XX *
> XXXXXX
> *XXXXXX
>
> (We have anti-aliasing, so X = black, * = grey)
>
> Easy. Now what if you have to do it at 5 by 3?
>
> XXX
> X
> XXX
> X
> XXX
>
> Now what about 4 by 2?
>
> XX
> X*
> X*
> XX
>
> It's no longer recognisable as an E: it's become too pixelated to read.
> This is why high res + large fonts looks nicer than low res and small
> fonts. This is why high res exists!
>
> Demonstration:
> http://examples.tobyinkster.co.uk/garamond-sizes
>
> ____
> 1. Yes, I do know about subpixel rendering, but this only works on some
> monitors, and it only goes a small way to improving things.
There is also "font hinting" which can cause the font renderer to
distort the shapes of the glyphs a bit so that they hit pixel boundaries
better and so don't require so much anti-aliasing at small sizes.
These hints are built into the font and designed painstakingly by hand,
which is quite a job, especially for CJK fonts. All that work and then
we just tell people not to use such tiny fonts anyway.
Windows fonts are quite good at hinting, and this is the reason why some
users of the recent Windows port of Safari have complained that the
fonts look "fuzzy": it's because Apple have built a font renderer into
their Safari port rather than using the Windows one. The one they built
in does less hinting and therefore you get more anti-aliasing at the
lower point sizes.
Navigation:
[Reply to this message]
|