Resolution indpendence

[Eirik Byrkjeflot Anonsen]

David De La Harpe Golden <david.delaharpe.golden at gmail.com> writes:

> Eirik Byrkjeflot Anonsen wrote:
>
>> That's a simple application of nyquist. Features smaller than 2
>> pixels can not be accurately represented
>
> That's a bit simplistic:  that'll tell you which
> features will be accurately preserved. It doesn't tell you if those
> features were important to preserve to make the reduced icon
> sufficiently visually distinctive and recognisable to remain useful in
> context.

True enough, but it does explain why a "less busy" icon "looks better"
at a small pixel count, with a fairly high degree of precision :)

And note that the problem isn't just "being preserved".  The problem
with aliasing is not that features "disappear", but that they appear
to be something different than what they really are.  And that is much
worse.

(If you're just using a naive rasterizer to render an image, nyquist
can be used to automatically detect features prone to aliasing.  And
it seems the opinion among people who have investigated is that in
general it is very hard to do better than a naive rasterizer.)

eirik