[Tig] new on the TIG wiki

[Bob Friesenhahn]

On Thu, 10 May 2007, peter swinson wrote:
> The reason I referred to the right hand samples as Nyquist limit was that
> the other person was quoting no point in sampling beyond Nyquist. i.e. 2K.

That person required your education.  The Nyquist-Shannon sampling 
theorem tells us that it is necessary to sample at at least twice the 
frequency of the desired image detail in order to avoid aliasing 
effects.  In fact this rule applies to a simple sinusoidal signal 
(e.g. "TV lines") so if the captured image contour deviates from that 
(thereby representing higher frequencies as represented by Fourier 
series), then a higher sampling frequency is required.

> All I know is that  only a few years ago everyman and his dog said 4K would
> never be needed, but we now seem to be moving in thjat direction!

In order to capture 2K worth of reliable detail it is necessary to 
sample with at least 4K samples.  Since normal images are not 
comprised of sine waves (very boring for the viewer!), even more 
sampling frequency is desireable.  This rule would apply even without 
the disruptive influence of film grain and other artifacts.

Bob
======================================
Bob Friesenhahn
bfriesen at simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/