> From: Laurie Solomon
>
> Yes, the tests were done prior to PSCS and I know of none done
> since. I am
> not sure if Adobe made significant improvements to the basic Bicubic
> formulation as much as they made its implementation more sophisticated by
> furnishing two subtle variations on the basic formulation.
As I understand it, bicubic is a linear (in the sense of linear algebra)
resampling filter. If you blow something way up, you always wind up with a
blurry result, if you zoom in on it. PS CS has added Bicubic Smoother and
Bicubic Sharper variants, but they merely tweak the high frequency response
of the filter, which you can see quite easily if you blow up some sharp
edges to 10x.
GF attempts to go beyond that by finding edges, and then trying to preserve
that edge sharpness when it upsamples. This is nonlinear processing, and is
in some sense artificial--and therefore not always effective. I find that it
works great on images that have distinct edges, e.g., architectural shots,
but sometimes creates edges where there were none.
I've posted a pair of examples, both involving blowing up by 10x a small
piece of an image that had some architectural edges as well as some non-edge
detail. You can see what I mean:
In the foreground, the artificial edge invention looks like some exotic
Photoshop special effect.
--
Ciao, Paul D. DeRocco
Paul mailto:pderocco@ix.netcom.com
----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or
body
