Re: Gradients
Re: Gradients
- Subject: Re: Gradients
- From: Armand Rosenberg <email@hidden>
- Date: Fri, 31 Oct 2003 15:29:59 -0500
I never said taking fewer measurements was more accurate, just a
valid way to obtain smoothing -- that's the point.
You don't want to use too few measurements, to the point where you
aren't representing the object accurately, but too many measurements
will give you more wrinkles than you want. There is a statistical
middle ground that maintains the shape of the object while also
smoothing it out. This is well known in the world of computational
simulations (in addition to color profiling). Missed anomalies (aka
singularities) are always a danger when using such methods, however.
Fortunately, most of the time in the world of printers anomalies are
rare...
Unless you have a very badly behaved object, statistically speaking,
fewer measurements (with intelligent interpolation of course) will
likely smooth it out. If you have as many wrinkles as you suggest
then I would venture to guess that no method known to mankind will
give you smooth gradients over a wide range.
That's the logic behind using fewer measurements, and it is perfectly
valid from a statistical/mathematical viewpoint. It's also been found
to work in the real world with many printers. Keep in mind that this
is by no means a panacea, so if you find an exceptional printer or
two, that should not be a surprise. But this technique is a lot
simpler and less expensive than any other method and often it works!
The mathematical smoothing with many measurements was the alternative
I suggested in my last message, already implemented in some packages.
I also explained some reasons why that method is not necessarily
better.
Armand
If instead
your intent is to smooth out those wrinkles (eg, for obtaining
smoother gradients) then certainly ONE valid way to do it is to take
fewer measurements.
How can that be more accurate?
How do you know if by chance you are measuring one of those wrinkle peaks
or valleys?
By taking lots of measurements, one can then mathematically average out
those wrinkles and get as smooth a surface as one want. Smooth too much
and one end up with a ball.
Bertho
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.