Hi,

I have a small website with basic info on image manipulation using APS
and PSP. I am currently updating and re-writing it. I have been
experimenting, trying to find a way to more or less "profile" a system
using just the system itself. My website is not aimed at people who
have a lot of experience or high end equipment. It is more for the
person who is just starting out. I am doing this because I figure
someone like this will not have profiling software available to them,
nor will they want to spend what such a system will cost.


My thinking goes something like this. Essentially, what we are trying
to do is to assure that none of the elements of our system (scanner,
monitor, printer) is introducing color shifts. Realistically, we know
that some shifts are going on, and that the various elements of the
system do not reproduce color the same way, nor are they capable of
doing so. But what we want is to get them as close to each other as
possible, so that we can be reasonable sure that what we scan looks
like what we see on the screen, and what we see looks like what we
print.

We have a way to measure color composition, the eyedropper tool. So,
if we had a standard sample that we knew the color composition, we
could use it to determine where color shifts are occurring and create
curves to adjust for them.

The problem is that we do not have such a color sample. But I reason
that we could make one. All we would need are some samples that we
KNOW are neutral. I am reasonably sure that just about any good
quality photo paper will be neutral. It might not be absolute white,
but it will be neutral. I think the same could be said for just about
any good quality black printer ink. The people that make these know
that neutrality is important, so (I would assume) they try to keep
their products neutral.

So if I can assume this, then I have a white and black sample. I don't
really know what the values for these samples will be, but I do know
that they should be fairly neutral no matter what values they actually
are. For a neutral gray sample, I suggest an 18% gray card used in
photography.

Now, suppose I scan these and measure the colors. If they are not
neutral, I know my scanner is introducing a color cast, I can identify
the cast, adn I can create a set of curves to adjust for it. If i then
save these curves, I can apply them to any image I scan to remove the
cast.

By the same token, I can create an image has various shades of black,
white and gray, themn print it. I can then scan this print, apply my
correction curves, and determine if my printer has a color shift. I
can then create another set of curves to remove this.

Does this make sense? I am writing a page for my site about this, and
I would appreciate any suggestions that you might have, Please
remember, this is not a site for professionals who have profiling
software and equipment, and so I try to think of ways of doing
without.


guntcher