There is no differance between resizing and
resampling as you call it : it's just differant
parameters to the same kind of algorithme.

In fact, only some programs give 2 differants
name for what is basicaly the same function.

Whenever you change the size of a picture,
you have to lost or create pixels. To get a
better result, it is possible to interpolate
half-tone pixel to get a smoother result.
(For example, if you have a black circle on
a white background, the program will use some
grey pixels to smooth the transition between
black and white). This is actualy a resize
followed by a smoothing.

Naturaly, that mean that sharpening is the
exact reverse operation : it will remove
half-tone between colors boundaries to give
sharper edge. If you sharpen a smooth resized 
picture enough, you would get the same result
(except for lost of information resulting from
the non-reversible algo) as if you had only
resized without doing any interpolation.

By the way, you are mistaken : I don't resize at
all scans of my cels : they are directly scaned
(opticaly) at the correct size with hard-edge
interpretation. If I wanted to get a better
result, I would have to scan at a bigger size
and then resize it to smooth it.

But if I don't do it, it's  because I have my
own reasons... It's on purpose that I don't
put scans of better quality on the web...

Cheers,
Yann Stettler