Digital BW, The Print

on 2/7/02 1:22 AM, Todd Flashner wrote:

Allow me to try to clarify something I wrote below, because as is it's
contradictory.

> If the camera then applies an algorithm to give each of those pixels an RGB
> component, that requires interpolation, and you'll end up 3 channels at 4
> pixels each. You've added color channels, through interpolation, which may
> or may not contain an accurate representation of the color they are tying to
> create, but you've added no detail.

Above I say that interpolation adds no detail.

Then below I say you may be able to take smoothness from the added detail.

I think I'm wrong about the below part. I was thinking of the three color
channels of scanned color film, where the channels don't always overlap each
other exactly, due to camera lens aberration, dye cloud composition, sensor
sensitivities, etc. But in the case of interpolated color I'd imagine the
registration between channels would be exact. You wouldn't gain tonal
smoothness, though you would still gain control over contrast and
"tonality".

But, nor would you loose detail from interpolation.

> Nor have you lost any, but depending on how you convert that interpolated
> color info back into grayscale, your file may stay as sharp as the original
> 4 pixel capture, or may get softer as you apply interpolated detail to your
> original detail. However, the consequence of this softening could show up in
> tonal smoothness.

My point in all this is that the interpolation we are speaking of is only
additive. It's not subtractive to not have the interpolation. Thus, from the
point of view of the amount of detail captured, it should not matter if the
camera is operated in grayscale mode or color, but tonality may definitely
be affected.

Todd