[filmscanners] Gamma and Color

["Dana Trout" <dana@troutcom.com>]

I'm trying to understand an effect that I'm seeing in Photoshop, which is
evidently there by design because it appears in versions 5.5, 6, and 7.

I have noticed that when I try to change the midtones of an image I am
getting noticable color shifts. For instance, if I have an image that's too
dark and I use either Gamma or curves to brighten the midtones, the RGB
ratios for each pixel do *not* remain the same.

Say I have a dark orange pixel with RGB values of 32, 16, and 8. I apply a
gamma of 2.0 and read RGB values of 90, 55, and 30. What has happened is the
green and especially the blue have been boosted much more than the red -- the
original ratio of R being twice as bright as G, and four times as bright as B
has now become R is only 1.63 times as bright as G, and three times as bright
as B.

An interesting exercise is to create a color step wedge and apply different
gamma values to it. You will see that whatever primary color has the smallest
value gets the greatest boost if gamma is larger than unity. This occurs in
the popular color spaces of Adobe RGB (1998) and sRGB. Note that if the
primaries are equal (that is, the pixel is gray), they all get the same boost
and the result is still gray. But if one color is dominant (foliage or
skintones, for instance), the hue will change. This is a real problem if I am
trying to preserve *both* skin tones and grayscale!

Does anyone know of a way of adjusting the midtones of a color image
*without* affecting the colors? And under what conditions would we *want*
gamma or curves to affect the colors when we adjust the RGB channel (as
contrasted to the individual color channels)?

Thanks,

  --Dana

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body