Archived from groups: rec.photo.digital (
More info?)
In message <5PCdnZSYAe-SrizcRVn-tQ@comcast.com>,
"Neal Matthis" <nmatthis@hotmail.com> wrote:
>Ah, that did it. I had linear checked. Hmm, so exactly what does that
>mean? I did notice that the histogram of the linear 16-bit image seemed to
>have all the pixels in teh first half of the histogram, which certainly
>explains why it was so dark.
Gamma-corrected images have luminance levels that represent the base of
an exponent (usually 2.2). This means that mid-scale (128 out of 255)
is not 1/2 as much light, but (128/255)^2.2 = 0.2195, or 21.95% of the
intensity, which is over 2 stops darker than the max of 255. In a
linear scale, 128 will be about half the intensity of 255, or 50%.
A lot of times you'll have nothing in the right half of a linear tiff
from a RAW file, because the cameras expose for JPEG, which usually
clips away the top 1 stop of highlights. If you don't touch the
"exposure" slider in the RAW converter, you will lose them, also, unless
you are using linear mode. So, you get two effects combined; a linear
image will have a histogram squished to the left even if the clipping
point is the same for both linear and gamma-corrected, *plus* the linear
tiff usually has an extra stop of dynamic range, so the content is
squished even further to the left.
The actual levels of what normally becomes 128 (out of 255) in a default
RAW conversion are around 200 to 400 (depending on the color channel)
out of almost 3900 levels (all 4095 are not used in the Canons). That's
only 5% - 10% of the full linear range.
--
<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS@no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><