Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 




      :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[filmscanners] Re: LS2000 Shadow Detail problem

>While doing scans of a transparency, I ran across a curious problem with
>the Nikon LS2000 losing shadow detail that was captured in the scan, but
>not retained when color management was used.
>I've detailed it with pictures and descriptions of my methodology at:

In addition to some of the good suggestions from folks on this list,  here
is something else which seems to be involved with this problem.

Thor Olson, a color imaging scientist whom I am corresponding with
privately, also suggested that gamut mapping may be playing a part in this
problem. He told me some very interesting information...

He said:

>I don't think it is an intrinsic issue about color spaces;
>it's probably one about gamut mapping between different ones.
>Gamut mapping is still a very active research topic.  One
>thing they've learned is that the best mapping is application
>and image-dependent, so it would not be surprising if the
>general purpose mappings performed by Photoshop do not meet
>the needs of all pictures!


>Most monitor spaces use an encoding for lightness that has a linear
>segment near
>absolute zero, which then splices into a power law curve for the
>perceptual levels
>above around 8%.  This is true for both sRGB and L*, and probably your other
>color spaces you tried.  The deep shadow detail near D=3, would all get
>into a few counts of digital range.
>A decent film scanner would be loathe to lose this information in this
>way.  It would
>probably use an encoding that was more like a power law all the way to
>zero, spreading
>the shadow detail across a larger number of counts.
>Of course when you try to display such an image, it will get mapped to the
>L* or sRGB
>curve with the linear segment, and the detail will be lost from view (but
>still encoded in
>the image of course).
>When you open the image with no profile conversion, PS treats the encoded
>shadows as
>if they were sRGB (even though the aren't) and the detail becomes
>visible.  You have
>told PS to consider those encoded levels as being at a higher luminance
>than they were
>originally encoded at.  It will show the shadows, but of course the tones
>(and colors) are
>not correct.

This seems to be the exact behavior I am getting. The pixel values for the
deepest shadow detail do not change with a profile assignment, but the
detail disappears visually. Curves cannot bring it back up either.

This is turning out to be a lot more complicated than I thought.

I don't know why I'm surprised. <G>


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


Copyright © Lexa Software, 1996-2009.