[filmscanners] Re: Scanner Profiling

["Doug Franklin" <franklin@shootingshark.com>]

On Thu, 17 Jul 2003 14:10:29 +1000, Julian Robinson wrote:

> I think what shaf was saying is that if the scanner is accurately
> calibrated color-space-wise, then it will take from the neg film the
> 'exact' colors as they are recorded on the neg film, complete with the
> color biases, the mask etc.

Yeah, if you built the profile from a target shot on the same type of
film.  The mask and gamut are different for every negative film.  The
gamut is different for every slide film.

> Then, because you have an accurate color rendition you can apply a
> known transfer function to convert to positive by removing the mask,
> inverting and allowing for color bias.

This has to be done on a per-film basis because the masks are all
different and the gamuts are different and the representation in
negative is different from film to film.

> I really get confused by what is the meaning of a color space vs "absolute
> color".

There's no such thing as "absolute color" once a human gets involved.
"Absolute color" in any reasonable, comparable sense can only mean
measuring the specific wavelengths.  All else is perceptual.  And
different people _do_ _not_ perceive the same wavelength as the same
color.  Women, in general, are more consistent in the way they view
colors, that's why the textile industry hired primarily women as color
checkers prior to digital colorimetry.

> If my scanner is set up accurately and is profiled for RGB.  I think this
> means that the scanner is set up to convert whatever it sees on a slide
> into a digital number representing the nearest best color that is available
> in RGB color space. Is this true?

True, for the medium from which the profile was generated. More or less
true for other media, depending on how "similar" they are to the
original medium in color response and gamut.

> Now what does it mean to say that a MONITOR is profiled for RGB
> color space?

It means that if you present it with an image that has been accurately
transcribed into the RGB color space, that the colors you see on the
monitor will be the same as the colors in the original image.

> nearest displayable equivalent 'color' on the screen, knowing the
> characteristics of the screen?

Pretty much spot on, there.

> If my understanding in above is correct, then I have more difficulty
> in understanding what Photoshop etc mean when you set up a profile/color
> space for them.  What does this mean?  What is the relationship between
> Photoshop color space and the monitor color space when a photoshop image
> is being displayed on screen?

OK. The basic deal is that every medium (the human visual system, films
of different types, papers of different types, CCDs of different types,
etc.) responds to incoming raw energy in different ways.  Every one has
differences in a variety of parameters: max/min perceivable intensity,
spatial and color resolution, linearity of intensity response, etc.

Color spaces encode the color response characteristics of a medium into
a set of functions.  These functions tell properly instructed software
what to do to get the colors of the image into a "space" (set of axes
along which we can measure things) in which value 34 means a specific
green.  That green might be perceived as a 30 green on some devices and
50 on others, and it might take a 22 or a 70 to get an output medium
(monitor, printer, etc.) to produce that exact green.

To encode the color response characteristics, color spaces define the
axes along which we can measure color, and the "linearity" of the
measurements on these axes.  The RGB space has three axes: how red is
it, how green is is, and how blue is it.  It also has three
transformation functions that map the intensity response, one for each
of the axes.

The CMYK space, often used in the printing industry, has four axes: how
cyan is it, how magenta is it, how yellow is it, how black is it.  Plus
the associated "linearity functions".

A specific color space is simply an agreement that 34 means some
specific green.  If that green is present in an incoming image, it will
show up as a 34 in the digital version.  If we present a 34 to an
output device, we'll get that same green back.  Each of the common
color spaces is simply an agreement among a different group of parties
about exactly which green is 34.

Knowing all of these transformations, I can

a) convert incoming images from different media or input devices into
colors that are directly comparable to each other (all the 34's mean
that same green)

b) convert an image so that it displays on different devices with the
same colors it had in the original (all the 34's still mean the same
green, and you see it)

So, basically, you take all of the incoming images, and use their
original (device) color profiles to transform them into the agreed upon
intermediate color space, often RGB.  Then, when you want to render the
image onto a display, you use the color profile of the display device
to ensure that the green 34 shows up as the particular green that we've
all agreed upon via the RGB color space.

> Any help including ref to a good basic text welcomed.

Evening's "Photoshop for Photographers" among others.  His is a pretty
good intro.

TTYL, DougF KG4LMZ


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