vuescan/IT8 target/PhotoShop

[Robert B. Peirce]

While the ICC adjustment Vuescan makes from Wolf's target is very good, I
want to tweak it a bit. One way is to sample the grey scale and adjust RGBL
curves to fine-tune it a bit.

The problem is I can measure the grey scale on a display of Wolf's target,
but I don't know what it is supposed to be. Does anybody? There are, as I
recall, a total of 24 rectangles. I think the extremes are supposed to be
white and black, although they aren't on my display; white is around 245 and
black is around 15. However, I have no idea what the other 22 rectangles
are supposed to be.

This may be a fairly standard thing, but I haven't seen it published
anywhere. I may find that any further adjustment isn't worth the effort,
but I would like to take a look at it..

 

[Erik Krause]

The problem is I can measure the grey scale on a display of Wolf's target,
but I don't know what it is supposed to be. Does anybody? There are, as I
recall, a total of 24 rectangles. I think the extremes are supposed to be
white and black,

No. If patch GS0 is white the target is overexposed, if GS23 is black
the dynamic range of the scanner is not sufficient. The should-be
values are listed in the text file you got with your target, normally
in XYZ color space, but from Wolf Faust fortunately in LAB color space,
too.

Be aware that those values are valid for D50 white point only. Before
you start to calculate with the formulas found at
http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html try an
easier way:

Use vuescans 'make IT8 target' task to output the target data as RGB
image in your preffered color space (AdobeRGB f.e.), same as your
scanned image. Then you can compare directly.

 

[Robert Peirce]

No. If patch GS0 is white the target is overexposed, if GS23 is black
the dynamic range of the scanner is not sufficient. The should-be
values are listed in the text file you got with your target, normally
in XYZ color space, but from Wolf Faust fortunately in LAB color space,
too.

Be aware that those values are valid for D50 white point only. Before
you start to calculate with the formulas found at
http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html try an
easier way:

Use vuescans 'make IT8 target' task to output the target data as RGB
image in your preffered color space (AdobeRGB f.e.), same as your
scanned image. Then you can compare directly.

Thanks. I forgot about the file.

 

[Wolf Faust]

The problem is I can measure the grey scale on a display of Wolf's target,
but I don't know what it is supposed to be. Does anybody? There are, as I
recall, a total of 24 rectangles. I think the extremes are supposed to be
white and black, although they aren't on my display; white is around 245 and
black is around 15. However, I have no idea what the other 22 rectangles
are supposed to be.

This is totaly correct. Actually, usualy the DMin value should be even
smaller than 245. Usualy it is arround 235 (after applying the
profile). You have to scale the luminance if you want white in your
image become RGB 255/255/255 on the monitor output device (also check
your used Render Intent of the CMS... for photo scans you usualy want
photographic intent).

To give another example with the opposite effect: you might measure a
luminance value of 90 on your film, but when printing on a recycled
paper, the max luminance possible (media white point) might only be 80
and any hightlights over 80 will be cut off when not scaling. So when
converting colors between different devices/media, extra action is
necessary and an absolute color accuracy is usualy not wanted.

This may be a fairly standard thing, but I haven't seen it published
anywhere. I may find that any further adjustment isn't worth the effort,
but I would like to take a look at it..

As Erik noted, you can look at the content of the reference file of
your target. The Lab values noted there for each patch can be compared
with the Lab data displayed by PhotoShop using the color picker after
switching the color mode to Lab (assuming PhotoShop uses the correct
profile setup for the scan).

You might also have a look at the qtProfileChecker tool shipped with
the free lcms profiler ( http://www.littlecms.com/profilers.htm ).

 

[Robert Peirce]

This is totaly correct. Actually, usualy the DMin value should be even
smaller than 245. Usualy it is arround 235 (after applying the
profile). You have to scale the luminance if you want white in your
image become RGB 255/255/255 on the monitor output device (also check
your used Render Intent of the CMS... for photo scans you usualy want
photographic intent).

To give another example with the opposite effect: you might measure a
luminance value of 90 on your film, but when printing on a recycled
paper, the max luminance possible (media white point) might only be 80
and any hightlights over 80 will be cut off when not scaling. So when
converting colors between different devices/media, extra action is
necessary and an absolute color accuracy is usualy not wanted.


As Erik noted, you can look at the content of the reference file of
your target. The Lab values noted there for each patch can be compared
with the Lab data displayed by PhotoShop using the color picker after
switching the color mode to Lab (assuming PhotoShop uses the correct
profile setup for the scan).

You might also have a look at the qtProfileChecker tool shipped with
the free lcms profiler ( http://www.littlecms.com/profilers.htm ).

Thanks for the help. I am noticing that a print of the scan seems to be
right on, so I don't think I want to mess with that. My hope was to
create some kind of PS overlay for the monitor, which is calibrated, but
which doesn't really match the target as closely as one might wish. The
problem, I am finding, is that transmitted and reflected light never do
seem to match exactly and that my attempts are not getting where I want
them to be, hence my comment that it might end up being a waste of time.

BTW, I am on a Mac and the last time I checked I could not use
littlecms's stuff.

 

[Robert Peirce]

I think I solved my display problem. I had been using a gamma of 1.8
because I was on a Mac. I just saw something from InkJet Mall saying
that you needed to use 5000K instead of 6500K for proofing. I
recalibrated my monitor at 5000K and compared Wolf's target to the
display. It was just as good as the printed target. Neither is an
exact match but I have come to understand that is never going to happen.
The important thing is they are all very close and that is what I needed.

BTW, I was very surprised at the difference 5000K made vs. 6500K.