[argyllcms] Re: ICC-profile shift histograms

From: Hermann-Josef Röser <posts@xxxxxxxxxxxxxxxxxxxxxxxxx>
To: <argyllcms@xxxxxxxxxxxxx>
Date: Wed, 15 Mar 2017 09:14:45 +0100

Good morning Roger,

... , to my knowledge, states that, 120 or 140 or 150 or 160 should give

you more or less the same XYZ/Lab values because it's all driven by the IT8
Referenz values.

This is how profiling is set up: the Lab/XYZ-values as specified in the
reference table are reproduced -- to within the accuracy of the model. But
this is not denied by my argument. I only cheat a little bit in that I use a
profile with exposure time X for a scan taken with exposure time Y. That is
all. I know, that profile X does not exactly represent the scan with
exposure time Y. But if I prefer the results for this or that reason, why
not? Of course, for a target scan with exposure time Y the profile derived
from a scan with exposure time X will not result in the L values given by
the reference table but be different according to the change in exposure
time. The values for a and b should be more or less untouched by this
procedure. This is what I see visually in the rendered images.

Best wishes

Hermann-Josef

--- Begin Message ---

From: "Roger Breton" <graxx@xxxxxxxxxxxx>

To: <argyllcms@xxxxxxxxxxxxx>

Date: Tue, 14 Mar 2017 21:09:08 +0100

Suppose, through Silverfast curves, you are able to make the RGB
values of GS10 exactly R=128, G=128, B=128.

I do not change the target scan via the SilverFast curves. I only choose a

different exposure time, e.g. instead of the standard

150 (in whatever units) I use 120.

Speaking in Lab-coordinates, this should only change the L-value, not a or

b. And this is what I want. You might also put it this

way: Exposure time is part of the scanner's properties which are to be

calibrated. In first instance, I am interested in getting

the correct colours, i.e. the correct a and b values.

Josef,

When you write "Speaking in Lab coordinates", at the time you create the
"target IT8 scan", you know that there is nicht Lab coordinates involved
hier -- only RGB coordinates, the scanner does not understand Lab at this
point.

profiling process is going to be such that Argyll is going to give you

"more or less" the SAME ICC profile.

I think it will not give the same ICC-profile in terms of exactly the same

numbers in the file.

The profile should give you the same Lab including the same "Brightness"
which is the "Lightness" value (i.e. L) regardless of exposure or other
adjustments.

If anyone wants to use separate profiles for different "exposure times", say
120 or 150, there is nothing that says that this is not what profiling is
intended to do?

For some reason, you may have better results, better "RGB to XYZ/Lab"
conversion at 120 but the science behind profiling, to my knowledge, states
that, 120 or 140 or 150 or 160 should give you more or less the same XYZ/Lab
values because it's all driven by the IT8 Referenz values.

Unless you cheat and change the Referenz values...

MfG / Roger

--- End Message ---

[argyllcms] Re: ICC-profile shift histograms

Other related posts: