Re: [Lcms-user] Strange behaviour with the Profiling utility

[<ma...@li...>]

Greg,

Seems you tweaked the scanned image of targed before profiling,
and now your are not applying same process to images.

The basic idea is to set the scanner driver to whatsever adjusts 
you are going to use. Then scan the target and make the profile.
Of course the profile would be valid only for these adjusts....

So, if you scanned the target using a white point (say D50) and
now you are using the profile with other white point, you would 
end with this problem.

This is the only explanation I than think. :-?  Otherwise this is not, 
of course, the expected behaviour.

Regards,
Marti.


----- Original Message ----- 
From: "Sullivan, Gregory (SNL)" <Gre...@hp...>
To: <lcm...@li...>
Sent: Saturday, March 01, 2003 1:46 AM
Subject: [Lcms-user] Strange behaviour with the Profiling utility


Hi,
I have been using the Little CMS Profiling tools to create profiles for
my scanners. To date, I have thought they have worked very well. 

However, I have stumbled across something which doesn't seem right, and
I'd like to get some feedback. 

The device I am profiling at the moment is the Nikon Coolscan IV ED film
scanner. I am profiling the raw scans from Vuescan 7.6.18. These raw
scans are not very neutral - they have a green colour cast. 

When I create a profile, the Profile Checker reports excellent results -
a max dE of about 2. 

The problem is that when I import the image into Photoshop (V7.0, on
Windows XP), the image takes on a yellow cast. I am tagging the image
with the LCMS created profile when I open the image. If I then convert
the image to my working space, which is sRGB for this testing, the cast
remains. It doesn't matter what intent I use - there is always a colour
cast - the greyscale is never neutral.  (absolute intent results in a
reddish cast, whereas relative intent results in a yellow cast).

If I use the TIFFICC utility to convert the image from the same scanner
profile to sRGB, the results look about the same as Photoshop, although
I have not yet inspected pixel values closely. 

Now, there is a way to get what I consider to be excellent results (a
totally neutral looking IT8 scan). All I have to do is convert to ANY
profile that has a white point of D50, using absolute intent. Voila!
Problem gone. After that, I can use relative or perceptual to go into
any other space. The Profile Checker reports that the white point of the
scanner profile is 5481K. I get the same result using TIFFICC - if I
convert to a profile with a white point of D50, the result is excellent.

Is this normal/expected behaviour? 

Greg.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Lcms-user mailing list
Lcm...@li...
https://lists.sourceforge.net/lists/listinfo/lcms-user