Re: filmscanners: Profiling negative films

["Richard N. Moyer" <dickmoyer@mail.earthlink.net>]

>In a message dated 4/25/2001 0:08:19 AM EST, michael@shaffer.net writes:
>
>>  Which isn't to say Ed has a few issues to work out.  Ed may have more to
>>   say, but I believe his evaluation of Q60 target results only in a
>>   matrix-type profile, which, generally is a simplification of a device
>>   profile.
>
>As long as the sensor is linear, there's nothing that can even
>theoretically be more accurate than a 3x3 matrix transform.  This
>is what VueScan uses.

Not withstanding previous comments regarding what should be expected 
in VS, it might be useful for you to post an example of your 3X3 
matrix transform "simulation" of a scanner LUT profile, and we will 
determine with ICC profile software the degree of compliance with the 
true LUT based scanner profile. I haven't seen yet a scanner LUT 
based profile that looks like a matrix developed profile. So, this 
begs the question regarding an assumption of linearity - .

>
>Only when there are nonlinearities, such as with inks on paper,
>do you need more than 3x3 matrix transforms.
>
>Regards,
>Ed Hamrick