Re: filmscanners: re:NewNewPolacolorInsight ColorSlide ProfileforSS4000 & +

[Wire Moore <wirem@home.com>]

on 10/8/01 4:52 PM, Alan Eckert at aeckert@cpcug.org wrote:

> Having gone through the elaborate IT8 slide profiling process detailed by
> Ian Lyons on his website, before I spend time on this new profile, I'd like
> to know whether it's likely to produce better results (e.g., better shadow
> detail) than my existing custom profile.  Opinions, anyone?
> 
> ----- Original Message -----
> From: Hemingway, David J <HEMINGD@POLAROID.COM>
> To: <filmscanners@halftone.co.uk>
> Sent: Friday, October 05, 2001 6:42 PM
> Subject: RE: filmscanners: re:NewNewPolacolorInsight ColorSlide Profilefor
> SS4000 & +
> 

This is an interesting question because the answer requires consideration of
some assumptions and norms for image data coding from the scanner.

Scanner drivers must all process the data from the sampling HW, at least
correcting for general film attributes and display gamma. Yet I havne't
heard of any standards for scanner output, and the different products work
according to their own rules. I notice that Polacolor's "Raw Color Positive"
and Silverfast HDR output seems to be a lot rawer than any I get from
Vuescan, Nikon Scan, HP Precision Scan or Umax Magicscan.

Whatever processing the scanner driver applied will be conducted according
to some kind of a profile, but not necessarily an ICC profile. Is this
profile considered part of the scanner? This leads to general questions
about layers of processing of image data with respect to ICC profiling.
Why/how does the ICC layer fit and what are the dependencies on and
interactions with the driver layer? Software for building ICC profiles
expects some coding, but the parameters aren't documented.

When does raw scanner data become cooked for an ICC profiling package :)

In a Creative Pro review of the Nikon 4000 ED,  the importance of using
"raw" data from Nikon Scan for making profiles for the 4000 ED is
reiterated, but there's no discussion of when/why/how the data can be
considered raw. If you use Nikon Color Management in the driver, is the
driver changing the data to the requested working space, or only assigning a
profile. Can/should you profile the output using Nikon Color Mangement as
opposed to not? If you think you know the answer to this question, then
change the context to Silverfast 5 AI and consider the IT8 based calibration
(not a color management system). Do the same rules apply? Why?

All that the Monaco EZ Color documentation has to say is "Make a note of
your settings when scanning the target. You will need to use the same
settings when scanning images for use with this profile." Later EZ Color may
reject you scan based on your settings. This seems to be a matter for phone
support.

This leads to related questions: Once I've found settings that work with a
profiling package, reusing my settings may result in a poor exposure of an
image. This poor alignment can substantially outweigh any advantage to the
use of the profile. But how far can I adjust the settings to get a good
image capture without upsetting the assumptions required for the profile to
work?

I don't have clear answers to these questions. Apparently, neither do the
purveyors or experts.

---

Here's how I handle it with the SS4000:

I consider the Polaroid film profile (e.g., "Color Slide 2b") to be an
integral part of the scanner. I would use any means at my disposal to get
the maximum amount of information from the film. Then I would profile using
these settings. After I have the profile, if an exposure adjustment is
required in the driver for some image, I do this (set levels).

Of course, once you get good at this, maybe an input profile is irrelevant
and a scanner profile starts to look like a waste of time. Especially if you
happen to notice that the scanner profile some times does more damage
(posterization) than good (accurate color) in your image.

Therefore, I would suggest you (Alan Eckert) use the Polaroid 2b profile,
then redo your custom profile on top of it, unless the custom profile is
somehow made using the uncorrected data from the scanner (filmtype "Raw
Color Positive"), in which case the 2b profile from Polaroid is irrelevant.

---

Question for David Hemingway: Is it appropriate to apply the Polaroid
profiles (in the Preferences folder) to the "Raw Color Positive" data in
Photoshop? I haven't tried this... Can the results effectively be the same
as selecting the profile in Polacolor? Or are the Polaroid profiles for
Polacolor consumption only?

Wire Moore