[argyllcms] Re: Number of patches - well behaved printer?

  • From: Rishi Sanyal <rishi.j.sanyal@xxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Wed, 23 Feb 2011 22:44:32 -0800

"Profile is between measured values and scanned values, mapping
scanned values to measured values. So it reproduces the chart."

Reproduces 'the chart' or the 'the chart on the film'?

Sorry, again I'm confused. Let's leave out my own film & just talk
about Wolf Faust's, say, Velvia 50 IT8 target. You confirmed that Wolf
(or whomever) will, after recording the IT8 chart on the film, read
the spectral values of the IT8 chart as recorded on the film (not the
theoretical XYZ values of the IT8 chart itself, but the values of the
chart as they appear on the exposed film, back-illuminated by a
standard illuminant). Then the profile will map the scanned values to
the XYZ (or whatever device independent tristimulus values) values
measured by the spectrophotometer of the image of the chart on the
very film you're scanning.

This does not create a profile that reproduces the chart's theoretical
color. It only reproduces the chart's colors on the film, correct?

If, on the other hand, I took a reflective IT8 target, illuminated
with a D50 lamp, and:
     1. Measured all chart values using a spectrophotometer, saved
these as XYZ measurements in a reference file
     2. Shot the same chart with the same illumination on my own
Velvia 50 film using a SLR
     3. Developed then scanned said shot in my Minolta DSE 5400 w/ all
auto corrections off
     4. Built a profile from this raw scan using the measurement file
created in 1.

*Then* I would build a profile that reproduces the original chart
(under that illumination). That would in a sense, I believe, remove
the film from the equation & attempt to recreate what the
spectrophotometer saw of that chart under that illumination.

Am I understanding things correctly?

Many thanks for your help,

On Wed, Feb 23, 2011 at 5:09 PM, Iliah Borg <ib@xxxxxxxxxxx> wrote:
> Dear Rishi,
>> But when you're building a profile from a Velvia IT8 target, say,
>> you're really trying to measure the tone response curve of your
>> scanner, not of the film, right?
> To build tone reproduction curve it is better to use something like Kodak #2 
> grey step wedge (tone is b/w reproduction)
> Profiles are addressing scanner+film+target development, one can't really 
> separate one from the other at this stage and using ICC. It is important to 
> realize that Wolf's development and your development are different; and 
> different enough. Creating the profile for a scanner we actually only try to 
> map the colour gamut of the film. Tone curve is an artistic decision as 
> Graeme rightfully pointed out; and more than colour. Image lives or brakes on 
> contrast. We have a b/w film which is something we can use - but try using ab 
> channels without L ;)
>> I thought that after recording the
>> IT8 chart on the film, Wolf Faust measures the spectral data of the
>> patches on the film using a standard illuminant, then stores these XYZ
>> values in a measurement file that the profiling software then uses to
>> map the scanned values to the spectral values.
> Right.
>> This means that the
>> scanner is trying to recreate the colors on the film, not the colors
>> of the original chart
> Profile is between measured values and scanned values, mapping scanned values 
> to measured values. So it reproduces the chart. If the chart accurately 
> covers the media gamut like Don's charts do - you are very likely to get a 
> profile that allows to render the abilities of the film pretty accurately.
> --
> Iliah Borg
> ib@xxxxxxxxxxx

Other related posts: