filmscanners: VueScan, LS30 & Workflow

["Bob Armstrong" <bob.911@btinternet.com>]

It seems intuitive to start by getting the maximum information out of the 
scanner and then make as few changes as possible - and as small as possible - 
to get the desired image.  Perhaps I'll be able to do this one day.  Yes, even 
one day would be nice :)

Using VueScan one can set black point to zero and white point to 0.001.  But 
this generally leaves a flat image which needs correcting by levels or curves.  
Alternatively, VS's white point can be increased until none of RGB exceed 250 - 
255 (for simplicity assume no specular highlights, also, there is rarely a need 
to change the black point).  Doing this necessitates:
  Do
     a full scan
     export to editor with histogram function
     look at max levels for each of RGB
     guesstimate change of white point
  until RGB close to but not exceeding 255

This gets the most out of the scanner and is somewhat similar to Ralf Schmode's 
workflow using Nikonscan.  However, Ralf can do this in Nikonscan from a 
preview before ever doing a full scan.  Even using the histogram in VuePrint 
(which loads files quickly) the time taken to carry out the process makes it 
unworkable.

I have wondered if any difference results from using VS's white point setting 
so that the image uses the full range of the histogram compared with the image 
using only part of the histogram range and adjusting in Photoshop using curves 
or levels.

I recently downloaded IrfanView to use for thumbnails and, looking around the 
menus, noticed a Count Colors Used function.  Scanning a raw VS file and using 
different white points, I took a note of the number of colours used. I then 
loaded these files into Photoshop and did auto levels on RGB with black and 
white points set to clip 0.00%.  The numbers came out like this, colours are in 
thousands and rounded to 3 sig figs:

White    VS     VS max             PS
Point     Clrs    R     G     B       Clrs       
0.001    574    205 204 208     573
0.040    816    249 248 253     816
0.100    832    253 255 255     798*
1.000    879    255 255 255     879

The 0.04 setting gives about the right level of maximum RGB.  Another negative 
produced a similar pattern of results but required a white point of 0.15 for 
optimum results.  I have not done any other tests.  It seems that one can get 
40% more colours from VS by stretching the image to fit the histogram by 
adjusting the white point.  Even more colours are produced where the white 
point is set to clip the highlights.  This is all very well but what the hell 
does it mean?  Should I be setting the white point to just below clipping 
before scanning?

Bob Armstrong

* Weird.  It seems that PS does something strange around the clipping point 
when clipped and unclipped channels are present.  Both 0.04 & 1.0 each produced 
exactly the same number of colours from VS and PS.