Technique for scanning color negs. on M1

[OldBikerPete]

While discussing development of C41 color on the analog site, I had occasion to bemoan the fact that - for color negative scanning, scanners and their software (in my experience) do not behave like enlargers. ie When I have repeated a sequence of actions precisely, the results do not repeat. The following is a dupliacte of a post i made towards the end of that discussion.
-------------------------------------------------

To my knowledge, recent scanner are quite clever enough.....<snip>.....

Too bloody clever by half. Duplicate scans of the same neg. were wildly different in color balance and had cross-overs - different ones. That's using the software which came with the M1 - both the Microtek and the Silverfast. I'm waiting impatiently for Vuescan to support the M1.



Although just in the last week or so I had an epiphany.

In setting up my development process I made use of some Kodak process control strips and a densitometer.

I had the idea of using the Microtek software which has in it the facility to enter custom Dmin and Dmax numbers. (I had been leaving the density range on automatic and the software had been arbitrarily clipping the highlights and shadows off my images and screwing with color balance). I had these density numbers to hand from process control strips, so I entered them into each of the R, G and B channels of the custom density setting and --------- voila -------- instant repeatability.

====================================
The numbers I used were:

R: 0.30 - 2.40
G: 0.80 - 3.40
B: 1.00 - 3.90

I was using 5x4 Portra VC
===================================
Gross color balance can be tweaked by making small changes to the Dmin values after viewing the prescan.



The initial scan (I use 48-bit scans) is weak and low-contrast and the negs I have scanned only use about 1/3 of the full density span. I save the scan as a tif and

then open it up in Photoshop. Add adjustment layers for (respectively) levels, brightness/contrast, hue/saturation, color balance and curves and I finish up with an image that looks like a good scan of a good transparency without any of the difficult and guesstimated color shifting I had to do previously.

 

[mts]

I am trying to do the same, i.e. produce a scan that is not manipulated by software or firmware to compare processing variations. I am using 35 mm and scanning with a very old Nikon LS-3500 albeit only with 24-bit scans.

This old unit is decently high resolution and for my purposes has the advantage of being fairly "dumb." The basic scan is made with linear gammas using only an autoexposure. Successive scans are consistent. Scans use RGB separation filters. For color negative film only the brightness and color balance has to be corrected to produce a correct grey level much the same as for color printing.

While better results can be had by setting gamma corrections for the separations, doing so confuses the comparison of films and processing. In color printing there is essentially no option beyond brightness and color balance, but scanning and digital manipulation can ultimately produce a much better result than I can make in the darkroom.