In article <dW7Ab.66148$>,
says…
Johan W. Elzenga wrote:
Mike Russell wrote:
Many people do prefer to make any large moves during the scan process, but it’s not really important to do so. One reason to get it right in the scan is to reduce the time spent processing a large number of images. Photoshop has more flexibility, but it is an extra step in what can be a fairly lengthy process.
There is another reason as well: If you make your corrections in the scanner software, you make correction in the full color depth of the scanner, which is usually more than 8 bits per color. If you make corrections in Photoshop, you will do so on 8 bits images. Unless you set the scanner software to output in 16 bits (but that is indeed time consuming, because all files will be double in size).
And, to date, and Bart’s examples notwithstanding, no one has shown me a photograph in normal gamma space for which this made a whit of difference.
On normal images with moderate corrections, it may indeed not be visible. Only the histogram shows the difference, but I agree with anyone who says that we are not in the business of making histograms, but images. It is however quite simple to show how excessive corrections can cause banding in a 8 bit image, while the same corrections on the a 16 bits version do not. A gradient works well for this. In a ‘real’
image the sky (or water) can be like such a gradient, so this is certainly not a theoretical possibility only. Yes, I’ve seen images with banding in the sky which was probably caused by excessive corrections in 8 bits color. As long as I don’t have a 16 bits copy, I obviously cannot *prove* that this could have been avoided if the corrections had been
done in 16 bits, but I’m pretty certain it would have been.
I agree with all your points,. including that the image you saw with banding in the sky, that would have been fixed with a 16 bit original, is somehow unavailable for anyone to actually see.
I have examples of 8 and 16 bit workflow in the tips section of my web site.
As far as setting levels at the scanner step, the most important thing is to watch that important details in the highlights and shadows don’t get clipped.
Very few scanners/software affect the actual data captured. Most apply the changes internally after the scanning step, even if it appears that it is all one process.
Personally I prefer to use curves rather than levels since you can control not only black and white points, but contrast in several regions of density. Levels applies a default curve that you can only control at one point.
—
Robert D Feinman
Landscapes, Cityscapes, Panoramas and Photoshop Tips
http://robertdfeinman.com