Thank you for a well-written and educational piece. This is of particular interest to me. I have a 5000 ED and quite a few 35mm slides that I have scanned over the past six or seven years. I've been disappointed with the results, but I'm not sure why, and wonder if Silverfast would help. I have a perception that the scanner is not focusing accurately, which may or may not be true. As an example, I've attached some images of the raw scan, a 100% crop of the raw scan, and a modified file that looks like the original slide in color and luminance. This was scanned in Nikon Scan 4 with ICE on, 16x, and in 16 (48) bit mode at full resolution. There were no gain adjustments or any other changes in the controls. I chose this image because it exposes (sorry) the detail limits quite well. My problem could be several things:
1. There's nothing wrong; this is the best it gets and I have been simply spoiled by the additional detail from current digital cameras.
2. The scanner does not auto focus correctly.
3. The film just isn't flat in the plastic frame holders. I was interested in the Epson 750 because of the wet mount capability, but you made a comment here that the dedicated (Nikon) scanner gives you slightly better results. I used to think this was the problem but now I'm not so sure.
4. The slide isn't that sharp. I suppose this could be the case, but this was shot with a 70-200 f/4L IS at f/8 on Velvia with a sturdy tripod, and my view through a 10x loupe gives me a better impression of sharpness. This, of course, could be my perception problem in #1 above. I use this same lens on my current 5DII with great results.
Based on the 100% crop, is this a good result? No sharpening or noise reduction has been applied at all.
Do you think this is a focus issue? Would Silverfast's manual focusing help? Nikon Scan allows manual focus, but the feedback is miserable. I haven't figured out a good way to use it effectively.
Is this a mounting issue??
Levels & Curves Applied to Approximate Slide
100% Crop (no adjustments)
Any opinions would be welcome.