I presently have a Colscan LS-9000 and the big brother to the Epson V500, the V750 and I'm reasonably pleased with the results. It becomes a rodeo with bad originals, but that's nothing new.
I have owned and/or worked with a lot of different scanners since 1988, yes, that is not a typo and I shudder when typing it. Medical imaging scanners, drum scanners of various sizes, freestanding flatbed scanners, all manner of desktop scanners, even police scanners! The picture quality from the last one can be, well if you use it wrong, arresting. ;-)
I can generally support the positive thing others say, though there are some caveats.
Given two things:
- A good negative or transparency
- Extra time
With the above, the equipment can produce respectably similar results using the flatbed and the dedicated film scanner.
Given the following extra options:
- A good, fast(er) computer with lots of space
- Fluid mounting accessories (and using them).
- Previously focus-calibrated flatbed.
These latter two options raise the bar and bring the results closer on screen. The differences are radically diminished when printed and are virtually imperceptible to the untrained eye. Here's a full-res (well, matching res) crop from a V700/V750 and a Coolscan LS-9000
There is a cost, there still isn't a free ride:
Using the same negative, the same computer (faster allows more for 'tricks' in a given time-frame) and fluid-mounting, it takes me approximately four times longer to scan with the flatbed to achieve these results.
The basics involve: oversampling by pixel-count and bit-depth (16-bits per colour), multiple tone-curves, multiple unsharp-masking applications and finally down-sampling is thrown in the middle.
Optically, the flatbed is inferior to the dedicated film scanner. The advertised high-resolution, beyond the actual resolution, of flatbeds is a function of overlapping sample areas.
The workflow entails restoring some of the subject contrast, then an application of large aperture (radius) unsharp-masking (there is direct relationship of edge-contrast and sample-rate a.k.a. resolution), a reapplication at a smaller aperture, a subsequent reduction in size (sub-sampling of the image in pixel-count and bit-depth), then final subjective adjustments and sharpening.
Variables to be attentive include the film grain, and how much you wish to reproduce, magnification ratio, and similar. These affect how many pixels render grain and therefore detail which in turn affects how much to oversample, the radius or aperture of unsharp-masking and how much to sub-sample mid-way.
Ultimately, this has the effect of 'tightening' the definition of grain by enhancing its edge-contrast without clobbering the subject detail-contrast (those elements are/should-be too large for the sharpening to directly touch in the first half of the process). The second half of the workflow is pretty much normal.
Of course, you don't have to follow this workflow -- I only do it for film that doesn't fit in my CS9000 (almost as good as a drum for most, good, film).
If you are not striving for large magnification and maximum information, let alone time considerations, there are easier and faster ways of getting acceptable results.
