I have not extensively pursued this yet, mostly because I'm quite happy with the results I'm getting with my film scanners.
But about 6 months ago I gave it a twiddle. My setup:
- Nikon D80 10mpx, using RAW.
- Copy stand to hold camera and bellows.
- Slide sorting box at the bottom of the stand, with 35mm 6 image film strip on it and a piece of AN glass on top to keep all flat. This is a vanilla slide sorter, the kind you get with a rectangular A4-size plastic diffuser flat viewing surface and a diffused light source behind it. I think the light is a low-voltage neon tube of daylight balance, because it flickers when started. 12V power source. These boxes are available at most photo shops.
- Nikkor 105/2.5 lens mounted on a cheap HK-sourced bellows.
With this setup I was able to obtain good results with b&w negatives and reasonable from positive "slide" film.
In a nutshell, I noticed this:
- The camera must be colour balanced for the light source, even though it is supposed to be daylight. I used the standard D80 auto colour temp setting and that produced the most pleasing colour balance, matching the original. Didn't record what the actual colour temp setting was.
- The diffused light source gets rid of most if not all scratches, imperfections and other surface defects of the film. Including most drying marks. No need for the infra-red d-Ice of the scanners and that won't work with b&w negatives anyway.
- Grain is less noticeable than in a film scanner. This might be a consequence of the next point.
- 10mpx is waaaaay too low for the definition I'm used to getting from my film scanners. It however produces perfectly acceptable "scans" from slides and b&w, at that definition.
- Getting the framing just right is fiddly with my setup and takes a long time.
- Flatness was not a problem, but field curvature from the lens was. Hard to focus precisely, mostly because the focus point changed slightly as I closed down the iris. I ended up using the lens wide open, with slightly less sharp results than I could get if I spent the time to get precise focus at smaller apertures: it took less time that way. Not sure if this focusing problem was a result of the lens itself or something else. The 105/2.5 is not a macro lens, so this might be the problem.
- For normal slides and negatives, I didn't find dynamic range a problem. The results were evenly illuminated and there were no "hot" or "dark" spots. Have not tried with "difficult" film images.
Things I'd like to try next time I give it a go:
- Higher resolution. I think the Sony A900 or similar would be ideal. It matches what I can get from my very best film images and is the cheapest of the FF dslr brigade. I don't need high ISO for this, either.
- A proper macro lens setup. Maybe an enlarger lens? I've got an old slide duplicator that might work better, perhaps?
- A less fiddly setup to move the film strip. I need some way of keeping the flatness I get from the viewer and the AN glass on top and still be able to slide the lot across to quickly " scan" the strip. I'm thinking along the lines of a "sled" sandwich.
Is it faster overall than a flatbed scanner? I don't think so, at least not with my non-dedicated setup.
Is it faster than a film scanner? Yes, but nowhere near the top quality of those.
Can it cope with difficult, contrasty slide film images? No idea, but those I can use my film scanners for.
For a quick and dirty scan of an entire roll of film, for cataloguing purposes and small image output (certainly Internet size or APUG gallery size) , it's a feasible proposition.
Of course: at the stage I've pursued it so far, it is not a quality replacement for my film scanner workflow. But it's something that I intend revisiting.