Hi Don,
Flare can be an issue in shadow areas, yes. That may or may not be a reason for a given individual to reject a DSLR slide capture, so I mentioned it only in passing. As to the quality of scanner lenses, I can't speak for the Crosfields, but here is a statement from a Mr. Christian Kammerer of Linos (now the owner of Rodenstock and other optical firms) on the Spindler & Hoyer objective installed in the Screen scanners you love to malign but which can deliver great results when used carefully:
"In addition I attached some optical quality data (MTF and Through focus MTF) adapted to your application. The data represent a nominal system without production tolerances. For calculation I assumed that Dainippon Screen implemented the lens according to it's specifications (microscope lens with 10 times magnification).
The MTF Datasheet shows the MTF Values over the imaging field for 20, 40, 80 and 160lp/mm. At 160lp/mm (approx. 8000dpi) the MTF is still l over 50%. which is very good. As it is an apo-chromatic lens, the MTF is good for all 3 scanning channels too. The through focus MTF (MTF for different focus position) shows, that the lens is very sensitive. In other words: changes of the drum position by 10µm will cause a significant loss in sharpness. But this is quite normal for a microscope lens.
In terms of imaging quality the lens you have should be very good for this application (presumed it is used as it was designed). If the Auto focus System of your scanner works well, the focus sensitivity should not be a problem. "
They obviously took some care in choosing this lens. "Very good" should be understood to mean as a microscope lens, which as you know must operate close to or at the Rayleigh limit. The issue of depth of field should be taken seriously in both scanning and camera copy work. 35mm slides in mounts are quite seriously curved, and this may dictate a lens stop of f/11 or smaller to render the entre image acceptably sharp. At such a small stop image softening from diffraction can be seen in large prints. This implies dismounting films from mounts and (in the case of 35mm) keeping them flat in some sort of glass carrier, preferably in fluid. This is impractical for bulk work. Stitching likewise woudn't be part of slide copying at 1:1, and one is stuck with whatever the system can sample in a single frame.
As to how much detail one needs to resolve for a sharp print, this is hard to define for various reasons, and I do agree completely that the value of an image can hardly be summed up this way in any event. The different, narrower question was whether a DSLR can make a copy sharp enough to pass critical scrutiny. In a way it's the same as asking whether one needs a really good enlarging lens and glass carrier for darkroom work. There's no question (at least by anyone who has actually done this sort of work) that one can get a sharper result this way. Will anyone be expecting the image and even film grain be crisply rendered right into the corners, at 10X enlargement? When the film itself starts to become a subject, the requirements get tighter. Sometimes this matters and other times not.
I have not seen noise in shadows I could definitely attribute to the drum scanner, though film itself is of course "noisy" and a sharp scan will render this more clearly. In in very dark areas one can even see light quantization (impact of single photon strikes) in the original image, which looks like noise--this applies obviously also to digital captures if the exposure is set just right, though it might be hard to distinguish from actual circuit noise. Needless to say, exposures should be generous enough to avoid these problems.
Don, I wish you would address the question of 24- v 48-bit captures, as it's part of the original question and a matter you've written about and shown examples of.
Cheers,
Mike
>
> Message: 1
> Date: Sun, 17 Jan 2016 17:35:50 -0500
> From: Don Hutcheson <don(a)hutchcolor.com>
> To: colorsync-users(a)lists.apple.com
> Subject: Does MF color slides scanning in 24 bit still make sense
> Message-ID: <D3216099-B816-471B-88B1-688675F37F4C(a)hutchcolor.com>
> Content-Type: text/plain; charset=windows-1252
>
> The main issue with camera capture vs drum scanning is not resolution nor dynamic range, it’s flare.
>
> Before discussing what I mean, let’s look at the question of resolution:
>
> Using a high quality, short focal length enlarger lens and sufficient stitching, any high-MP camera can easily equal or exceed the so-called “sampling resolution” (pixels per inch) of any drum scanner, for whatever that’s worth. Pick your pixel count and you can achieve it, subject to a little care in the stitching.
> But how much resolution do you need? The sharpest low-ISO slide films were capable of perhaps 60 line pairs/mm at 40% MTF, which, according to Nyquist, needs a sampling rate of about 6,000 dpi, give or take. But good luck finding a film original with such extraordinary image detail. It would have to have been captured (a.) on Kodachrome 25 or similar, (b.) with a very fine macro optic at optimum aperture and (c.) on a perfectly stable tripod, etc. Most of the images we think of as “tack-sharp” have about half that real resolution if they’re lucky, and we love them.
> Bottom line - whatever sampling resolution you decide you need, you can get it by camera stitching. Albeit with patience.
>
> Of course there’s the question of whether camera-captured pixels are as "sharp” as drum scanner samples. I won’t get into that, except to say that you’d be surprised how cheap some drum scanner lenses were. The notion that Crosfield, Hell, Screen and ICG spent the farm on high quality apochromatic microscope lenses is a little exaggerated.
>
> Next let’s look at dynamic range.
>
> Most good slide films have a dynamic range of at least 4.0 (max. - min. dye density) which corresponds to an f-stop range around 14. Interestingly, most drum scanners struggled to get anywhere near 4.0 DR. The log amp (logarithmic photomultiplier tube amplifier) in Crosfield scanners was artificially limited to about 3.0 but I was able to squeeze about 4.3 out of ICG scanners, with a bit of PMT noise.
> Few digital cameras can approach a 14 stop range in one shot, but with careful HDR, they can easily exceed it. However this is where flare begins to muddy the waters (or shadows).
>
> The flare problem.
>
> The one area in which drum scanners easily trounce flat-bed scanners is gross optical flare. The only flare you’ll find in a drum scanner is within a few microns of the scanning spot, the actual flare radius depending on the scanning plane optics. However all line-at-a-time (a.k.a. “CDD” or “flat bed”) scanners suffer from a much more severe problem that I call “linear flare", manifested as streaks of unwanted lightness where light and dark image areas meet. What makes this flare offensive is that it only occurs at right angles to the scanning direction. These directional flare lines are caused by the fact that light image areas must be illuminated at the same time as dark areas in the same scan line are being sampled. Drum scanners don’t do this.
>
> Camera captures also suffer from optical flare in the lens itself, but lens flare is often less noticeable (unless you go looking for it) than flat-bed scanner flare, because (a.) it’s “radial” and therefore symmetrical around the bright detail and (b.) it is often mistaken for flare in the original, i.e. caused by the original taking lens. Unfortunately, if you rely on HDR to expand your camera-copy dynamic range, the copy-camera’s lens flare can actually be magnified. That’s because the HDR process typically relies on some form of blurred masking to blend light and dark exposures. The blurred mask can itself contribute to ghostly shadow anomalies if you’re not careful.
>
> The good news is that scanner-produced flare (whether linear or radial) is mostly noticeable when you have to exhume dark shadow detail, i.e. open up an under-exposed transparency. Normally-exposed slide scans usually look fine.
>
> OK. That’s enough typing for a cold New Jersey Sunday.
>
> ........................................................
> Don Hutcheson, President
> HutchColor, LLC
> Washington, NJ USA
> don(a)hutchcolor.com
> M: 908-500-0341
> ........................................................
>