"I'm not sure if this is a problem for all CCD scanners. Just for comparison I had a 35mm Pro 160S color neg scanned on a Flextight X5 at 8000 ppi. I have to say that the results were actually quite good. The optical resolution of the X5 is definitely in the same league with the best drum scanners for 35mm frames. I couldn't detect any grain aliasing in the scan. But on pixel level there was a scan line pattern visible in the highlights of the blue channel and to a lesser degree in the green channel too."
Unfortunately the Imacon is grossly overrated for 35mm resolution, which might actually account for the scans looking better than you expect. It's actually returning less detail that the pixel count would suggest. That has always been a big problem with Imacon, not to mention the blooming and lack of density range. That you're also seeing artifacts on a color neg scan is further evidence that it doesn't have the density range.
"If the micron value is the diameter of the aperture than the optical resolution of the scanner could be better than the micron value suggests because with finer steps these larger analysis spots would overlap. This would reduce contrast but still get you some more resolvable detail as long as you've got enough contrast on film. It's the same effect how diffraction reduces contrast when the central diffraction pattern is larger than the sample size."
From what I've been told, the micron size is indeed the diameter of the hole in a metal disk that rotates in position behind the lens and before the light is split to the three pmt's. There has to be an overlap in the spot in order to have a smooth transition as the drum moves in front of the lens. I do not know exactly how this is programmed when you ask for other than the automatically designated apertures, but I do know that if you manually set an aperture that is too small for the requested resolution, there will be gaps in the overlap.
"On my ICG scanner there are 9 numbered apertures, 1 being the largest and 9 the smallest. There are no micron values. ICG does not want to put any details about their aperture sizes into the public because the whole optical system including lens design need to be taken into account. In the scanning software I can chose to use larger apertures than the automatic choice but this only softens the scans."
I'm pretty sure that the highest end ICG uses the same 3.17 micron aperture that the Howtek/Aztek uses. Does your scanner do an optical 8000 ppi? If it does, it pretty much has to have that size for the smallest. I'm sure if you contact them they should be able to tell you. It shouldn't be a huge secret. If selecting a larger aperture only softens the scans, then something is awry, or you're already using an aperture (unknowingly) that is optimum for the res you are requesting. All drum scanner work on the same basic principle, but differ slightly in optics, physical specs, and of course, software.
"So I never make any use of this option, and I haven't experienced the sort of grain aliasing that I've seen demonstrated by Howtek and Aztek scanners. Maybe ICG is taking a different approach."
The software used for scanning makes a huge difference, particularly in color negs. Lenny Eiger uses Aztek's Digital Photo Lab, which in my experience, is anything but... It was the worst at scanning negs, black and white and color, than any of the software packages I tested. While it did allow manual setting of the apertures, it also had a nasty habit of plugging shadows and blocking highlights and had a non existent color management implementation. Moving the same scanner with film mounted on the drum between Trident and DPL, Aztek's software showed poor color (well you also had to guess as to what color profile to assign) but much harsher tonal transitions, particularly in out of focus, creamy bokeh type of areas, the type of areas where you want smooth transitions.
I've spent a dozen years scanning all types of film with Trident and just haven't seen what some people call "grain aliasing". What I have seen, for instance, when requesting a 4000 ppi scan from Portra160, is a gradual softening of the grain as the aperture is opened from the "auto" setting of 6.35 microns to 9.4 to 12.7 to 16 and then finally to 19 microns. At somewhere between 13 and 16 microns, the softening of the grain gives over to softening of detail but up until (usually) about 16 microns, I don't see any loss of detail. Specific emulsions will yield optimum results between 13 and 19 with grainier high speed emulsions maybe going up to 22 microns.
"If the physical aperture limits your resolution to the spot size then your effective resolution of the scan wouldn't be better than from a 2000 ppi scan, except for maybe some slightly smoother transitions. But have you compared a 4000 ppi color neg scan to a blown up 2000 ppi scan where you used the same aperture?"
Funny you should ask, because in the earlier days of Trident, and perhaps still in there, there was a setting where you could limit the optical resolution when the software was set to scan negs. You could limit it to whatever you wanted but the default was 2000 ppi, which was probably perfect for the prepress scanning done in the early to mid 1990's. I did quite a few comparisons and came to the conclusion that scanning at the requested res and setting the aperture manually yielded superior results to simply rezzing up a lower res scan in Ps. And for one, if you did select 2000 ppi, your auto aperture setting would be 12.7, not bad, but for most low speed color negs, just a tad on the small side. By using the "wrong" aperture, you're still getting the higher number of "steps" in the long direction, giving slightly better rendering of both detail and grain. The difference isn't great but it's there.
Okay, now off to bed. Got a big studio shoot tomorrow. Who like to light polished, hammered stainless??? Yeah, just what I thought.