Have we reached scanner sharpness limits yet?

[leicaglow]

John's question about keeping two scanners got me thinking: Have Nikon's newest scanners reached the theoretical

limits for sharpness yet? New IR technologies, like ICE4 make better dust free scans, but other than that kind of

technology, could more be done to provide sharper or higher resolution film scanners? I.e., should we be looking

for a Coolscan VI or 10000 at some point that goes beyond 4000 dpi?

[franklin_polk]

I don't think we have reached scanning limits quiet yet, but I wouldn't hold my breath for a Coolscan 6000. Nikon seems to pretty much have stopped development of scanners (as evidenced by the discontinuance of the Coolscan V), sadly. Drum scanners can provide much better quality scans than the Coolscans, but it seems, like Nikon, there really isn't any new development going on, plus most people don't have 50 grand to spend on a drum scanning setup.

[kelly_flanigan1]

Film and scanner Scanners sales volume track the dow jones average; except the peak occured a few years back with scanners. The Dow Jones divided by pi is the limit of a scanners resolution. The target price of a scanner for an amateur is 180 times this; since 180 pi ohms is the impedance of free space.

[ellis_vener_photography]

At this point in time there are a few things that limit a scanner's ability to make scans that are virtual digital copies (meaning as faithful as a copy (second generation can be) version of the original piece of film are:

 

Film flatness.

 

--This is addressed by wet ("fluid") mounting of the film against a rigid glass surface - possible with the AZTEK film holder for the Coolscan 8000/ 9000, with drum scanners and at least theoretically possible with an Epson V750M for medium and large format film.

 

Absolute resolution.

 

--Partially addressed my holdingthe film flat and in the precise plane of focus for the scanner, definitely possible with a good drum scan but the Nikon Cool scan at 4000ppi comes close enough for al lbut very large prints.

 

Real world dMax capacity that matches or exceeds the dynamic range of film .

 

--We are actually really close to that with the Coolscans. Wet mounting helps.

 

The operators skills.

 

--Those can only be learned through diligence and application.

 

Color fidelity we have: use a 16 bit per channel color space that is tuned to film (AKA Chrome 100 from http://www.josephholmes.com )

[mendel_leisk]

The Minolta Scan Elite (versions I and II) *were* available, with 5400 dpi. With mine (the first version), I did some

tests downsampling to 4000 with scans of decent slide film, and decided 5400 dpi was worth it. For a lot of my slides

it was overkill, just rendering ever more detail out of film grain, but I saw enough cases where actual detail became

indistinct, going from 5400 to 4000, to justify scanning at the full 5400 dpi. File size more-or-less doubles in that

step.

 

Anyway, film *grain* is the ultimate limiter, not much point in going *much* higher.

[leicaglow]

Kelly, if they were only that cheap!<g>

 

Ellis, thanks for all your knowledge on this subject.

 

Mendel, That's really why I asked. I've been shooting a lot of digital, but buying a Canoscan (albeit old and only 2700 dpi), I love the look of the grain, but it doesn't seem quite high enough to see the grain in some cases.

[melmann]

Michael,

 

Like Mendel I have the 5400dpi Minolta scanner that I use with Kodachrome 64 slides shot on my OM1. A full resolution

scan results in a 255Mb file for each slide and I find the quality of my lens starts degrading detail before the grain does (not

that Kodachrome shows grain....). I archive my slides to digital at around 1400dpi, giving me files of about 14Mb in size

which are easy to work with for processing and printing, or resizing for websites, and I have no problem with the detail I get.

Only rarely do I re-scan at the higher level, and then it's usually because I want to crop down to a very small portion of the

slide.

[mendel_leisk]

The Coolscan's at 4000dpi are pretty good. While the dpi is a little lower (than 5400), I think the depth of focus is better, and the lens is good quality. I have a recently acquired Coolscan V as well as my Minolta scanners. My first Minolta was a Scan Dual II, which scans at 2820 dpi, just a bit more than what you're contemplating. I did my Tri-X with that. Recently I recscanned a few with the Scan Elite 5400, the difference is amazing. The 2820 dpi scans are fine in their own right, but if you compare the 2820 with 5400 it's a little depressing, the difference.

 

Anyways, I think the Nikon's at 4000 dpi will deliver good grain. As I said, I think they have a slight edge in overall sharpness, due to light source/lens/depth of focus, which makes the most of that 4000dpi.

[Ed_Ingold]

Grain is not the only factor which determines the resolution of film, nor is it the most important factor. Halation, optical and chemical diffusion play a significant roll. Using an high contrast resolution target is perhaps the most reproducible standard, but ordinary scenes have a much lower contrast, hence present less detail.

 

I have attached an MTF curve for Fuji Reala, one of the sharper films for landscapes. This curve is typical of film, and represents the resolution independent of imaging using a lens. The contrast is fairly constant below a certain frequency then rounds off and takes a plunge with a corner frequency (for Reala) at about 30 lp/mm. You will notice that the MTF 50 is only about 50 lp/mm. THe results in a camera, using a lens of a scene with average contrast will be much lower. MTF 50 is a good measure of how much detail will be visually appararent.

 

The resolution of a Nikon scanner is about 80 lp/mm, which is somewhat greater than the practical resolution of most film. Because of the properties of film and other uncertainties, there is relatively little to be gained by scanning at higher resolution.<div></div>

[leicaglow]

Edward, you crack me up. I was hoping you'd jump in with your science lesson. I know grain isn't the only factor in film sharpness, but am I missing something in thinking it's a major part of what I'm trying to capture with a film scanner?

 

Thanks all. This has been very, very useful to me.