Nikon Coolscan dilemma

[matej_izgrada]

I'm a serious photo enthusiast shooting exclusively film. After saving for a

long time (college student) I have just enough to buy Nikon Coolscan V.

 

However, I'm in dilemma after comparing it with Coolscan 5000. It costs 50% more

(where I live), and it would mean saving for long time to come, but it does seem

to have some interesting features. What I would like is to hear your experiences

on whether or not this 50% more money is worth it.

 

Here's the situation...

 

I have a lot of film that will need serious scanning, meaning it's occassionaly

very dense. I need those dense areas in as much detail as I can afford, so a

Dmax of 4.8 for 5000 would be better than 4.2 that V has.

However, I know Dmax is limited on one side by noise scanner introduces, and I

also know that using multisampling technique reduces that noise. Multisampling

of 10 images increases Dmax by 0.5. I am aware that only 5000 has multisampling,

but I know how to do it manually in Photoshop.

 

So given all this, will Coolscan V suffice for a few difficult scans? Or would I

get much better results with 5000 that would justify higher price? (I presume

Dmax of 4.8 combined with multisampling would further expand the possibilities)

 

 

As for 14bit/ch vs 16bit/ch precision -- is there any application where you can

actually see that difference?

I know Dmax and bit precision are two separate things and that 16bpc is nothing

without a decent Dmax. So does 16bpc only fill for higher Dmax on CS5000, or

could Coolscan V also benefit from more precise A/D conversion?

 

 

I also have a few important shots (negative film) which are underexposed. I'd

like to try and get out as much information off the film as I can. Is there

something 5000 has and V doesn't that would make a difference in this effort?

 

 

CS5000 is twice faster than CS-V, but speed isn't an issue.

 

 

Thanks for your replies!

[Ed_Ingold]

The only feature of exceptional value in the LS-5000 is the ability to use a roll feeder and automatic slide feeder, neither of which you mention. Scanning takes a significant amount of time, and using a roll feeder can cut the processing time by up to 2/3rds by reducing the feed time (external time to the industrial engineers). Most of us can't afford to sit in constant attendence to a scanner, so there tends to be a lot of idle time between film strips.

 

The LS-5000 is also much faster per scan than the LS-50. Again, time is a significant burden in scanning, making it impractical for professional use. For an hobbyist (or for fine art, which is practically synonymous with "hobby"), time is less important. You are able to get extraordinarily good results with a film scanner when your time does not have a price tag. A faster scanner means you can have more of a life outside of scanning.

 

It is questionable whether there is a noticeable difference between 14 bits/channel and 16 bits/channel, or if any sensor can actually deliver 16 bits/channel. The bit depth has little effect on dynamic range, rather the size of steps within that range.

 

DMax is not a measure of dynamic range, as commonly thought, but the maximum density in an image that the scanner can measure with respect to its noise floor. This is an issue with high-contrast color reversal film (e.g., Velvia), and with overexposed B&W negative film. It is never an issue with color negative film, nor with reasonably exposed B&W.

 

Multi-pass scanning works by averaging successive readings. Since thermal noise is (essentially) random, averaging reduces the effect of this noise on the image. You must double to the number of passes to effect a noticeable improvement on the previous level. This may allow a little more detail to be uncovered, but adversely affects the resolution, due to mis-registration between passes. I find it of little use with a Nikon film scanner, but an essential element when using flatbed scanners, which seem to be inherently noisy.

 

Nikon film scanners perform multiple passes within the same scan line, which minimizes the mis-registration effects. Flatbed scanners usually repeat the entire scan, and mis-registration can be significant. I see no way to emulate multi-pass scanning in photoshop, although you could apply gaussian filtering to reduce the appearance of noise. You could overlay multiple passes of the entire scan, at the risk of major loss of sharpness due to mis-registration.

[rombon]

As already said the main difference between two Nikon 35 mm film scanners is time, needed for a scan. Photographers seem to systematically underestimate the time needed to scan even moderate number of frames. I have serious problems explaining people how notorious time consuming is scanning. Beside even with the slower model the scanning time is not long enough for someone to focus on any other serious work on computer between two inserings of the film. So it is really very important to have fast scanner if you are doing volumes of film. The quality of scans with both models is good and almost identical.

 

Regards, Marko

[gene_e._mccluney]

This issue of the time it takes to scan is a good topic. For me, scanning b/w negatives, or

color slides that I have processed myself (and still in filmstrip form, not mounted) I can

average about 24 scans in about 8 hours, and this includes the adjustments and dust

spotting in photoshop. I have the Super Coolscan 9000ed, but I am referring to scanning

35mm format film at the highest resolution. I scan and adjust and dust spot at a detail

level good enough for very large prints. I only want to scan a negative once. I can always

downsize later. The point I would like to make is the actual "scan time" the scanner takes

to produce the initial scan of a negative or transparency is only a part of the overall time it

takes to get a finished file suitable for printing. I think I am safe in saying that NO scanner

produces an ideal image, in and of itself, and you will be required to open the scan in

Photoshop and make adjustments to get exactly what you want. Your speed and skill in

using Photoshop will greatly affect the total time needed to produce a printable file.

 

McCluney Photo

[matej_izgrada]

Edward,

I'm aware of incredible usefulness of automatic feeders, and as much as I would like to have an automatic roll feeder, it's price is comparable to Coolscan V (3/4 of its price here). So as it is, these feeders are useless to me.

 

On a summer job I had in highschool, I used Nikon LS-2000 to scan big amounts of negatives, and I'm painfully aware of the time it takes. However, given the time I've been saving to buy a film scanner, waiting for a scan will be pure enjoyment. :) I'd like a faster scanner, but only speed isn't worth the difference in price in my case.

 

You say Dmax isn't as important for color negatives as for B&W film. Why is that? Can't color film get very dense too?

 

As for simulating multipass scanning in Photoshop, I successfully replicated the technique of noise averaging from multiple passes (on flatbed scanner examples) by mixing layers with equal weighting factors, but I ignored the possible effects of misregistration as there's nothing I can do about it...

I was kind of hoping loss of sharpness wouldn't be too dramatic, and the revealed details from dense areas of positives and B&W films (which I have tons of) would be worth it.

 

 

Could someone tell me how obvious is difference in Dmax between 4.2 and 4.8? Will I be able to do without it if only occasional shot needs diggiging deep into dense areas?

I have some color negatives with overexposed faces I would really like to salvage if at all possible, so I was hoping 4.2 would suffice.

 

 

Gene,

I too was planning to scan each film only once, get out of it as much information as I can and then rest easy knowing that should anything happen to my films, I have them backed up in 4000dpi/16bpc.

 

 

Thanks for great answers, everyone!

[Ed_Ingold]

Matej,

 

You misread my response. To rephrase, DMax is ONLY important for high-contrast reversal film (e.g., Velvia) and overexposed B&W. It is NOT important for color negative film, nor properly exposed B&W. The dye used color negatives is not nearly as dense as in reversal film. If you are as experienced as you imply, you must have seen many color negatives ;-)

 

I miss the salad days when time was not important. It's a shame they're not making any more of it for us who need it.

[matej_izgrada]

Edward,

 

I heard many times that Dmax only plays a role in positive scanning, but after inspecting some color negatives the other day that appeared pretty dense in bright areas, I started wondering why do people never mention that. I didn't realize it was a dye issue.

If I had a lightbox to directly compare negative and positive film over homogenous diffused light I might have realized that, but right now the way I look at films is put them against nearest lightsource so it's tricky to compare :)

 

Thanks again for explaining!

[scott_turner]

<p>Matej,</p>

 

<p>Others have given good and extensive advice here, but I'll add one thing. You ask:<p>

 

<p><i>Could someone tell me how obvious is difference in Dmax between 4.2 and 4.8? Will I be able to do without it if only occasional shot needs diggiging deep into dense areas?</i></p>

 

<p>I highly, highly recommend that you read <a href="http://www.scantips.com/basics14.html">this section</a> of Wayne Fulton's excellent "Scantips" website for a better understanding of DMax/Dmin and dynamic range in scanning. Be sure to check at least the linked page, and the one following.</p>

 

<p>Among other things, you'll learn that the numbers you're quoting for the Nikon scanners (and for most others on the market as well) are pure marketing fluff. They're merely derived (calculated) from the <b>theoretical</b> maximum value possible from an A/D of a given bit depth. Given that no A/D can achieve theoretical perfection, well... y'know. And that's not even considering sensor issues and limitations. As Mr. Fulton says, <i>"If comparing a 4.8 spec with 4.2, don't believe all you read, neither value is achievable in any CCD scanner."</i></p>

 

<p>I've used and compared the Nikon V, 5000 and, more recently, the 9000. As others have mentioned, there are substantive differences between the V and 5000, but they're mostly about speed and/or convenience. In terms of the real-world ability to deal with deep shadows that you're concerned with, I've seen absolutely no difference between these three scanners, even when pulling up shadow detail in post-processing. None. Their real-world dmax is essentially identical.</p>

 

<p>Scott</p>

[ilkka_nissila]

Well the A/D isn't the problem, it's how much noise you have in the detector and how much light enters it.

 

The Coolscan 5000 achieves a better dynamic range by increasing the scanning time - this is the multiscanning feature. If you select 16x, it will spend 16 times as long on each row, and average the outcome, in priciple reducing noise by 2 bits. So that's the difference. But scanning times will be longer if you use this feature.

[scott_turner]

<p>Ilkka,</p>

 

<p><i>Well the A/D isn't the problem, it's how much noise you have in the detector and how much light enters it.</i></p>

 

<p>That's exactly my point. The Dmax numbers put out by scanner manufacturers are typically based on the theoretical capabilities of the A/D, resulting in numbers that are pure fiction and aren't even remotely possible in a CCD-based scanner. Worse still, even when you accept that they're not "real" you still have to come to grips with the fact that they're also useless in any comparative sense. A stated Dmax of "4.8" in one scanner may or may not be better at dealing with shadow detail than the same manufacturer's "4.2" scanner, because both numbers are based on something with little relevance.</p>

 

<p><i>The Coolscan 5000 achieves a better dynamic range by increasing the scanning time - this is the multiscanning feature.</i></p>

 

<p>Actually, it's multi-<i>sampling</i>, and while this may be true in theory, in my experience it doesn't mean much in practice. I've used all of the current model Nikon scanners, and have found all of them to be very, very noise free and astonishingly good at getting into the shadows of dense transparencies. If there's real benefit to be had from 16x multi-sampling, it's awfully, awfully minimal, and it requires an awful lot of patience to get to.</p>

 

<p>In any case, that feature still has nothing to do with the stated numbers offered by Nikon.</p>

 

<p>Scott</p>

[matej_izgrada]

Thanks Scott for addressing the essence of my remaining questions!

 

I was wondering, regarding misregistration talk above, about your experience with multisampling: how much does it influence sharpness? And is that only obvious at 4000dpi?

[scott_turner]

<p>Matej,</p>

 

<p><i>"I was wondering, regarding misregistration talk above, about your experience with multisampling: how much does it influence sharpness? And is that only obvious at 4000dpi?"</i></p>

 

<p>Since the 5000 works by taking multiple samples at each scan line before the stepper motor moves things, I doubt you'll see any visible difference in sharpness at any resolution.</p>

 

<p>Note that using 3rd party software (either Vuescan or Silverfast AI), you can do multi-<i>pass</i> scanning with the V as well. Here, while the Nikon's stepper is extremely accurate, you will begin to see some accumulated registration error as evidenced by a slight softening of the averaged result as you increase the number of passes. I've never tried to quantify that softening as, again, I haven't found the feature to be worth the effort. It's not awful, but by, say, 8x or so, it's certainly noticeable.</p>

 

<p>Good luck, what ever you decide.</p>

 

<p>Scott</p>

[scott_brim]

Just as a point of interest here, I've taken Edward Ingold's advice about using Silverfast with the CS 5000 and find that it does a much better job of reducing noise in dense areas of Kodachromes when using multisampling than does NikonScan. But the software is expensive, however.

[rich815]

Get a used 4000.

[matej_izgrada]

Well this is bordering with science fiction (at least in my world), but I just got an offer from the local Nikon dealer where I get Coolscan 5000 for the price of Coolscan V! I intend to take them up on the offer.

So I guess 10 years of waiting paid out after all :)

 

Thanks for all your replies and explanations!

[jtk]

The 4000 has focus issues that were solved with 5000 and V.

 

V with Vuescan isn't good for multipass in my experience..if multipass is crucial due to bad (underexposed) slide film, 5000 is the answer...though the examples I've seen have not been worth the effort.

 

Envisioning print by holding film up to a bad light source (intense and not diffused), is futile practice as the intense/odd lighting can't be replicated in print, unlike lightbox vision of the film (assuming print is the goal).

 

The traditional (wet darkroom) workaround was duplication of a difficult slide using proper Ektachrome dupe film or camera film with pin-registered silver mask. Takes more skill than the original underexposing photographer had, by definition.

[rich815]

This is the first I'd heard of the 4000 having focusing issues. Have not had any with mine, but could be that I use Vuescan 99% of the time instead of NikonScan? Do you mean the corner focus issue with the narrow DOF if the neg is warped or not flat? Reports I see say the 5000 still has that particular issue.

 

John, could you point out where this issue with the 4000 has been discussed. I googled and could not find anything.

[jtk]

Rich, I think my recollection about 4000 came from deep within discussion of other topics, probably 2006...generic title like film flatness, autofocus, whatever. The V has plenty of depth-of-focus for mounted slides and the motorized carrier does a great job of flattening warped strip film, but it's not able to deal with significant curl at end of strips if they're not supported in the accessory strip film holder. I also recall discussion of custom antinewton glass carriers for both my V (using modified strip film carrier) and 5000, so probably 4000. Might be worthwhile to browse using "antinewton"