Bob,

I think I mentioned this in an earlier post, but you CANNOT use ICE with B&W. It just doesn't work. It will create really nasty artifacts. It also does not work very well with Kodachrome.

If you use the software that came with your scanner (mine is the Nikon Coolscan 4000), you will find that any negative or slide which is really dense (i.e., you can't get much light through it), will have all sorts of problems. This is especially true with slides that are designed to be projected in front of a 500W bulb. The problem is that the little tiny lights in your scanner emit such a weak light that when they go through the dense sections of a slide or negative, there just isn't much signal left to read on the other side.

Make sure you don't place your scanner next to a sunlit window. The ambient light can screw things up. Also, use Vuescan as your scanning software. It can make a huge difference, and it has all sorts of extra features specifically designed for scanning dense slides and negatives, including the ability to scan the same frame multiple times and then average the results. It also lets you control more of the scanning parameters which can make a big difference in the color you get from those really dense color materials.

Even with Vuescan, and even having scanned over 50,000 slides and negatives in the past few years, I still find that I must spend many minutes scanning the troublesome slides and negatives. There just isn't any one single thing that gives you good results for each situation.

John,
I've got the 5000 ED scanner, not really certain what they've 'improved' over the 4000.
Certainly the film density is a big problem, struck the issue way more so with 8mm film in the Elmo, the CCD was way past its use by date. I improved matters then by upping the lamp power and adding fan cooling.

But the LED lamp in the 5000 is pretty damn bright. I can get image data out of slides that I can barely see even held in front of a 100 watt lamp but some of them have really odd stuff in the denser areas of the slide and I don't think it's the scanner is adding it's own footprint to the image, some of them have this quite organic color shift in the black areas, it sure looks chemical rather than digital in nature.
Running 16 passes of the scanner to reduce CCD noise makes not a shred of difference, dialing in more analog gain at times helps a lot but that wierd stuff is still there in the blacks.
Having re read the manual I have better results on some of the older faded color slides by turning ROC off and GEM up, got rid of the weird red artifacts.
You're dead right about ICE and B&W slides, learned that the hard way!
Silverfast seems to be the best of the best software wise but very expensive, my concern is just where it's aimed at, is it better at adding the 14th coat of wax to already great slides or at pulling more useful images out of horrid slides taken with cheap cameras of poorly lit scenes.
And at the end of the day as I'm more than certain you've discovered all this experimenting take an aweful lot of time, I'm way past the point of even thinking about how much I'm loosing on this job.Running 5000 slides through the scanner with the auto feeder it was marginally OK, spending several days manually tweaking 200 of them blew any chance of a viable return, so now I'm just writing that off to 'training' but other jobs start to bank up.

I guess it's the same as any of the other work we get sucked into in this line of work. The clients who need the most effort on their jobs don't have the money for it. If they did they'd have used better cameras, paid for decent film and had it properly processed in the first place. I'd add, bought a flash gun to that list as well, no one in the family ever owned one!

Bob.

I can't speak about Silverfast because I haven't used it. However, I can tell you that it is virtually impossible to get good results with the Nikonscan software when scanning dense material, especially Kodachromes that are slightly underexposed. However, with Vuescan, I can get usable results, but it took a lot of tweaking to get there, and I have found that the newest versions of Vuescan don't work as well as the older versions.

You are absolutely correct about the color shifts in the dark areas. Until I figured out the "recipe" for this, I was consistently getting some really horrible blue-green shifts in the shadows of Kodachromes.

I downloaded Vuescan to give it a whirl and so far the results from it are worse than what Nikon Scan can manage. With a lot of manual tweaking I can get close but not as good. It's got some quite useful tools which make it easier to highlight what's wrong but its processing capabilities don't seem as capable as Nikons own software with the 5000 ED.
Mind you the Nikon software uses up a lot of CPU doing its thing compared to Vuescan, whether that's anything usefull or just a lot of badly written code might be another question.

I've taken a hard look into some of these slides and well I think most of the problems are just on the slides. The blacks are deep red, the grain is visible with nothing more than a magnifying glass and the depth of the density is very shallow apart from the bits where there's no emulsion. On some of them I can get a watchable image doing a fair bit of work in PS, masking areas so I can bump up the levels and reduce grain while leaving the blacks alone.

Bob.

I don't have a densitometer so I did a crude measurement with a Sekonic light meter, difference between no slide and slide in a light path goes from f22 to f1.4. Somehow I think I'm pushing it uphill trying yo get any better than what I'm getting now.

Vuescan's implementation of GEM is not as good as Nikon's. The ICE implementation, however, is generally better. ROC is totally different than Nikon's and sometimes better and sometimes worse.

It is the exposure controls that are generally so superior. However, you have to know how to use the settings. Here are the key settings for scanning Kodachrome slides (you have to set each tab to "Advanced" to be able to see all these settings).

1. On the Input tab, set Media to Image, NOT to slide. I spent a lot of time talking to the author, Ed Hamrick about this. Long story, but this setting works better. This only applies to slides, not to negatives (for which you SHOULD use the "negative" setting).

2. You can sometimes get better results by playing with the RGB analog gain controls. This is on the input tab. I generally scan a well-exposed neutral color slide (like a landscape that doesn't have lots of strong primary colors) and then look at the histogram. If one color predominates, I set the the RGB gain controls to compensate. Currently, my starting point is Red Analog Gain-2.0, Green Analog Gain-1.5, Blue Analog Gain-1.0. I'd recommend NOT messing with this control until you've tried ALL the other things. This is a tweak to be done only at the end of the process, and only when you've almost achieved your goals.

3. On the Filter tab, I use Infrared Clean set to "Light." One neat thing about Vuescan is that it saves the RAW scan and you can output several versions of the same scan without having to re-scan the slide. This is very useful for Kodachrome, which can't be cleaned at all using ICE in the Nikonscan software, but can somewhat be cleaned using this light setting. However, sometimes even the light setting will introduce artifacts, most notably around severe light to dark transitions. The solution is to create one file using the clean setting set to Light. Then, turn off the clean setting entirely, and then simply click on the "Save" button. This will apply the different processing to exactly the same scan. I then line up the two scans in PhotoImpact (you probably use PhotoShop) in two different layers. Every place I see an artifact, I just erase the artifact area and let the scan without the cleaning (which is on the bottom layer) show through. This usually takes less than a minute, because you don't have to be very precise with the erasing, since the lower picture is pixel for pixel the same, except for the dust that has been removed or the artifacts that have been introduced.

I don't usually use any of the other options under Filter.

4. The Color tab is key. I generally use either Auto Levels or White Balance. If I didnt choose "slide" back on the Input tab, but instead chose "Image," then there are far fewer controls available under this tab. I set Auto Levels, and then set both black and white point to 0.02. This gives me every bit of dynamic range, although the final image may lack punch and contrast. I prefer to fix this in my photo editing program because once you lose the highlight or shadow, it's gone forever. I sometimes further tweak the balance. For my Kodachrome settings (with my Nikon 4000), I use Brightness 1.0, Brightness Red 1.06, Brightness Green 1.0, and Brightness Blue 0.91. View Color RGB.

You'll find hundreds of posts all over the Internet about the problem of Nikon scanners creating slides with terrible color shifts, especially those from Kodachrome and other 'chrome films. After reading a lot of those posts, this is my solution.

Out of all of this, the main thing to try if you want to spend another few minutes with VueScan, is to change the input setting from Slide to Image. Very counter-intuitive, but it is what the software author himself recommends -- you'll even see it in some of his own documentation, and you'll certainly see it in all of his posts in various photo forums.

John,
thanks for all that.
I was indeed using the 'slide' setting!
Will go back and try with 'image'.

With both Vuescan and the Nikon software I've been using the analogue gain controls, can be very useful. Also shadow and highlight controls although the Shadow enhancement can do wonders but I find you have to get just the right setting, there seems to be some form of 'knee', possibly in the film and if you get the setting to just that point I can get better detail in the blacks, a point or two either way makes a huge difference.
The other approach I've used is to turn everything OFF in the Nikon software and save as 16bit targa files and then work on them in PS, of course the file sizes are huge!

The really interesting part though is as I said that many of the problems are in the slides themselves. It seems many of these slides were developed by the client, some of them were taken on a 1/2 35mm format camera which he then enlarged and printed to full frame 35mm. Just how and using what emulsion is lost in the mists of time, none of the film that I've checked has the usual identification numbers etc on the edge. The is somewhere out there a web page that gives all the Kodak manufacturing details from the batch numbers on the film which might be usefull if there were any numbers!

Bob.