A guide to creating digital scans from 35mm film with a Nikon Coolscan 4000 ED

I started writing this post in March 2006, so why did it take more than three years to complete? Well, it was partly a lack of time – scanning is a time-consuming process and it’s only in the last few weeks that I think I’ve finally cracked the image-scanning process. In the meantime I’ve considered outsourcing the job of scanning my old negatives and slides but just can’t bring myself to use an overseas service for something so precious (anyhow, the service I’ve heard good things about – Scan Café – is not available in the UK). Anyway, here goes, my long-overdue guide to creating digital scans from 35mm film.

Nikon Super Coolscan 4000 EDA few years ago, I was advised that, rather than switch to digital photography, I should switch to transparency film and buy a decent scanner. I did exactly that, but it took me some time to save up for the scanner, by which time I had a mountain of film to scan. A year or so later, 6 megapixel digital SLRs had become affordable and I switched to digital capture, keeping my film body as a backup although it’s hardly seen the light of day because the digital format provides me with so much freedom.

I’d still like to scan those slides (especially as many of them are from my honeymoon) but it’s a time-consuming process and, up until now, the odd frame that I have scanned has left me unimpressed with the quality. Even though 4000PPI is a lot to ask of any device, I couldn’t believe that a Nikon Super Coolscan 4000 ED (which cost me just short of £1000 at the time of purchase) would offer poor quality scans and after contacting the Nikon European Customer Support desk I found that it the scanner is actually excellent – it’s just that the user (i.e. me) didn’t really know what he was doing!

After Nikon had set me in the right direction, I googled a bit, read the NikonScan user manual, and learned lots. This post is a quick summary of how to get the best from a (Nikon) film scanner with Nikon’s NikonScan 4 software. Most of this information is lifted from the user manual but a short-ish article on the web should be quicker to read than a 150 page PDF.

In common with many other film scanners, the 4000 ED offers Digital ICE3 technology. Actually, this is three technologies from Applied Science Fiction that will increase scanning times significantly; however they will also affect the quality of the resulting images:

  • Image correction and enhancement (ICE) is about removing dust and scratches – even clean negatives tend to show up dust when scanned at such high resolution, so I always use ICE on its fine setting (even though this has a slight effect on the overall sharpness of the image).
  • Restoration of Colour (ROC) is used to restore color on old/faded images (I may try it on the foll of film I dropped in icy water whilst helihiking on a glacier!).
  • Grain Equalisation and Management (GEM) is used to remove the signs of grain on an image.

For most of the applications that I use, the default settings are fine and my initial scans used the defaults. That was the mistake I made with my image scanning and is the reason I’m writing this post! From my support call with Nikon, I learned that I don’t necessarily need to use all of the ICE3 technologies together – in fact, it was the use of ROC and GEM at their default levels (5 and 3 respectively) that was causing my scans to look so bad. I also chose to turn off the Unsharp Mask (it can always be applied later with a pixel editing application such as Adobe Photoshop if required) – similarly I would ignore changes to the curves, LCH and colour balance (although Nikon’s advice to reduce the blue channel by 0.5EV on my sample problem image made a huge difference).

Something else that’s worth noting is to zoom in to the required depth before generating a preview scan, as zooming after performing the preview will appear pixelated (and no zoom will do little to show the effects of the digital processing). It’s also worth ensuring that automatic exposure is selected within the software preferences. Note that the warning symbol is displayed if changes are made to the settings after the preview scan is performed.

Multisampling is another option that can increase scan times dramatically but which should also increase quality as the scanner makes multiple passes over the film. The idea is that it reduces electronic noise (the real values should average out, whereas the noise will be more random), rendering a more accurate image with smoother tonal changes. Personally, I’ve had great results using Noise Ninja as a Photoshop plugin instead – it’s faster and it works on my digital images too!

The boundary offset is a setting used with strip film to line up individual frames within the scanner..

The NikonScan application is able to scan images in either 8- or 14-bit colour although 14-bit images are actually converted to 16-bit for editing purposes. I wrote about this in a separate post but basically each channel is recorded separately for each pixel with 256 levels on an 8-bit scan and 16384 levels for a 14-bit scan (which actually requires 2 bytes for each pixel), so for an RGB (red, green, blue) scan:

  • 1 byte (i.e. 8 bits) x 3 channels (red, green and blue) x 5959 x 3946 (pixel count) = 70542642 bytes (67.3MB).
  • 2 bytes (i.e. 16 bits) x 3 channels (red, green and blue) x 5959 x 3946 (pixel count) = 141085284 bytes (134.5MB).

For CYMK (cyan, yellow, magenta, black), this would increase to 4 channels, so file sizes will be one third larger. Of course, these figures assume an uncompressed file, and does not take into account any overheads of the file type – Nikon Scan supports RAW (NEF) (read-only), JPEG (JFIF), JPEG (EXIF-compliant), TIFF, BMP (Windows) or PICT (Macintosh).

When considering the required resolution for a scan, it’s worth knowing that pixels per inch (PPI), is not equal to dots per inch (DPI). In fact, the scan quality will depend on the output device:

  • Commercial (dye sublimation) printers use continuous halftone, measured in lines per inch (LPI). For this, the artwork PPI needs to be set at twice the LPI of the output device.
  • Inkjet printers use something called simulated halftone – 240DPI should be adequate quality for most prints. A 4000PPI scan from a 35mm negative (actually 24x36mm) on my scanner is 5959 x 3946 pixels. Printing at 240DPI will allow print sizes of around 24.8″ x 16.4″.
  • Window PCs display monitor graphics at 96PPI; Macs display at 72PPI.

It’s also important to understand the way in which a computer represents the colour in an image. This is all controlled with colour profiles, which I still find a little confusing but, thankfully, NikonScan handles for me using the Nikon colour management system (CMS). Some points that are worth noting though:

  • Windows PCs use a gamma value of 2.2; Macs use 1.8.
  • Gamut is a term used to describe the range of colours that are displayed – a narrow gamut will be vivid with saturated colours, whereas a wide gamut may appear low contrast and flat.
  • Each monitor is generally provided with a monitor profile (or one can be created in software).
  • An RGB profile also needs to be set (e.g. Color Match RGB on a Mac, or Adobe RGB on Windows).
  • The CYMK profile can be left at the factory default setting if the output will be printed on a variety of printers, or a custom CYMK profile may be provided with a specific printer. Any other imaging applications used (e.g. Photoshop) should also be set to match the colour profile as it is not passed between applications.

Further reading

The following links provide additional information that may help to produce good scans:

Leave a Reply