Thursday, 21 March 2013

Scanning 35mm Slides...If Only They Were All Kodachromes

As well as the 4 x 4 cm Superslides I have written about previously, I have scanned thousands of 35 mm full frame (24 x 36) and half-frame (18 x 24) transparencies, dating from 1956 to 2002.

Whenever I read an account of slide scanning, the writer has concluded that Kodachromes had survived the passage of time in better condition than any other transparencies. I can only agree. But many of us were tempted by the home processing of Ferraniacolor or the marketing that promoted Agfacolor CT18, for example.

My Ferraniacolors have actually survived well. The colour is maintained including the yellowish tinge they had originally. They now look better scanned with their colour corrected than they ever did projected when newly processed.

Two films stood out as poor. The first, the old film of the two, is Agfacolor CT18. A few slides have turned a purplish shade over the years but the main problem is ‘grain’. I have put ‘grain’ in inverted commas because the term seems to cover a multitude of sins. Scanning the films, I noticed a very great difference between films. In other words, the degree of ‘graininess’ was similar in slides of a similar subject taken on one film but different on a different film. Thus, ‘grain’ varied from acceptable to dreadful. Such a difference clearly indicates a difference in manufacture or in processing.

The effect was particularly noticeable in half-frame transparencies because the size of the grain was greater in proportion to the image.

In an attempt to get better scans, I searched for more information. What is the ‘grain’ and are there hardware/software solutions to getting better scans?

In colour film, ‘dye clouds’ rather than ‘grain’ is a better term for the final result of film development. Is the scanner, though, simply picking up what is there or is what is there being exaggerated by the optical scanning process? The latter is known as ‘grain aliasing’ and there has been some discussion of this phenomenon (see below for links to websites). I do not think the problem with the Agfacolor CT18 slides at the poor end of the apparent graininess spectrum lies with grain aliasing. There was no apparent change in structure or position at different scanning resolutions or in different scanners. There was no effect of changing the scan direction across the slide. Furthermore, the same structures could be seen in the transparency under a microscope. By now, I was convinced that the ‘grain’ had a physical presence that was not being increased by the scanning process.

I tried several standard grain-reducing bits of software including Nikon Scan (when Nikon supported the software for their scanners), VueScan and Epson Scan during the scanning process. There was a little improvement but not much. If the graininess apparent in the scans was really there, were there any software solutions to remove it. Because graininess is a form of noise in the image, I tried the standard noise reduction controls in Photoshop and Aperture with little effect.

Then I then came across an article by Norman Koren on Grain and Sharpness in Scans and Enlarger Prints (website address given below). He recommended Neat Image and showed some examples of grain reduction in scanned images. Within minutes, I had loaded the trial version, installed it as a plug-in in Aperture and tried it on a few scans. The result was that a few minutes afterwards I paid for it and had the software in operation. I can only describe the effect as amazing. Only the worst two films still show obvious graininess and they were really bad to start off with. Even with them there has been a significant improvement, the worst effects of graininess being apparent in areas of sky.

So far I have been dealing with Agfacolor CT18. The other poor but more modern film was a Fuji transparency film (not Velvia) I used in Botswana in 2001. That showed a similar problem. Again, there was considerable improvement using Neat Image.

With the Agfacolor CT18 scans, I have never been completely convinced that the problem was in the emulsion per se. The appearance I can best describe as a reticulation (not the old burst blister appearance of reticulated film but think of the pattern of a Reticulated Giraffe) could have been in the base layer. I even wondered if the films had been varnished after processing and rubbed one over with alcohol to see if any coating could be removed (there was no effect). Then I came across a 2002 article by Nick Rains on Fuji 'Pepper Grain' The Mystery Resolved (website address below). The problem, as Fuji apparently admitted, was in the film base, not the emulsion. The effect was not apparent with wet, drum scanning, only with dry. Since the effect could be removed with Digital ICE technology (mine could not) we clearly are not dealing with the same phenomenon but I don’t think we can eliminate the film base from consideration entirely with the 1960s CT18.

This following is a comparison of enlarged images of a scanned full-frame  35 mm transparency taken in 2001 on a Fuji transparency film before (A) and after treatment with Neat Image (B). You should be able to see the difference. Finally, the best I could get out of this slide and, really, you do not want to see the untreated version.

A Scanned 35 mm Fuji transparency before and B after application of Neat Image.
The full frame version of B is shown below

Returning to where I started, why did we not take everything on Kodachrome? Well, Kodachrome was more expensive and it was slower. In the mid-1950s and using film speeds published in the BJ Almanac, Kodachrome was only 10 ASA. To put that figure into perspective, the ‘standard’ ASA setting on my Nikon D700 is 200 or more than 4 stops faster than the old Kodachrome. Ferraniacolor was 1 stop faster at 20 ASA and Agfacolor CT18 (18° DIN) was 50 ASA — more than 2 stops faster than Kodachrome. The next version of Kodachrome at 25 ASA brought that difference down to 1 stop.

For half-frame films, processing by Kodak was a hassle, whereas other processors, like Agfa, offered mounting in card or plastic after processing. My wife with her Olympus Pen FT avoided Kodak for that reason.

I think I have now finished with trying to get the best possible scans from my old slides. There will be one more article in this series but that’s it for 35 mm and 127 film Superslides. Somebody asked me the other day whether I would like to go back to colour film from digital capture. The answer is: NO! Only the luddites (and the real luddites at least had the excuse that their livelihoods were at stake) don’t realise that we have never had it so good. That’s my view on colour; to monochrome I shall return another day.

Website Addresses

Norman Koren Website (follow links there to earlier studies on grain aliasing): 

Fuji ‘Pepper Grain’ on The Luminous Landscape

Neat Image Website†

†I had an unexpected benefit from using this software which I will cover in a future post.