|
|
Mountain Art and Digital Technology
|
|
In 2005, Gordon Stainforth and myself applied ‘digital restoration’ to the images compiled by him for The Crux, a photographic exhibition for Kendal Film Festival. This presents 25 classic photographs from British mountaineering and rock climbing from 1880 onwards. A principle of digital restoration is to maintain or even uncover image integrity, meaning to remove artefacts that are added at each image stage as a result of whatever processes it goes through. We rarely had original material to work from, and were dealing with the grain structure produced by the original film or plate chemical processing, noise from the original film scan or print, noise from a scan of that print, and in some cases the screen dot pattern in scans from a book reproduction added to any characteristics of that scanning process. Sometimes a duplicate transparency added its own emulsion characteristics to the original film. Most of the work consisted of grain/noise reduction with a complicated sharpening process that incorporated some image processing. Although not used for The Crux exhibition, digital restoration can also include Fourier processing. This is the image processing version of the hi-fi filter that removes ‘crackle’ from audio recordings. The work enabled us to clarify a third figure and section of rope in the Abrahams brothers’ 1915 Scafell Central Buttress picture, that wasn’t clear in the scan we had to work from (Fig.3). The most fascinating and difficult picture by far for me was T. Howard Somervell’s picture of Edward Norton at 28,100ft on Everest in 1924, and I did slightly lighten the area surrounding the figure to help distinguish it from the background. The area round the ice axe remains particularly puzzling (Fig.4). It helps to be a mountaineer in restoring mountaineering pictures. There comes a point working close in where it is difficult to tell if certain image detail is actual rock detail against the snow, debris left from film processing chemistry, or system noise looking similar to film grain. Rock against snow has a characteristic feel to how it looks; the bottom line is - if unsure, leave it in. Digital systems create colour images based in a different science to that of photography, which uses Densitometry (the measurement of density or implied brightness), where essentially colour is not separated out from brightness in the measurements. Digital colour is based in Colorimetry, which uses the spectral composition of colour, and transformations between different colour spaces that do allow us to separate brightness from colour in the same way that the eye does. Colour science isn’t new, but digital colour is, and when it was realised that their combination could provide significantly improved digital preservation of the visual information in art and other heritage collections and archives, there was an explosion in the funding made available for research. Art reproduction has become a hot topic in colour science labs worldwide, and perhaps leading the research is the Rochester Institute of Technology’s Munsell Color Science Laboratory (MCSL). Together with the Image Permanence Institute, MCSL recently completed a survey in benchmarking the digital capture of artwork in 52 American museums(7). The study is remarkable both in defining the problem (current practice being based in photographic science, with different institutions taking their own approach), and the answer (use of colour science methodology with an art historian’s input). In spite of considerable staff dedication, there are many inconsistencies in imaging practice, and a lack of understanding of colour science principles. It also gives a recommended best-practice methodology for institutions that are unable to provide scientific skills or facilities. Since the background work culture of these scientists is that of ISO committees in photography and imaging, it seems reasonable to predict an eventual ISO standard for the digital preservation of heritage. Funding bodies could well begin to favour heritage digitisation projects from institutions that show they are following the practical guidelines, and are aware of the scientific lead being given. It may also be possible for the applicant to educate the funder! A major problem in digitising photographic collections can be the scale of such projects, especially if planned at print resolutions. The Paul Nunn Archive for example, recently donated by his family to The Fell and Rock Club, contains over 17,000 images. Conventional individual scanning at 15mins/image for 25 hrs/week works out at 3½ years of work for one operator. I would propose that such projects are scaled down into stages. Large collections can be batch processed relatively cheaply direct to low resolution, but scientifically more accurate, ‘raw’ files. This gives a good quality, repurposable, thumbnail image collection that enables cataloguing of the content, perhaps as a basis for selection of the important material for a high resolution digitisation funding application. It also enables the selection of individual images for higher resolution scanning for specific purposes. Above all, it makes archives accessible without the cost of a large scale digitisation project, rather than them being hidden away. The new technology has other advantages for institutions concerned with heritage preservation. It gives electronic accessibility via internet websites where art artefacts are fragile. A state-of-the-art example is the Vatican Library Project(8), which was able to make previously unseen art available worldwide. Content management system websites now enable easily-managed picture galleries that can hold thousands of images. Also, superb quality short-run art reproductions are available via digital inkjet printing. Because the process is not designed for mass print runs, much better quality inks, pigments, art papers and paper receiver coatings can be used, and a properly produced fine art print can now have a longevity rating (period before discernible fade) in excess of 200 years. References (1) Berns, R. S. (2001). The science of digitizing paintings for color-accurate image archives: A review. J. Imaging Sci. Technol. 45 Preserving our Mountain Art, and Digital Technology: A new imaging landscape. By Tony Riley, March 2007 (First published by The Alpine Journal in their 150th Anniversary Issue, 2007)
Mountaineering and photography seem to have strangely aligned histories, maybe by developing over a similar period. Certainly high light levels in mountain regions made them a natural environment for getting the best from the first very slow film emulsions, and the very large dark chambers of the early plate cameras were ideal for reducing the ‘flare’ that degrades image contrast and detail. The unending supply of great pictures in climbing publications is a tribute both to the popularity of photography with climbers, and their documentary and artistic skills. A love of mountains finds natural expression in photography, and few climbing teams are without a camera. Each advance in camera technology has been rapidly assimilated by climbers, through early rollfilm formats to 35mm then the compact cameras. They have been slower to change to digital cameras over the fifteen years that consumer models have been available, especially given that digital cameras have been outselling film cameras for the last few years. There are probably good reasons for the slower uptake of digital cameras in mountaineering, mainly loss of battery power in extreme cold. But digital cameras have advantages in cold conditions, namely fewer moving parts to freeze, far less detail-obscuring image ‘grain’ than film, and more accurate colour reproduction. For the professional cameraman/photographer, film equipment can be degreased and low-viscosity lubricants used to combat freezing. I well remember burying a movie film camera in a snow slope at 25,000ft, rather than carry it back down only to have to haul it up again. I was surprised to find it working a week later. But improved image quality in digital images over film? This is a whole other debate, outside the scope of this article, but I refer to the reduction at low temperatures of the ‘grainy’ texture that is actually electrical noise in the image signal. It is easily seen in digital images in areas of even tone like skies. Scanners tend to produce even more signal noise than cameras. Fig.2 shows a close-up from a 3200 ISO exposure, which exaggerates the effect. Low sensor noise gives a much better signal to noise ratio. Low temperature improvement is so pronounced an effect that some scientific digital cameras and professional scanners incorporate cooling systems to reduce noise in the charge coupled device (CCD) that initiates the image capture sequence. We accept grainy texture in photographic film images, but the better digital camera images get much closer to reality by reducing it, since it can obscure fine detail. Digital capture also enables more ‘accurate’ colour information after data processing than photographic film dye. How technically ‘accurate’ is an image to a scene anyway? Much film used on expeditions has been through successive cycles of extremes of temperature, and this can easily cause both selective and overall colour casts in the colour layers that are designed to work in parallel. Some films even deliberately distort colour rendition for aesthetic reasons. Film dyes can only produce a limited range of colour compared to the response of a silicon CCD, given the colour gamut of its pixel filters and subsequent processing. An impeccable scientific experiment(1) in comparative colour accuracy between film and digital sensors, when recording real scenes, placed a custom built IBM system top and a consumer digital SLR second, as having the least colour error. Digital sensors also record a wider range of subject brightness (dynamic range), and produce much lower noisy/grainy images in low light levels, than film. Indeed one method of ISO speed rating for digital cameras is based on the acceptability level of noise in the image(2). Fortunately, most photography doesn’t have to produce accurate colour, merely a pleasing picture(3), since pictorial images are rarely compared with the original scene, and perceptual processes can be shown to distort colour memory(4). The great mountain photographer Ansel Adams recognised very early in his work both the way film distorted tonal distribution and how the picture in his mind differed, compared to the way the actual measured light in the scene was distributed tonally. He sought a way of recording an image that would allow him to print, not the image he ‘saw’ in front of him, but that image in his mind, “previsualisation” as he called it. Working predominantly in monochrome to achieve this, he co-invented the Zone System, which reinterprets film exposure in the light of perceptual behaviour. How would Ansel Adams have worked today? Probably like photographer Joseph Holmes(5), who straddles the art/science divide in much the same way Ansel did, using colour science and colour management to solve the same problem the Zone System did, but with more control. Colour science is based on defining colour in terms of it’s spectral content and plotting it in three-dimensional ‘spaces’. In doing so, it takes account both of the spectral sensitivity of the cones in the retina, and the subsequent reworking of this retinal response into a different signal set before the information ever reaches the brain. Joseph Holmes works with very high quality scans of large format film, and has designed the colour spaces used by many discerning photographers. Colour management refers to the practice of measuring or profling the colour characteristics of individual cameras, monitors, scanners and printers with a spectrophotometer, and recording any differences in a digital file that can adjust colour appropriately in image production workflows. Many professional photographers have now added a spectrophotometer to their equipment, using it to calibrate, profile and monitor their colour workflows. Colour ‘accuracy’ is however crucial when we consider the digital preservation of our heritage collections and art reproduction. Whereas archivists are rightly concerned with the preservation of the original artwork or artefact, even now creating the UK’s first low oxygen storage facility at the British library, ‘Digital Preservation’ is concerned both with capturing scientifically accurate image information, and maintaining its portability to future technology. Cameras and scanners sense ‘raw’ image information, but introduce proprietory changes in processing this data, so it is important to record the raw data prior to the changes. The individual characteristics of the capture equipment, including lighting, should also be profiled for image re-evaluation in future technological systems. How ‘accurate’ can you get in digitally preserving heritage, and how is it quantified? Colour can be compared to its reproduction, giving units of colour difference on a scale called, among others, deltaE(6). Up to 2 units of difference are considered imperceptible to human colour vision. From 2 to 5/6 is considered acceptable, and more than this increasingly unacceptable to the average observer. In the colour science lab, accuracies of less than 1 unit have been achieved.
Tony Riley qualified in Imaging Science with a dissertation in the measurement of colour difference in art reproduction (9), and he has a special interest in the digital preservation of heritage. Having spent most of his life as a mountaineering cameraman, professional photographer and lecturer, he has enjoyed photographing mountain areas for over half a century, and now runs an art gallery in the English Lake District, also providing scanning and digital printing services through his online picture library. Feedback and discussion are welcome - contact This email address is being protected from spam bots, you need Javascript enabled to view it . |
