This is a special guest post by Elizabeth Mitchell, who was part of the Factum Foundation team which digitised the Cartoons.
The Raphael Court is currently undergoing a nine-month refurbishment to mark the 500th anniversary of Raphael’s death. The refreshed gallery and its new interpretive approach will transform the way museum visitors experience and understand the Cartoons. The Cartoons are lent to the V&A from the Royal Collection by Her Majesty The Queen.
As part of this project, the V&A and Royal Collection Trust entrusted Factum Foundation and Momart to carry out a high-resolution recording of the seven Cartoons, supported by the Royal Commission for the Exhibition of 1851. In this post, Factum Foundation reveals the challenging process of undertaking this unique digitisation project.
The recording of the seven Raphael Cartoons in colour, 3D and infrared is one of the largest-scale art digitisation projects ever undertaken. Over a period of five weeks in August 2019, a team from Factum Foundation for Digital Technology in Conservation worked around the clock to digitise the 115 sqm of the Cartoons’ surface, recording at a resolution of up to 10,000 points/cm2. It was a project which required meticulous planning and close coordination with the teams from the V&A, Royal Collection Trust and Momart: before any recording could take place on a Cartoon, Momart technicians had to remove the huge frames and protective glass from in front of it – a task that required tight coordination and handling skills. On any one day, up to four 3D scanners were to be found at work in the galleries, poised on scaffolding three metres above the ground, with colour photography of the Cartoons taking place by night.
But scanning and photographing the Cartoons was only the beginning, and was followed by extensive processing work in Factum Foundation’s workshops in Madrid, as the team started merging, stitching, and modifying the recordings into a form in which they can be viewed and studied.
Processing colour
Colour is surprisingly difficult to record: if you try taking a photo of a painting and then hold it up alongside the original, you’ll probably find that even though the contours are well reproduced, the colour will appear completely different. Similarly, if you look up a famous artwork online, you’ll find it reproduced in a rainbow of shades.
Factum Foundation’s photographers used a process called panoramic composite photography to record the colour of the Cartoons. For each cartoon, they took thousands of photos at regular intervals across the surface – ensuring that every inch was captured in focus. Some of these photos include a colour checker, which allows the recorded colour to be adjusted in relation to standard RGB values.
Back in Madrid, these photos were then digitally stitched together to form a single, ultra-high-resolution image for each Cartoon. The initial composite image is distorted because all of the photos are taken from a single central point – something which has to be corrected in post-processing. Colour processing is a slow business: the original photos overlap by 60% or 70% with the photos on either side of them, and the task of selecting the most in-focus image for each part of the painting can only partially be performed by computers – it also requires substantial human input. It takes about 300 hours to complete the colour for each Cartoon.
This panorama has been corrected for distortion. Each red outline marks a separate photograph (in fact, three separate photographs); where the outlines are relatively straight, computer software has selected the most in-focus parts of each image; where they are more uneven, the in-focus parts have been chosen by hand.
Processing surface relief
Recording surface relief involves another process altogether. When you look up close the Cartoons resolve into a complex and highly textured landscape, each made up of nearly two hundred sheets of paper pasted together; these composite backing grounds were later mounted on canvas in the 1690s. The surfaces were recorded using the Lucida 3D Scanner, a close-range, non-contact laser scanner which captures 3D surface data for low relief surfaces (designed by artist and engineer Manuel Franquelo with the support of Factum Arte and Factum Foundation). The scanners recorded the surface of the Cartoons in ’tiles’ measuring 48 x 48cm. The larger Cartoons, such as The Death of Ananias, needed up to 117 tiles and took about 150 hours to record. To record the upper parts of the Cartoons, some scanners were raised onto scaffolds, and up to three scanners at a time were sometimes used on a single Cartoon to speed up the recording process.
In post-processing, the hundreds of ’tiles’ were again digitally stitched together to create a single continuous image of the surface of each Cartoon. This image is known as a depth-map – a representation on a 2D plane of 3D data, using gradations of tone to represent depth. The resulting 3D relief data has many uses: it can be merged with the colour recording to create an integrated screen image showing colour and 3D together, it can be printed out as the basis for a facsimile (a replica of the artwork which is indistinguishable from the original at ordinary museum-viewing distance), or it can be viewed by itself, giving conservators and others a clearer understanding of the material composition of the cartoons without the distraction of colour.
Infrared photography
Finally, the Cartoons were recorded using infrared photography. The low wavelengths of infrared light partly pass through the paint layers of the Cartoons, but are absorbed by carbon, revealing the preparatory drawings – also called underdrawings – made using black chalk or charcoal.
Viewing the data
Displaying vast composite images like these is a challenge in its own right: even opening one of these panoramas will cause most computers to crash. Comparing multiple different image sets (which can weigh up to 140GB) is even harder. To deal with this problem, Factum has developed a browser which only loads the specific part of an image which a viewer is zoomed in on at any one time, allowing a smooth viewing experience for vast quantities of data.
Recording and processing vast surfaces like this is a long, slow process, involving months of work by the teams involved, but it also has its own distinct magic. Watching the colourless surface of a painting re-materialised line by line on a computer screen, complete with every crack and crease, or seeing a Cartoon for the first time without its protective glass, make the colours stand out with extraordinary clarity and transforms the way in which you see these paintings. The results – which will be available for visitors to explore both in the gallery and online when The Raphael Court reopens in November 2020 – will allow everyone to explore the surfaces of the Cartoons inch by inch, as if with a spotlight and a magnifying glass.
The vast amount of data from the recording of the Raphael Cartoons gives rise to practical questions that go beyond the complexity of the scanning itself. How can a recording project of this scale transform the way the data is processed, stored, disseminated and displayed? To what extent can modern technology bring to life previously unknown qualities of an object? Factum software engineers are continually working to improve the automated processing of data: to speed up processes, improve alignment, and understand and exploit the elasticity of the cloud in which the data is stored and through which it is processed. The recording, processing and outputting of data from the Raphael Cartoons project has proved an important spur to the development of such processes, opening the field for new reflection on the role of AI in the preservation of cultural heritage.
