Illustration

HP-Illustration-lThe study of 15th-century illustration is ill served by digital tools currently available. The Incunabula Short Title Catalogue (= ISTC) does not include systematic information on whether an edition contains illustration; the Gesamtkatalog der Wiegendrucke (= GW) does, but details are not searchable with the same degree of reliability. A few large publications were compiled at the turn of the 20th century, with an art historical angle and city-centred approach (Paul Kristeller, Early Florentine woodcuts, London, 1897; Wilhelm Ludwig Schreiber, Manuel de l’amateur de la gravure sur bois et sur métal au XVe siècle, 5 vols in 2 parts, Berlin, 1891-1911; Victor Masséna, Prince d’Essling, Les livres à figures vénitiens de la fin du XVe siècle et du commencement di XVIe, 4 vols, Florence and Paris, 1907-14; Albert Schramm, Der Bilderschmuck der Frühdrucke, 23 vols, Leipzig, 1920-43; Max Sander, Le livre à figures italiens depuis 1467 jusq’à 1530: Essai de sa bibliographie et de son histoire, 6 vols, Milan, 1942; Carlo Enrico Rava, Supplément, Milan, 1969, etc.).

There is instead the need to track the re-use, exchange, and copying of woodcuts to better understand the working practices of printers and to highlight otherwise undocumented links among them; also, to expand the investigation beyond national boundaries to detect the international borrowing and copying that did take place, for example Parisian features introduced in Venetian illustration of books of hours in the 1480s. And of course there is the need to easily retrieve in which other edition a certain woodcut has been appearing.

 

Methodology

In 2011-12 Clementina Piazza, a University of Oxford MPhil candidate, conducted an in-depth survey of all current European projects relating to fifteenth-century and later illustration. In particular, she browsed through a series of tools that thanks to the advancement reached by the new technologies could substantially facilitate and refine this kind of research, shedding more light on the circulation and reutilization of woodcuts. After having examined different systems of cataloguing images, she ended up focusing on major changes in terms of accessibility and research on images which came from Oxford.

Professor Andrew Zisserman, of the Visual Geometry Group, part of the Engineering Science Department of the University of Oxford, and Dr Giles Bergel, of the English Faculty, recently developed an open source sofware for the ‘Ballads Project online’ at the Bodleian Library. It applies image-matching software to enable an automated recognition of similarities. It has an upload functionality to build up corpora of material to be matched. The only major complication is the production of metadata. The Ballads projects could use the detailed cataloguing of the Ballads’ collection previously prepared by Dr Alexandra Franklin. This work still needs to be done for 15th-century illustration.

For her pilot project on Johannes Hamman’s illustrated editions of Horae: ad usum Romanum, that is books of hours, Piazza elaborated an effective method to survey how each woodcut appears in every editions, meaning not just the mention of their measurements and general description but their recurrence for each page and throughout the gatherings of the edition.

Piazza could thus retrieve the exact number of blocks that the printer had used and their lay-out and recurrence throughout the text. A key element in determining the number of blocks that may have been used by a printer was the collational formula, which was also the basis for designing the dedicated metadata spreadsheet. Piazza’s work successfully set the basis for further research in this field. Therefore, the 15cBOOKTRADE Project includes the position of a DPhil student, Matilde Malaspina, who is in charge of a systematic census of illustrated Venetian editions as well as of capturing images and producing metadata for many of the woodcuts contained in them.

 

Developments

Recently, the application of an image-retrieval system to 15th-century printed woodcut illustrations has been devised within the 15cBOOKTRADE in collaboration with the Visual Geometry Group, Faculty of Engineering Science, University of Oxford.

image retrieval screenshot

Images are uploaded in the system and catalogued with a unique identifier, consisting in:

ISTC n°_MEI n°_foliation

From this very simple sequence of numbers, it becomes immediately clear in which edition the image is, in which copy and where exactly in the copy.

 

This image-matching system follows the same idea of text-retrieval, which allows users to retrieve documents containing query words. Its goal is to match and align a source image to a dataset.

There are two stages:

1.Matching the source image to one or more of 1000s of images in the dataset

2.Aligning the matched images (for comparison)

screenshot matching regions

When applied to 15th-century printed book illustrations, this system allows the user to know fast and precisely the number and locations of:

– Reoccurrences of the same cut in different editions;

– Reoccurrence of the same cut within one single edition.

This contributes to highlight or to explore further cases of exchange or copy of the printing material between different printers, the evolution of the printing practice of one single printer from one edition to the other, as well as  the working practice, as far as woodblock material is concerned, of one single printer.

 

This system for image-retrieval is just the first step towards the creation of a more articulate tool for cataloguing and researching the use and circulation of 15th-century printed woodcuts, which will include both instance-level search (based on image-matching) and category-level search (based on keywords).

 

This second level has been developed over the past months through the use of a software for digital annotation of images, recently supplied by the Visual Geometry Group. This tool allows to create and assign specific metadata for research to each separate cut composing an illustration with no need to physically fragment the image.

ImageAnnotation4Thanks to a combined application of the two tools, the data gathered by the 15cBOOKTRADE will be used as a pilot for the creation and application of a new and cutting-edge tool for detecting the presence and use of early printed images in books, which is the ultimate objective of the project.

 

The collaboration between the 15cBOOKTRADE and the Visual Geometry Group is supported by the Engineering and Physical Sciences Research Council’s grant EPSRC Programme Grant Seebibyte EP/M013774/1.