About DIAMM

• Introduction
• General information
• Technical information
  • Image capture
  • Image quality
  • Archiving
  • Conservation
  • Metadata
  • Digital restoration
    • Introduction
    • Process
    • Samples
• Rights protection
• Project team
• News
• Sitemap
• Contact us

More information regarding techniques employed in virtual restoration may be found in the DIAMM publication: Image Restoration Workbook (see News: Publications)

Palimpsest (Oxford, Corpus Christi College MS 144)

Corpus manuscript 144 is unusual (though not unique) in containing leaves from a 14th-century music book that were scraped, refinished and re-used for writing another text in the 15th century (the Libermetricus de nova poetria of Geoffrey of Vinsauf). The music leaves were also trimmed, removing still more of the original musical text. First discovered in the 1970s, the music in this manuscript has until now been virtually illegible: it is clearly present, but a continuous composition could not be transcribed. The following pictures show a folio of Corpus MS 144 before and after recovery undertaken by DIAMM. Description of the process is given below in Process.

The images on this page have been provided for viewing only by kind permission of the President and Fellows of Corpus Christi College, Oxford: please do not download or reproduce them.

The images show a much reduced version of a digital image that was captured at high resolution using conventional daylight balanced lighting: the first image shows how the page looks to the naked eye without the aid of ultra-violet light. The underlying music is barely apparent, and that overlaid by text is unreadable. The second image is of the same page after digital restoration. In between is an animation that shows the gradual transformation of the document.

The first stage of enhancement involves the removal of the overlaid text. This is done in several stages, and the text is replaced in this case by a neutral background colour. Once this is done, the contrast level is slightly altered to darken the underlying music and lighten the background. The next stage was reached by lightening the background colour, heightening the contrast between the ink colour that should be retained and the colour of the parchment which can be discarded. However, because the ink has faded or is there is very little of it left, in places its colour is almost inseparable from the background.

Compared to an undamaged source this is still not very easy to read, however there is enough information now for someone familiar with the style of notation of the period to read every note on the page.

Dirt Removal (Stratford, Shakespeare Birthplace Trust, DR 37 Vol. 41, back cover)

The commonest form of ‘damage’ found in medieval fragments is the simple accretion of dirt. Since parchment was much tougher than paper it was often used as reinforcement for binding or to form binding or a wrapper itself. The Stratford source is a fascinating case study: a large vellum leaf from a music manuscript (or loose-leaf collection) was reused as a wrapper for paper documents. For many years it was believed that there was either no music written on this face of the leaf, or that if there was it was permanently lost. Inside the wrapper the leaf is remarkably clean and clear and the music is in extremely good condition.

The images below have been provided for viewing only, by kind permission of the Shakespeare Birthplace Trust, and may not be reproduced without permission of the Trust.

In this case a similar process was applied to the Corpus fragment, though here there was negligible overwriting to exclude. However the dirt accumulation on the surface was very dark, and apparently obscured any notes that may have appeared. On the top left corner there were some shapes that appeared to be notes when the image was magnified, and the colours of these artefacts were searched across the image and darkened, while areas that were clearly background were lightened. The result is far from beautiful but again is now fully transcribable, not just the notes, but the text too (not visible at this small size).

It is evident though from this source that even the most basic restoration techniques requires editorial decisions: the restorer had to decide which colours were notes, and which were background. On the original image a string of rising minims was visible in the middle of the top half of the page. On restoration these minims disappeared, and turned out to have been in the gap between two staves. Evidently a different decision on which elements to darken and which to lighten would have produced a quite different result: it is believed that this line of ascending notes must be an offset from another leaf, now lost, and a second restoration process must be undertaken to attempt to recover some of the lost offset.

Palimpsest repair and editorial process (Florence, San Lorenzo, MS 2211, f. 82v)

The main problem with the type of restoration shown above is that the appearance of the original source is ‘falsified’ by the restoration process. It is possible to come to the restored image and see the naturalistic colours which would imply this is the condition of the original source. In presenting digital restoration it is actually much more profitable, rather than darkening existing colours, to replace a ‘found’ colour with a completely unlikely colour, such as green, blue or purple. The effect is very dramatic: not only are the notes extremely clearly readable, but there is no question in the mind of the viewer that this is a restored image, as it cannot be natural. In addition, the areas of colour selected by the editor/restorer are clearly apparent, enabling later users to make some evaluation of the accuracy of the restorer’s work. It has been found that this type of restoration, using bright colours, yields more readable results than other processes with a more ‘natural’ finished appearance.

The two images below are from a major palimpsest manuscript source in Florence. The music manuscript was completely scraped and refinished, often to the extent that all that remains on a leaf are the very edges of the music. Some pages however are better, and what can be seen once the overwriting is removed is eminently transcribable, particularly as it is now known that many of the works have concordances in other sources which can act as a check against the found music in San Lorenzo 2211.

The images below are provided for viewing only by kind permission of the Chapter of San Lorenzo, Florence. Please do not copy or reproduce them without permission.