Photo-masking the Mildenhall Treasure

This application enables the creation of a high quality 3D model of an archaeological artefact via process known as photo-masking. There has been a revolution in 3D modelling in recent years and it is now relatively easy to construct such models from ordinary digital photographs. Isolating the object depicted in these photographs, and masking out the background, is an important first step to achieving high quality results. The final 3D model will be made publicly available and is useful not only for basic documentation purposes, but also for graphical displays in museums, for inclusion in gaming and virtual reality environments, or for identifying different sub-styles in otherwise similar types of artefact (that might tell us about the date of the artefact, where it was made, or by whom).

This particular photo-masking application will help to model 3 pieces from the fabulous Mildenhall Treasure. Roman Britain, 4th century AD Found near Mildenhall, Suffolk, eastern England (1942 or '43)

This hoard is one of the most important collections of late-Roman silver tableware from the Roman Empire. The objects were found during ploughing near Mildenhall in Suffolk, eastern England, in January 1942 or '43 and were declared Treasure Trove in 1946. Although no coins were found to give a reliable date, the tableware's style and decoration is typical of the fourth century AD. The artistic and technical quality of the silver objects is outstanding, and though we do not know who owned them, it was probably a person or family of considerable wealth and high social status. To learn more about the Treasure, you can visit the British Museum website here

We would like people to draw a polygon around the object that they see in each photograph in order to identify its outline and exclude the image background. This allows the 3D modelling process to concentrate on the object itself and ignore irrelevant background information.

If you are interested in what a 3D completed model looks like, please have a look at an example here for a MicroPasts palstave model.

This project is on behalf of:

Find the code on GitHub