Developing a Simple and Inexpensive Reflectance Transformation Imaging (RTI) Dome for Teaching and Research Needs, Part One: Background
Reflectance Transformation Imaging (RTI) is a process that captures multiple directions of light in multiple photographs and computationally recombines the photographs in order to store a format which allows the arbitrary changing of the direction of light in the resulting image. It has many uses for heritage items such as coins, seals, and small archaeological artifacts, specifically because it allows for highlighting small surface details, which are often of high significance for the study of such objects. This post will not go into detail about the process, for which I refer you to the description and documentation on the Cultural Heritage Imaging site.
While RTI can be done by hand, this process is time-consuming and best reserved for larger objects; smaller objects are generally imaged with a dome which provides a set of lights in enough directions that all the directions of light can be reconstructed programmatically. These domes are not readily and inexpensively available; commercial examples can cost in the neighborhood of €10.000–30.000. This means that many scholars and institutions with limited need for RTI have no access to them, either for digitization or teaching and experimentation.
At the ZIM–ACDH, I teach a course on Imaging Techniques for Heritage Management, where we learn about multispectral imaging, photogrammetry, and RTI, amongst other practices. The ZIM–ACDH, partnering with the library and funded by the Faculty of the Humanities, acquired a multispectral camera, and photogrammetry is simple enough to try out with a cameraphone. However, the manual process for shooting RTI is slow enough that I don’t include a practical component in my course, and borrowing a dome from another university proved impractical. This drove my desire to have an RTI dome for teaching purposes. Our Professor for Digital Museology, Chiara Zuanni, also teaches a course focusing on the digitisation of 3D materials, which touches upon RTI. At the same time, we have a strong research interest at the ZIM–ACDH in diplomatics, driven by our Head of Institute, Georg Vogeler (this is after all the DiDip project blog), and RTI is an excellent process for capturing good detail for (often-hard to image well) seals that are attached to medieval charters. Given the interest in the department, Georg asked me to look into pricing, with the intent to possibly apply for funding for a departmental dome.
With this in mind, I started looking into both commercial and home-made RTI dome projects, trying to get a handle on the differences and how they might serve our needs. After a frustrating experience with the MSI camera, where the provision of a complete system by a professional supplier led to frustration with parts that we could not service ourselves, I was personally keen on the idea of a system which we could understand and maintain, independent of manufacturer support, which led me to pay particular attention to the various home-made systems that are documented.
The original home-made RTI system to take off was Brian Benchoff’s Affordable Reflectance Transformation Imaging Dome, originally a project submission (and winner) on Hackaday. While the documentation is excellent, and this project clearly inspires what follows, the project is rather complex. A fascinating alternative from Tim Zaman is made entirely of printed circuit boards; while apparently inexpensive to produce, I must admit that, being an open dome made of lightweight materials, it (perhaps wrongly) is a bit outside of my personal comfort zone for our needs. While the Austrian Academy of Sciences (ÖAW) has a commercial dome (which I am informed cost around €28.000), they also have a large dome which they built with the participation of students. The Cologne Center for eHumanities (CCeH) has a very helpful site documenting their experience with the construction of a dome, but their dome requires a custom circuit board, and they cautioned me that they were currently experiencing some issues with it. Projects by Richard Benjamin Allen (RTIPy) and Ted Kinsman (An Easy To Build Reflectance Transformation Imaging (RTI) System), while still requiring extensive soldering, were significantly simpler in their implementation, due to pushing more of the work onto the microcontroller of the Arduino, and were the most immediate starting points for me, thinking about how to approach the project.
The list, above, is not comprehensive, but these were the projects which most influenced my thinking about what to do locally, including dissatisfaction with a number of issues, including requiring specialized electronic skills (soldering at the very least) and triggering systems which generally limit use to one make (and sometimes, specific model years) of cameras. While we have a colleague with good soldering skills (and I have basic soldering skills), the latter issue was more of a problem: the Digitisation Lab in the university Special Collections uses Sony Alpha cameras, the ZIM–ACDH owns a Nikon DSLR (but no macro lenses), and I personally own a Canon T3i and two excellent macro lenses. So, a system for our use would either have to have its own integrated camera or it would need to be camera-agnostic. Both of these concerns drove many of the decisions that I made when coming up with our own prototype system, which I will outline in the next post.