3D Modelling with Smartphones, interior spaces

Background

3D survey techniques are becoming ever more popular to document and study movable and immovable cultural heritage. 3D models of objects and sites or monuments offer researchers new possibilities for interpretation and research. A 3D model of an archaeological site at a particular moment of the excavation for instance makes it possible to return to this situation also after it has been further excavated. 3D models moreover make excellent tools for teaching and outreach, offering audiences the possibility to freely engage with objects, turn them, scrutinize details, and so on, without having them physically touch them.

Among the available techniques for the creation of 3D models, digital photogrammetry is currently the most widespread. DSLR cameras are a lot more cost-effectiveness than for instance laser scanners. They are also much more portable and allow for the registration of colour as well as shape. However, professional cameras are still an additional piece of equipment that remains costly for many researchers, students, and cultural enthusiasts both inside and outside of the UK. At the same time, all these interested parties very often possess a smartphone, the cameras of which have improved tremendously over the last couple of years. For this reason, together with my brilliant partner, the Virtual World Heritage Lab, we started to test if smartphones would in the meantime not be a readily available and appropriate tool for the capture of photos from which 3D models could be build?

A previous fellowship focused on exploring the possibilities of smartphones for the 3D modelling of smaller objects. The characteristics of several current smartphones (Android and iOS) of different price ranges were tested and the effectiveness of their images for photogrammetric use was tested. These experiments were conducted in a laboratory of the Politecnico di Milano by Dr Laura Loredana Micoli and Professor Gabriele Guidi. Their results exceeded all expectations. Based on their experiences and advice on optimal methods for data acquisition and post-processing, a protocol or ‘how-to’ manual was designed. It contains easy-to-follow steps to allow also non-specialists to capture photos of an object with their smartphone that could then be used to create the 3D model using data processing software.

Set-up of the current project

In this follow-up project, we are building on lessons learned to test if it is possible to arrive at high-quality photogrammetry and 3D models of walls and interior spaces again simply using the camera on your smartphone. Although there are some similarities with the previous project, differences are great as well. The objects to be captured are much larger in size, they cannot be moved to a location where they are easily reachable and where the lighting conditions are ideal. This last aspect is particularly important for photogrammetry as light conditions can drastically influence results.

Case-studies

The first phase of the project again consists of tests both in and outside of the laboratory Politecnico di Milano. Dr Laura Loredana Micoli, assisted by Umair Malik and advised by Professor Gabriele Guidi, selected three case studies. A first case-study is an external wall, about 7.5x3 m in size, flat, but with bright-coloured graffiti (Figure 1). As a second case-study a portion of a white internal wall at the University was chosen, characterized by the presence of two black drawings (Figure 2). Finally, the third case study is a portion of the wall inside the so-called 'square tower' (Figures 3-4). Originally one of the towers of the carceres of the Roman circus in Milan, it was later transformed into the bell tower of the church of San Maurizio. The wall is plain but shows the typical signs of rebuilding and renovation throughout the centuries, from the Roman period up to the 19th century. Access to the tower was kindly granted by, Dr Anna Provenzali, the Director of the Archaeological Museum of Milan, to which the tower currently belongs. This last case study in particular is rather typical for situations encountered on archaeological sites and in cultural heritage buildings.

These case studies give us the possibility to examine the impact of various potentially problematic aspects in the photogrammetric imaging of architectural/archaeological surfaces, e.g., a very large surface to be captured, an entirely flat surface, a almost plain surface, low lighting or, conversely, very harsh lighting conditions.

Determining the quality of smartphone photos

Laura and her team have again created accurate reference models of these case-studies, using a terrestrial laser scanner (Figure 5). The models produced with scanners show even the smallest details of the wall finish. As these are important for archaeological and architectural analyses, the same level of detail will need to be visible in the smartphone-generated models for them to be useful.

Figure 5. 3D digitizing of the test objects with laser scanner (Photos: LMM)

The same surfaces are now being captured with a smartphone. More in particular, based on the excellent results of the previous project, a Samsung Galaxy A7 smartphone is used. Additional equipment includes a rigid mobile phone holder, a led illuminator ring and a telescopic selfie stick. The illuminator (Figure 6), which is very popular for selfies and extremely cheap, is highly useful especially in indoor settings as it provides a continuous illumination of the scene and a proper autofocusing of the camera. The selfie stick, equipped with a rotating head, makes it possible to get the phone and camera close to surfaces high-up or in any other way difficult to reach (Figure 7). Finally, a portable power bank is useful to ensure that the illuminator can be used when there is no other source of power.

Images are shot in the highest quality mode and saved in a simple .jpg format. A freely downable phone app ensures that the camera is set to take an image ever 2’’. After every photo, the photographer needs to move sideways, maintaining a constant step and an equal distance from the wall, and repeating the same operation at different heights. This ensures a ca. 80 percent overlap between adjacent images necessary to be able to process the photos afterwards.

Of the first case study, a total of 223 images have thus been taken. They were processed using Agisoft Metashape Pro, one of the software packages more extensively in archaeological and museum applications (Figures 8 and 9). The analysis of the first try-out is showing that the reliability of the resulting image is the highest in the centre of the object, degrading towards the sides. This teaches us that it will be necessary to take this into account from the beginning and that it will be necessary to shoot more images at the peripheries if we want to arrive at a proper 3D digital model.

The quality of the resulting model is currently being compared to that of the standard produced with laser scanner.

2021-22 | 3D Modelling with Smartphones, Interior Spaces

TORCH Heritage Programme