Hello, my name is Vincent Miailhe. From a scientific background in mechanical engineering, I branched out after my studies towards heritage and, more specifically, archeology. As an archaeologist for more than 25 years and agent of the National Institute of Preventive Archaeological Research (Inrap), I am specialized in topography and especially photogrammetry. I work in the French territory of Poitou-Charentes as well as abroad in Sicliy, Saudi Arabia, Iraq, Libya, Palestine, Jerusalem, and Syria in the framework of French archaeological missions under the Ministry of Foreign Affairs, and I teach photogrammetry and mapping for geographical information systems at the University of Poitiers.
Aware that archaeological excavation is a destructive discipline, it is important to collect all the three-dimensional information of the remains uncovered before their removal. Topography using a total station does not collect all of this information. So I turned to photogrammetry in 2011 to meet this demand, as well as for other reasons. The first is the cost—a digital camera is less expensive and more easily transportable than a laser station. The second reason is the possibility of generating high definition orthophotographs of the remains, both in plan and in elevation, a very popular practice in architectural archaeology, a discipline in which I often intervene within Inrap.
Initial Forays into Photogrammetry
I started photogrammetry with PhotoSynth for practical reasons: my computer was not powerful enough and manual point matching, which was required by the most popular software at the time (PhotoScannerModeler3D), took too long when working with more than a hundred images. PhotoSynth allowed me to align a large number of photos and create a dense point cloud. I carried out my first photogrammetric survey on the castle of Coudray-Salbart, an imposing fortress of the 13th century AD in the town of Echiré in the Deux-Sèvres (France).
This digital surface model combines both a conventional topographic survey carried out with the aid of a total station and photogrammetric survey only of the castle. I used 3DReshaper to join the two surveys and 3ds Max to complete the missing areas of the mesh, including the top parts of the castle that could not be recorded because I did not have a drone. This model remains simple and without texturing, but this first test allowed me to see the many possibilities that I could draw from photogrammetry even if only the topometric data. My second model and first that I textured is the abbey of Maillezais in Vendée.
Tools and Workflow
All my photogrammetric surveys (Olympus OM-D E-M10) are coupled with a topographic survey completed with the aid of a total station (Leica station TS06). I cannot conceive of photogrammetry without a topographic survey. It allows me to scale the 3D model, to georeference it—essential in my work as a cartographer—and, of course, what is most important for me, to verify the accuracy of the 3D model that remains the very heart of my topographical work. I work with different software: PhotoScan, MicMac, 3DF Zephyr, CloudCompare, MeshLab, AutoCAD Map 3D, 3DReshaper, 3ds Max, QGIS.
Now all my professional interventions on complex remains or artefacts are done by photogrammetry. My most complex subjects have been:
- the curtain north-east of the castle of Coudray-Salbart, entirely scaffolded on its two faces (Image 1), leaving less than a meter of space to take the 6,500 photos;
- a network of cellars on two levels under the Hotel Claveurier in Poitiers.
These two interventions required a great work of topography with complex pathways in cramped and difficult to access places like spiral staircases or underground casing; but without this topographic work, I could never have realized these two 3D models. I love challenges.
Sharing Data on Sketchfab
If digital surface models allow me to extract vectorized planes and orthophotos, what can I do with these 3D models? Most of the time this data remained on my computer, my own colleagues could not open it. This is where Sketchfab comes in. All at once it is a communication tool during conferences, allows for dissemination of discoveries to people who can not visit a site, and conserves the memory of cultural heritage often threatened with destruction. Now, I take more pleasure in doing photogrammetry because I know that these works will be shared beyond the scientific community.
I take care to put the 3D model in its historical context, including an explanatory text, and in its geographical context by sharing plans or images of the environment. It is important for me to transcribe the sensations that I have during my surveys. That’s why I like adding sounds to my models, like the sound of the wind in a gallery (e.g., in the hypogeum that I recorded in Palestine), or the water that flows in an aqueduct conduit, or croaking crows in a castle… These are small details that, I hope, immerse users when consulting my models on Sketchfab.
Practical Applications for Photogrammetry
The digitization of heritage also makes it possible to restore the original appearance of a monument. During a photogrammetric survey on a part of the church of Saint-Eutrope in Saintes (Image 2), I masked out the protection grids of the stained glass windows and I was able to obtain a model of the Roman chevet with holes in the bays. Having done photogrammetry inside the church, I was able to show the stained glass as it could be seen from the outside, restoring the original appearance to the Church of St. Eutrope.
I find a lot of interest in the use of 3D printing, especially for the blind. It is difficult for them to perceive a monument of great dimension or an artifact that they cannot touch—a description is often their only alternative. With 3D printers, one can easily materialize digital models. Through these impressions, blind people can touch the object and get an idea of the volumes and the details of a monument or an artifact. I did this with the Arch of Germanicus in Saintes (Image 3) from a photogrammetric survey.