'Point cloud' of the Three Graces by Antonio Canova © Factum Foundation
Photogrammetry is a 3D recording technique that employs 2D images to create a 3D model of an object or surface. It involves taking hundreds of overlapping photographs of an object from many different angles and processing them using specialised software such as RealityCapture (RC) or Agisoft PhotoScan. The digital 3D model can be used for study or outputted as a physical object via 3D printing or CNC milling.
Photogrammetry is a fundamentally democratic technology: it can be done by anyone with a phone camera and access to free photogrammetry software. However, in order to produce the kind of high-resolution 3D models that accurately capture the geometry and surface detail of an object, a high level of skill is required for both data recording and data processing. In general, the better the camera sensor and resolution capacity of the lens, the better the data will be. The skill of the operator, the quality and number of images captured and the distance of the camera sensor to the surface of the object also play a large role in determining the resolution and detail of the 3D data. Under ideal conditions, photogrammetry can record 3D data of a surface to an accuracy of 100 microns, on par with other close-range 3D recording systems available today.
Factum Foundation currently employs photogrammetry – sometimes combined with white-light or LiDAR data – to produce geometrically precise, but also highly detailed, 3D models of artworks. The technique has a number of advantages over traditional 3D scanning technologies, including the possibility of recording colour information at the same time as 3D data. Photogrammetry is also inherently ‘portable’ – in most cases the equipment (camera, tripod, flashes) can fit into a small camera bag, making it a particularly useful tool for recording at remote or dangerous sites.
Long-range or ‘aerial’ photogrammetry was developed in the 19th and 20th centuries as a cartographic tool to produce three-dimensional maps of the surface of the earth. In the 21st century, advancements in computer processing power have resulted in the rapid growth of close-range photogrammetry, which can be used to capture both the shape and fine surface detail of 2.5D and 3D objects. The technique continues to evolve: for example, long- and close-range photogrammetric recording is now being coupled with drone technology to facilitate the documentation of large cultural heritage sites – operators are able to photograph at heights and angles otherwise impossible to reach without scaffolding.
Factum Foundation’s Manuel Franquelo gathering photographic data using a drone, for the 3D model of the Cochno Stone, Scotland’s largest and best example of the Neolithic or Bronze Age © Factum Foundation
Photogrammetry software works by identifying common features on an object's surface across multiple images. Each of these ‘features’ can be described in 3D space by a series of coordinates (x, y, z). A grouping of such points is called a ‘point cloud’. During post-processing, the points are also triangulated (connected to one another by lines) and conjoined with flat planes to produce a 3D model – a 'geometric mesh’ composed of vertices, edges and flat planes. By applying digital colour management methods during a recording session, it also becomes possible to simultaneously obtain both 3D and accurate colour information about the surface of an object.
The images below show the process of producing a 3D model using the software RealityCapture:
An object is photographed from multiple angles. The software detects the camera positions and plots them in 3D space as seen in the image above, where each triangle represents a camera position. Screengrab from within RealityCapture © Factum Foundation
The screengrab above shows 'control points' that had to be chosen manually by the user inside RealityCapture in order to produce the 3D model of the "Cellini Bell" - the user located 'common points' in different images. Photogrammetry software does this automatically for millions of other points found across images of an object. Image © Factum Foundation.
The point cloud up-close. Each individual point in the image above represents a common pixel found in more than one image. Screengrab from within RealityCapture © Factum Foundation
The blue lines represent how images are interconnected to create the point cloud. The software recognises common points in different images and plots them in 3D space. Screengrab from within RealityCapture © Factum Foundation
Designing and imparting training courses in cultural heritage digitisation with local partners is a large part of Factum Foundation’s work. The Foundation has demonstrated the potential of photogrammetry as an adaptable, relatively simple technology with which trainees can produce excellent results even after brief training periods. Photogrammetry also represents a useful transferable skill that people can later use for different applications that do not necessarily need to include cultural heritage digitisation.
The Sepulchre of Cardinal Tavera at the Hospital Tavera in Toledo (Spain) was made in 1552 by Alonso Berruguete, one of the greatest sculptors of the Spanish Renaissance. As shown in the images below, it was badly damaged during the Spanish Civil War. The Sepulchre was recorded in 3D using a mixture of photogrammetry and LiDAR for a restoration project in conjunction with the Casa Ducal de Medinaceli. Marble has traditionally been viewed as a highly complex material for 3D recording and for photogrammetry in particular. However, with the right imaging equipment and processing experience, it is possible to obtain highly detailed, precise 3D models. The 3D model of the Sepulchre will be used to produce a digital restoration of the four Fates, whose faces suffered the most from the vandalism in the war.
The Jamnitzer or "Cellini" Bell is a tiny silver bell thronging with insects, lizards and all manner of plants cast in fine relief. Although once believed to have been the work of the Italian sculptor Benvenuto Cellini, the bell is now firmly attributed to the German goldsmith Wenzel Jamnitzer. Factum Foundation were commissioned to produce a facsimile of the bell for Horace Walpole's recently restored Strawberry Hill House. Recording small objects in 3D at high-resolution is fiendishly difficult. Photogrammetry makes this possible, but the process requires a huge amount of patience and precision. Recording the bell with photogrammetry was further complicated by its shiny surface and it had to be recorded twice: the first time using cross-polarising filters to reduce shine in the photographs, the second without to obtain true-to-life colour. A further difficulty was caused by the fact that, during the recording break, the object had to be moved back to its case - moving an object during a photogrammetry recording 'confuses' the software, which cannot use the object's surroundings to identify common points. In order to solve these issues, control points had to be added manually during post-processing. However, the results of the recording show the potential of using photogrammetry to record cultural heritage at high-resolution.
The digital model was later printed and gilded to produce the facsimile.
For more information about Factum's training programmes in photogrammetry, please contact email@example.com.
Otto Lowe / Recording, processing, training / Factum Foundation
Pedro Miró / Recording, processing, training / Factum Foundation