Photogrammetry is the science and activity of creating a 3D model based on a series of photographs. By photographing a subject from a wide variety of angles, with consistent lighting and no motion, the images can be processed by software to create a 3D model.
The above example is a screenshot from the Agisoft Photoscan software, available for PC and Mac, the most popular photogrammetry app.
Based on the principles of stereoscopic vision, photogrammetry is actually an old technology, originally developed during WWII to locate enemy positions based on aerial photos. This application is still used often today by drone camera enthusiasts.
Learn more about using photogrammetry with drone cameras from Sean RG Barberie, whose work is shown in the above screenshot.
With smartphones and powerful desktop computers, this technology is now more widely available, and is used in professional sports, the video game and VFX industries. In fact, the newest iteration of Google Earth uses photogrammetry extensively. How extensively? The whole freakin planet!!!
Visit https://earth.google.com/web/ and navigate to any location in the world, then click the “3D” button in the lower right hand corner, click and drag with your middle mouse button, and be amazed.
Photogrammetry software goes through a series of steps to create a point-cloud from the photos, then builds a polygonal mesh from the point cloud, and finally texture maps the colour information from the photos directly onto the mesh. UV maps are automatically generated for the model, but can often be chaotic and disorganized.
Above is a simplified series of steps in photogrammetry reconstruction. Software will first produce a sparse point cloud, then a dense point cloud, and finally a solid trigon mesh.
Photogrammetry requires photos to be taken in a systematic way, as haphazard photographing will introduce errors in photogrammetry measurements. The key to a successful photogrammetry reconstruction is in the redundancy of “look angles” of a subject, i.e. consistent and methodical overlapping in images. The ideal is to obtain photos such that each point on the surface of an object is captured by 9 overlapping photographs. Achieving this ideal set of photos can be greatly enhanced by the use of a rig, where cameras are arranged in fixed positions, with all measurements between cameras and placement having been built into a lightweight but solid frame. Photogrammetry rigs decrease the time required to obtain photos substantially.
Rigging multiple cameras together will also require a separate piece of software to coordinate and synchronize the cameras, such as Breeze Systems.
Photogrammetry applications for science, biology and medical visualization are promising. Archeologists and anthropologists have used photogrammetry to study and preserve artifacts and wall carvings in remote locations. Cultural Heritage Imaging is one such group that is dedicated to using photogrammetry to preserve archeological treasures. Their detailed production process is a great resource to learn more about the science of photogrammetry.
Use of photogrammetry for biological specimens is still in early stages, since the reflections on many biological subjects work against the ideal photogrammetry process. However, as this technology continues to improve, it’s presence and usefulness to the medical illustrator will become ubiquitous.
If you are interested in trying to create something with photogrammetry, check out Paul’s tutorial here.
Paul Kelly, MScBMC (1T1), CMI
BMCAA Treasurer
IG: @keloskopy
Twitter: @pizawl