Blog

Leveraging Drone Technology and Reality Capture for Design and Construction

Posted by dan kemme

Drones + Photogrammetry = Reality Capture

Ever since Canoma was released in 1999, the idea of taking two-dimensional digital images and creating 3D virtual models has promised tremendous possibilities for recreating reality. Canoma allowed the construction of the existing world based solely on manual perspective input from just one or two images. Virtual building sites and streetscapes could be constructed quickly from a handful of photographs.

Canoma Interface
Canoma Interface

While revolutionary for its time, the intervening decades have seen a tremendous change in what technology makes possible. Rather than manually telling software how to interpret vanishing points, we can now take drone images with embedded GPS locations and have software reconstruct the scene in 3D. Using hundreds of photos. With incredible accuracy.

Drones

DJI Phantom 4 Pro
DJI Phantom 4 Pro

Why drones? In theory, any images taken with any kind of camera can be used for photogrammetry—the process of comparing how the perspective in each image changes as you move around an object. But drones offer multiple advantages for this work:

  • Each image is tagged with its GPS coordinates, which allows the processing software to instantly understand each photo’s position in relation to the others;
  • The drone can be programmed to fly an automatic pattern that guarantees the photos are taken from the correct locations with the ideal amount of image overlap;
  • Even when the site conditions require hand-flying, software can be used to automatically trigger the camera shutter based on the drone moving a certain distance from the previous photo – say, every 6’ vertically and horizontally;
  • Drones can easily take pictures from locations that would otherwise be very difficult or impossible to reach.

The key is that the object being photographed is surrounded by images that have specified amounts of overlap. It’s the overlap that allows the software to compare how points in each image change as the photos move around the object. That’s how 3D relationships are determined.

Photogrammetry

Point Cloud example
Each "point cloud" location ties to multiple photographs

Photogrammetry has been used since the dawn of photography, and in its simplest form is merely measuring distances using photos.
For our purposes, photogrammetry is the ability to extract 3-dimensional data from 2-dimensional images. We do this with software such as Pix4D, which compares each pixel of each image to all the other images, and then places those pixels in 3-dimensional space with coordinates.

The result is known as a “point cloud.” The process requires a lot of computer processing power, but can achieve results that are striking and very accurate.

Densified Point Cloud
Densified point cloud with more than 120 million points; view Sketchfab model

Reality Capture

So much of what we do is about both creating reality and documenting existing reality. With tools like drone imagery and photogrammetric processing software, we can very quickly capture existing site and building conditions. This can be done with considerably more accuracy and speed than either recreating a building virtually with old drawings (that may not record subsequent changes)—or heaven forbid—hand measuring existing facilities.

This video illustrates both the process and the results of reality capture for Lusk Hall, the administration building at the New Mexico Military Institute. This building was used as a study for the workflow and data collection of drone reality capture.

While the basic geometry of the building is relatively simple, the brick and stone detailing would have been almost impossible to record by hand. Using this technique, the building’s existing conditions can be recorded down to each individual brick, and each carved face of the upper tower.

Dekker/Perich/Sabatini is continually updating our abilities with all types of reality capture and reality visualization. Using drones in this way, beyond simple photos or videos, is another way we leverage our internal expertise to provide value for our clients and information to our teams.