Capturing reality for objects, buildings, and cities: Reality capture explained

Pete Kelsey, Former Strategic Projects Executive, Autodesk: If we don't have a computer model of all this stuff, how do we create one?

We're using a bunch of tools to do exactly that.

Some laser-based tools known as lidar, photo based tools, just a camera to use what we call photogrammetry.

Brian Mathews, Former VP Platform Engineering, Autodesk: So photogrammetry is a process of getting a three-dimensional model or object from photography. Photogrammetry works by having an object and taking different pictures from different positions. And the computer goes in and looks at every single pixel of each image and every other image, and it makes a database of features. A unique shape from this angle of this photo is matched to one from another photo, and that tells us where in space the camera was when the photo was taken. And once we know where all the cameras were, when the photos were taken, we can now triangulate based on those triangles and create a three-dimensional model of all of the points.

Kelsey: Lidar is a laser-based technology. It can be mounted in an airplane. We can put it on a tripod.

It's a piece of gear that shoots a laser out up to a million times a second, and it creates this, what we call a point cloud. This incredibly rich, photorealistic cloud of points that we can then bring into the computer and start converting that raw data to actually solid models.

Mathews: These tools are very important because the old way of doing documentation of important historic sites was really done through surveying. And in traditional surveying, you could maybe make a few hundred measurements in a day. A laser scanner today takes about 1 million 3D measurements per second. You're not just finding the edges of buildings or the corners of the nose at Mount Rushmore, you are getting the entire experience of what is there. And that allows us to do better visualizations, better ways of studying how to care for these objects.

Kelsey: What we're doing here is not only a good fit for critical structure, but it can work for any physical thing that hasn't been modeled in the computer. So a building, a car, a thing—we've got solutions for that.