• Home  /
• Blog  /

# Can I use PhotoModeler to model an object from a single photo? For example, a photo from a book?

There are some objects that can be modeled from a single photograph with PhotoModeler. A few of the tutorials shows just that. There are some requirements though and many photographs you come across cannot be used.

In some fields, such as forensics or archaeology, it is quite common to need measurements from unknown photographs. These unknown photographs can come from archives, historical files, police, bystanders, etc. In these cases, there is little or nothing known about the camera that took the photographs. PhotoModeler normally requires information on the camera (focal length, principal point and format size) before photographs taken with it can be used. The Inverse Camera feature of PhotoModeler overcomes this problem.

To use Inverse Camera some information about the scene or object is required. This information is called a Constraint. InverseCamera can use known 3D points or Axes Constraints. You may be asking, “If I know the 3D positions of the points already why would I need PhotoModeler?”. What PhotoModeler solves is the case of knowing some points or shapes in the photographs but not the information required. For example, you might need to get the relative position of two cars that appear in a few photographs from an unknown camera. You know the dimensions of one of the cars but not the dimensions of the other car nor their relative position. This is a case Inverse Camera can give you the answers you want. Inverse Camera is carried out by using Constraints. The most common form is done with Control Points. Single photo Axes Constrained projects can also perform Inverse Camera to a certain extent.

If you are using Axes Constraints, you have to have a photo with the correct perspective. With Axes Constraints you can solve for focal length if you have a strong two or three point perspective and you can solve for focal length and principal point if you have a strong three point perspective. A single point perspective photo has just one vanishing point and the horizon is near the center of the photo. If a photographer was standing in the middle of some train tracks and took a photo looking down these tracks and the tracks vanish in the distance to a point near the center of the photo then that is a single perspective photo and Inverse Camera will not work. A two point perspective photo has two vanishing points. A photo taken at 45 degrees to the corner of a building but the camera still parallel to the ground plane (which makes the vertical edges appear vertical in the photo) is a two point perspective photo. If the perspective is strong enough PhotoModeler can solve for the camera position and focal length for this photo. You would mark two sets of lines or edges and set up two Axes Constraints. A three point perspective photo has three vanishing points. A photo taken at 45 degrees to the corner of a building with the camera pointed towards the top of the building (which makes the vertical edges appear to vanish to a point near the top of the photo) is a three point perspective photo. If the perspective is strong enough PhotoModeler can solve for the camera position, focal length and principal point for this photo. You would mark three sets of lines or edges and set up three Axes Constraints.