How do Shapes-based projects and Points-based project differ? The main difference is which algorithm runs to create the 3D data. You decide whether a project is to be a shapes-based project or a points-based project at the beginning when the project is first created in the Getting Started Panel.
Here is a table of differences with further explanation below:
|
|
Shapes-based project |
Points-based project |
|
Photos oriented by |
Shape algorithm |
Relative orientation algorithm |
|
Coordinate frame (before scale/rotate) |
Defined by base shape |
Arbitrary (free-network) or defined by control points |
|
Points are computed by |
3D intersection math |
Bundle adjustment algorithm |
|
Shape’s reference frame |
The base shape |
Tied to points |
In a photogrammetry project we need to solve not only the 3D model but also the positions and angles of the photographs (called the camera station orientation) and we need to know or solve the camera’s internal parameters. In a Shapes-based project the photos are oriented when enough edges are marked on the base shape and when the shape-algorithm runs. The shape-algorithm is responsible for solving the shapes’ parameters and positions as well as any camera parameters when needed. In a points-based project the photos are oriented by a relative orientation method based on six or more referenced points. Single-photo points-based projects are possible too using control points (in which case the photo is oriented by a point-resection algorithm – see Control Imports in Importing and Setting Up Coordinate Systems for more information on control).
Every project has a 3D coordinate reference frame. This can be thought of how the x, y and z axes line up with the project data. In a shapes-based project the coordinate frame is defined by the first shape added to the project (the x, y and z axes will line up with some of the edges of the first (or base) shape). In a points-based project with no control points the coordinate system is somewhat arbitrary (hence the importance of using the Scale/Rotate feature in points-based projects). If a points-based project has control points then these define the coordinate frame.
When points are marked and referenced across oriented photos they become 3D. In a points-based project these points are first computed by light-ray intersection math but later are computed more accurately with a bundle adjustment algorithm. In a shapes-based project referenced points are only ever computed by the light-ray intersection math. These points may be less accurate than they would be in a comparable points-based project.
When a shape is added to a shapes-based project it will always be related in some way to the first shape added to the project (or the base shape). This may be via relations to the base shape or via a shared ground plane. When a shape is added to a points-based project the shape must be either tied to some existing 3D points or be related to a shape that is in turn tied to some 3D points.