Controls Points are known and pre-measured locations in a 3D coordinate frame. For example, a survey of a building may indicate the accurate positions of the foundation corners to be (450.2,210.3,32.4), (480.5,210.0,32.0), (480.2,270.2,32.0), and (450.1,270.0,32.1) feet in a city wide mapping coordinate system. These points are known before you start measuring any other part of the building and these points can then be used to "control" the size, position and orientation of the remaining measurements.
Control imports define the project’s coordinate system so in this way they are like a Multi-point Transform. The difference is that control points are carried in the processing/optimization and can modify the shape of the 3D model. The project adjusts to match the positions of the marked control points, for example when taking points from a survey and using them to orient a project in the same coordinate system. Control points are also used when solving for unknown cameras using Inverse Camera features (see Inverse Camera solves for), for single photo projects, and for projects with weak camera geometry.
Control points are also used when solving for unknown cameras using Inverse Camera features (see Inverse Camera solves for).
Here are several cases when Control Imports are used in PhotoModeler:
• For Inverse Camera (recreating parameters for unknown Cameras): Control Points can be very useful when trying to recreate a description of a Camera from photographs of unknown source, which is often required for Perspective Matching.
• For single photograph projects: When you are extracting information such as Surface Drawings, photo-textures and orthophotos (see Orthophoto Export) from a single photograph, Control Points can be used to set up the planar geometry and to solve for the Camera Station location.
• For tying together and improving the accuracy of projects with large numbers of photographs spread out over a large or long area: If you are modeling a large area with many overlapping photographs, errors will slowly be introduced into the model as you go from overlapping photo to overlapping photo. For example, modeling a very long wall without any overall photographs using just photographs that overlap as you move along the wall will generally give a result where one end of the model is rotated or skewed relative to the other end. A sparse network of control points along the wall will help alleviate this problem. Control Points can also be used in a SmartPoints Project, to help tie long scenes together. See Using Control Points With SmartMatch.
• When Camera Station geometry is quite weak (i.e. low angles): Control Points can be useful when the Camera Stations are in poor positions. Two cases of this are: a) the Camera Stations have similar angles, and b) the project contains only two Camera Stations. In both these cases it is important to get accurate positions and orientations for the Camera Stations or the resulting 3D data may be inaccurate. Having accurate Control Points marked on these photographs can help PhotoModeler considerably in determining the true positions of the Camera Stations and hence improve overall accuracy.
• When the source of Control Points is of much higher accuracy than PhotoModeler and accuracy is a prime concern in the project: Control Points can also be useful in creating accurate 3D surveys. For this to be true in the normal circumstance with PhotoModeler, the Control Points have to be more accurate than PhotoModeler or they have the potential for degrading PhotoModeler's accuracy. For example, if you are using digital camera with a resolution of about 2500 pixels across and the Camera Stations all have good angles then you could expect accuracy from PhotoModeler of about 1 in 2000. To improve this project with Control Points, they should be accurate to about 1 in 10000 and the points should be well distributed in space and be well distributed on the photographs.
See Marking/Pinning Multipoint Transform, Control and Pinned Import Points for information on marking and assigning control points.