There are a number of parameters that are available within a relationships properties that are directly used to compute the relative point deviations and influence its use in optimizations as well as for reporting.
Allows more emphasis to placed on the selected relationship. Often times weighting can be tweaked to bring a part in tolerance. For more details see Normalize Weighting
When a fit constraint is added to a relationship it defines a “dead” region where the relationship has no influence. For example, a high limit can be set at +0.1 and low at 0. This will define a region from 0 to +0.1. When the solution reaches this region, the relationship will stop influencing the solution and the fit will only continue to optimize if other relationships are involved.
Sub sampling is useful for large data sets. Often all of the data is not required for the fit.
Relationships provide sub-sampling controls to increase processing speed for large data sets. These include two categories:
Fit using every i-th point
Using the Fit using every i-th point control provides the ability to display a reduced number of vectors that continues to increase in number as the data set increases in size. When i=1, all the point deviations are used for display and calculation. As i increases so does the number of skipped points.
Fit using no more than n points
Using the Fit using no more than n points control provides the ability to place a hard limit on the number of points used in the relationship for display and reporting purposes.
Projection options for comparison between a point and an object offer several methods for compensation. See Projection Options
Tolerances can be set for any relationship and apply to all the individual point deviations for both alignment and reporting purposes. Individual tolerances can also be applied directly to nominal points and these point tolerances will be picked up and used within a Group to Group ( and Group to Nominal Group) or point to point relationship. These individual tolerances will then be reported for the relationship report unless the override option is checked within the relationship properties dialog. Point to Object relationships will not use point tolerances as it is assumed that tolerances on measured points should not apply to a surface analysis.
Outlier rejection is a great way of removing stray data for the fitting solution. Stray points can often impose difficulties for the optimization.
Desired Measurement Count