I’ve been working on a drone mapping project in RealityScan, and after the second alignment, the result was a single component with 2736 out of 2768 cameras being used. Prior to the alignment, all of the control point residuals were less than 2 pixels after careful inspection and deletion of blurry and marginal measurements. After the alignment, many of the control points’ residuals were higher, several of them were above 4 pixels, and the majority of reported surveyed ground control point positional residuals averaged 0.03 us-ft or less.
In the screenshot below, on the left side, we see the first several lines of the results of the second alignment and next to it on the right is exported .csv with my manually inserted header; Point,Y,X,Z. The values in the .csv file are not intuitively communicating anything meaningful to me. What are these values and in what units are they expressed?
In the next screenshot below, I’ve selected the first survey point: HM1-01 and among other things, the original estimated error settings from when I set up the project are shown; i.e., 0.16 & 0.26 us-ft. All points were located with GNSS RTK with most falling under mixed tree growth and increased multipath; e.g., and why the error estimates were jacked up a bit.
The Alignment report for the selected HM1-01 shows a total deviation of 0.09 but the table (first screenshot, left side) shows an error of 0.03 us-ft, and the exported .csv shows no obvious correlation to either of them.
Is it possible (?) to somehow export a least squares adjustment GCP report (.adj) that includes a header, units, and at least (?) :
PointName, X, X residuals, X error estimate, Y, Y residuals, Y error estimate, Z, Z residuals, Z error estimate
Why this ask?
Having a ready list of post-adjusted GCPs that can be sorted can help to reveal possible errant surveyed points. This inadvertent process of RealityScan discovering a survey issue happened earlier in the season in this same challenging environment.
For example back in May before the dead trees were cleared and low limbs trimmed, RealityScan calculated a less than agreeable error for GCP9 and GPC11. The resulting GCP coordinates from additional occupations made RC happy.
In July GCP11 and GCP4 were used as backsights for some total station work and the GNSS receiver was set on auto repeat while also serving as the backsight. Each green dot represent 180 epochs for a single data point. Individually, they may look agreeable, but over time they hint at why I chose to jack up the RealityScan project estimated error settings.
