I’m trying to combine aerial imagery and lidar in RealityScan 2.0.1.
The images have been taken with a large-frame airborne camera at approx. 4km altitude, the lidar scan was done in a different flight at much lower altitude. All data is already registered/geo-referenced. During the lidar import, I have generated virtual cameras at an altitude of 150m. The lidar point cloud only has intensity, no color.
I can align either the lidar data or the images, but if I try to align both, the resulting component only contains registered poses for the lidar data. If I turn alignment off for the lidar data, the images get aligned. I’ve added a number of tie points to both images and lidar point clouds, but this didn’t resolve the issue.
I’ve tried different combinations of alignment settings, including turning off prior orientations and positions, but the results stay the same. What am I doing wrong?
After alignment do you see both components in the 1Ds view? If so, are both of them georeferenced? if they are, you can try to use the Merge georeferenced components setting in the Alignment settings. Were the laser scans imported as georeferenced in a proper coordinate system?
But I suppose the best option there is to use control points, which should be placed properly (at least three control points placed on at least three images and LSPs, the same for both components). For this case it would be better to keep the georeferencing only for one source (the one you believe more).
Alignment produces one component, showing only the lidar camera poses as being registered. I’ve tried aligning the images without the lidar data and then merging the components, but this did not produce a result. The lidar data was imported as georeferenced and I did use the same coordinate system for all data.
I tried using control points (in tie point mode), but this did not result in lidar and imagery being merged, the imagery was ignored.
How do I deselect the georeferencing of one component during merging? With the “Use camera priors for georeferencing” setting?
Can you show a print screen of a 1Ds view after alignment?
For merging components you need to export the aligned components and then use those components in a new project.
Can you show the control points placing? Also, 1Ds view with the control points?
In this case the option Use camera priors for georeferencing should be set to Yes. Just select the wanted data/group of data in 1Ds view, like:
and then in 1Ds/Selected inputs set the Prior pose/Absolute pose to Unknown for wanted inputs:
I’ve now managed to align point clouds and images, in two different projects using different altitudes for the lidar virtual cameras and using tie points only. I’m not entirely sure what I did different now other than closing and re-opening RealityScan, but at least it works.