Hi everyone,
I’m trying to merge two components in RealityScan but I’m running into issues due to changes that occurred between the two data acquisitions.
Here’s the context:
-
I performed a LiDAR survey of a bridge during the summer. At that time, there were guardrails and handrails installed.
-
More recently, I conducted a UAV (drone) survey, but in the meantime, the guardrails and handrails were removed.
Now I’m trying to merge both datasets in RealityScan, but I suspect that the structural changes are causing problems with alignment.
The software likely can’t find enough common features between the two components.
Question:
Is there a way in RealityScan to manually merge components, or to exclude the inconsistent areas from the matching process?
Or is it necessary to switch to RealityCapture for more advanced control?
Any advice or workarounds would be greatly appreciated.
Thanks in advance!
Hi Christina, there is quite a lot of posts about merging such data.
The best option for you will be usage of control points.
Hi, I've used several binding points and ground control points, but the processing freezes and crashes.
I think the different geometry of some elements is causing the processing to freeze.
How many points have you used? How are they placed? In which step are you getting the crash?
Such small difference in geometry as handrails shouldn’t affecting the process. As I suppose the bridge stayed the same.
In the extreme case you can remove those parts from your scans and try then again.
I used eight GCPs, evenly distributed and not aligned. This doesn't just apply to the railings, but also to the balconies. I've added photos to clarify.
I used eight GCPs, evenly distributed and not aligned. This doesn't just apply to the railings, but also to the balconies. I've added photos to clarify.
Now
Before
What was the main part of the scanning? The bridge’s path, pillars or the covered areas?
But still I suppose this should work. Also, it is possible that those removed parts will be part of the produced mesh.
Can you show the print screen of the placed control points? Are they measured on botch components (laser scan and images)?
What do you mean that they are not aligned? When did it crash?
Both the scan and the UAV flight were performed on the entire bridge. The scan was performed first, and at the time, the walkways shown in the screenshot were present at the top of the piers. Since then, the walkways have been removed, and at the time of the UAV flight, they were no longer there. The alignment starts but stops at about 20%, then after about 4-5 hours it crashes.
Here is the distribution of the GCPs.
Control points were taken from the fixed and georeferenced lidar cloud, and were collimated onto both the cloud component and the flight component.
OK, so there the walkways won’t be a problem.
As you showed the component it is created from laser scans, but those control points aren’t assigned to the component (they are not blue, they are grey). The control points have to be placed over at least three images for image component and at least two/three LSPs for laser scan component. What I see is just a suggestion where such control points should be places, but it is not assigned.
What are your alignment settings? How many images do you have in your project? What is your RAM? What kind of crash are you receiving? Do you see some red error there?
Do you have also the component of the images? If so, is it georeferenced?
The control points have been placed on both components, just as you say, except that the original project is processing the union between the two components. I created a new project with the two components just to show the control point placement.
I have 8622 images in my project. The two components, laser and image, are both georeferenced.
My PC has 64 GB of RAM.
I don't get a red error, but a dump message.
RealityScanCrash_9448_76676-667_2_0_0_118619_RS.dmp (524.8 KB)
Now
Ok, it is possible that your RAM is insufficient for such work (or you are on the edge of it):
- 2,000 images - 16GB RAM
- 4,000 images - 32GB RAM
- 8,000 images - 64GB RAM
- 16,000 images - 128GB RAM
Also, you can try to use the latest version.
There are also some issues for such big components - the merging of the components can take quite long time. Try to import the images just as images into the project, not as the component.
The laser scan is georeferenced in local or global coordinate system?
The laser component is georeferenced in EPSG 32633.
Thank you for your feedback.
