Strange scan problem

[scottkim]

We are using the FARO S350 scanner, and sometimes we run into the following problems that we cannot understand in outdoor scanning.

That is, when there is a nearby object to be scanned and a building background behind it, a part of the object to be scanned is cut into irregular shapes and positioned at a far distance instead of the original location. This is made by the scanner, not the registration software.

The attached screenshot was taken in Navisworks. In View-1, it's not a perfect view, but the shape of the brace is visible. And View-2 has a cut off part of that brace attached to a structure about 200 feet behind it.

View_1.png

View_2.png

FARO says this is a common problem that can occur with any other brand of scanner.

If you have any experience with this problem while using FARO or any other brand, please share it. I want to know if this is a FARO-only problem or is it only my unit's problem.

Thank you!

[Leandre Robitaille]

What software have you used to do the registration? If you look at the scan in that software does it have the same issue as in Naviswork? Did you try importing the original .fls from the scanner in another software than the one you used to do registration?

This is defenitivelly not a common error, I never saw this in any of my projects or projects my team did. At first glance it looks like and offset picture to the background due to the laser not capturing points there...but is it actually points that have been move back? Hence creating false recorded coordinates at another location...now thats an issue :/

[scottkim]

Same result for Scene, Cyclone and ReCap. So this is from the scanner.

[Tanguy Nédélec]

Not that I have a solution sadly, but I've seen this happen too, with a X130. Only noticed it on spheres however, i.e. the top points of the sphere are sometimes "projected" on the background, several meters away from where they should be.

[landmeterbeuckx]

FARO says this is a common problem that can occur with any other brand of scanner.

LOL, hooray for their support and disrespect for their clients (once again)

[smacl]

FARO says this is a common problem that can occur with any other brand of scanner.

While difficult to prove, I can't think of any reason to believe the above. Point cloud data is commonly stored in voxels, octrees or similar structures for reasons of speed and compactness, where data is grouped into cubes. A missing or misplaced bunch of data in cube shapes suggests to me an indexing error in this data structure. As such, it would be an instrument specific bug on the onboard or field processing software. I'd be interested in hearing FARO's reason to claim that this is not a problem specific to their product.

[yankoch]

Hi,

Are you using large coordinates in your project?

Can you please explain the main steps in your workflow (I mean: registration, export, conversion)?
I've been using Faro scanners for 10 years now and never had this issue, even with the S350 that I tested on several outdoor projects.

[flmnt]

We have also observed this problem with our X130, S350.

Seems to appear when there is a distance discontinuity. i.e. Edge of roof against distant background..
So not a problem with your particular scanner unit. More of a generic one with Faro scanners at least.

But is it a bug in their SDK, or in the way the scanner store the data?
I would guess that the scanner store a raw binary data stream that the SDK then voxelize (octree)

As for the project coordinates, doesn't seem to be the problem since the scanner store local data relative to the scanner itself along GPS metadata.
And the problem appears immediately on data import

[smacl]

But is it a bug in their SDK, or in the way the scanner store the data?
I would guess that the scanner store a raw binary data stream that the SDK then voxelize (octree)

Could well be a bug in the SDK or the scanner firmware. The last SDK we've got for use with SCC is FARO LS 7.1.1.81 x64 which shows a modified data of 16/12/2020. We've had one previous version 7 SDK and a version 6 prior to that. Could be worth running the data against systems built with current SDK versions to see if that is the issue. We can check here using SCC FLS import if it would be of benefit. So far as I know, Cloudcompare also uses this SDK and is freely available.

[scottkim]

First of all, I would like to thank everyone who posted a reply. And I would like to ask for the experiences of users who use other brands.

Again, this is the result of the scanner raw data, not the registration process.

View_5.png

This problem, as explained above, has been recognized in my experience only under certain conditions. (Outdoors, when there is a building wall or a large structure in the background other than the scan target. The above picture was scanned inside a pipe rack and the structure is in the background.)
In many cases, the cut object grows in a curved shape by 20 times or more depending on the distance, and is newly located in several places, such as around the wall of a building. Then I clear it from the registration software if I notice it. For a large project, it is a very difficult process and reliability issues arise about the results.

Regards,