TLS Laserscans - Registration incorrect

[Dieter]

Dear Agisoft team, I have now tried the procedure with TLS scans and photos several times and each time I get a different result. There is now version 18, no one knows which script is correct or which one to use, no one knows exactly how I can add registered scans to an existing project without destroying the registration, all in all I can say for myself that the function "Reading in registered laser scans" DOES NOT WORK CORRECTLY.

I ask again to firmly integrate all these scripts into Metashape to make the whole thing a little more transparent.

There are now many entries here in the forum where almost all users say it doesn't work, how will the Agisoft team react to this?

For me and our office, I can say that we urgently need this function.

dieter

[Dieter]

Found in agisoft freshdesk:

"In case the laser scanning locations are considered as precise and you would like to avoid image matching procedure between the TLS data and digital images, it is possible to apply the exterior orientation data from the Reference pane to the loaded laser scanning data using the following script:

https://github.com/agisoft-llc/metashape-scripts/blob/master/src/quick_layout.py

If the project contains also the common digital images to be aligned, it is recommended to align such image set first and then load the TLS data to the same chunk and run the mentioned script. Note that the digital images alignment should be referenced in the same coordinate space as laser scanning data in order to be properly processed together."


If I follow this workflow, Metashape destroys my entire registration of the photos via GCP.


I always have to import the scans first, then the photos, then mesh the photos with the scans, then set control points for the photos, then run the script.


And with a lot of luck it will work.



My question about this: Has a user ever brought this workflow to work?

[Alexey Pasumansky]

Hello Dieter,

In order to analyze the problem on our side, can you please provide the detailed instruction (describing the steps that you follow), so that we can try to reproduce it on our side on our sample data (TLS scans and overlapping images for the same scene)?


If there are also any screenshots that may illustrate the observed problems, please send them to support@agisoft.com or attach to the reply (providing that there's no sensible information).

[Dieter]

Hi Alexey,

i proceed as follows:

Editing a photogrammetry with coordinate systems. This includes aligning the images, optimizing the camera alignment and checking a control point.

This project is then fully balanced, calculated and assigned the appropriate coordinate systems.

In addition, I read in a scan with the coordinates of the same coordinate system of the images. There are no identical Points between the Pictures and the Scan!!

Then I assign accuracies for the coordinates of the scan of 0.001 meters. This means that the scan should be included in the calculation unchanged. Then I run the script. The same coordinates now appear in the calculated values ​​as in the original data. The errors contain 0,0,0 as the deviation between the original data and the calculated values. That's all right and good.
So far everything is going as it should - apparently (see Picture 1).

However, if I now create a dense point cloud, generate a mesh or calculate anything, the entire coordinate reference is completely destroyed. Updating the transformation also results in completely implausible values (see Picture 2).

Image 1 is after applying the script, all coordinates are plausible and correct.

Image 2 is after pressing "updated  transformation", then nothing is correct anymore.

[Dieter]

Any news here?

[Dieter]

Again, any news here?

[Alexey Pasumansky]

Hello Dieter,

In 1.8.4 release (published yesterday) when you add pre-processed laser scans from TIFF files (all at once) they will be added as multi-camera system, but during the alignment the relative orientation of the sensors in the system will be preserved (just to not enable adjust rotation/location for the slave cameras in the Camera Calibration dialog).

This approach is an intermediate solution, as in future major Metashape updates the concept of the laser scans will be revised. But I hope that it could at least partially work for your needs.

[Dieter]

 

[bassistas]

Hello Dieter,

In 1.8.4 release (published yesterday) when you add pre-processed laser scans from TIFF files (all at once) they will be added as multi-camera system, but during the alignment the relative orientation of the sensors in the system will be preserved (just to not enable adjust rotation/location for the slave cameras in the Camera Calibration dialog).

This approach is an intermediate solution, as in future major Metashape updates the concept of the laser scans will be revised. But I hope that it could at least partially work for your needs.

I've tried this method and I get a successful alignment but the orientation of the model is wrong. How Can I keep the orientation of the laser scans? (rotation/location for the slave cameras is unchecked)

[Dieter]

You must define the tls as fixed.

To do this, click the checkboxes for position and rotation (only tls) before the camera orientation.

[bassistas]

You must define the tls as fixed.

To do this, click the checkboxes for position and rotation (only tls) before the camera orientation.

Where do I find these checkboxes? Also do I need to run the quicklayout script first?

[Dieter]

In the reference tab you only have to click the top two boxes for the TLS, the box on the left is for position, the box on the right is for rotation. As a prerequisite, you must have read in coordinates for your Tls. The script doesn't run in the latest version anyway, so it should work without it. This is how it worked for me.

Much luck