Align laser scanner cloud on georeferenced cloud. Global shift problems

[Dann]

Hello everyone, I'm new here. I ask you for help hoping to solve a problem that I have in cloudcompare for a few days. I have two terrestrial laser scanner files from the same area but captured at different times. One of the files is geotagged correctly and one is not. My goal is to be able to superimpose the file with wrong georeferencing on the one that is correctly georeferenced to georeference them both but this causes me problems with the global shift. When I import the .e57 files, cloudcompare suggests a transformation. The suggested transformation is not the same for the two files, (-498600.00; -4184100.00; 00) for the correctly georeferenced file and (-1027000.00; -4201000.00) for the other file. If my workflow is correct, to achieve my goal I need to use the same transformation values ​​for both files, then proceed with the alignment of the two clouds and once aligned correctly save. The problem arises when I use the same values ​​for the global shift, the cloud on which I apply the transformation other than the one suggested loses quality, gets damaged, showing the non-uniform data, arranged along strips. Do you have any idea how I can solve the problem? I also tried to import the badly georeferenced file, disable "preserve global shift on save" on the global shift / scale window, save in .e57, import into a new project, align with the georeferenced cloud and check if it solves the problem but when I try to save gives me an error ([10:44:01] An error occurred while saving 'File name': the third-party library in charge of saving / loading the file has thrown an exception).
Thanks to everyone, I love cloudcompare but this time I'm not being able to bypass the problem.

[daniel]

So actually, if you are going to align the two clouds eventually, then you don't care about the Global Shift that is applied to the to-be-aligned cloud (the one that is not correctly geotagged). Its Global Shift will be overwritten by the good one when the Align tool will finish its work.

So you should follow CloudCompare advice and use 2 different shifts.

[Dann]

Maybe I'm wrong but I noticed that if I perform the ICP alignment with different global shift, when I close everything and reopen the files are not in the same position. I noticed that to be in the same position they must have the same global displacement at the time of alignment and if I force the same global displacement one of the two clouds is ruined. However, now I have solved it by making a point-to-point alignment in order to overlap the two clouds, save, close and this time I can reopen the clouds with the same global displacement value without damaging them and finally I apply the ICP fine registration.

[daniel]

Ah yes, sorry, I was referring to the Alignment (point-to-point) alignment tool.

Looking at the code of the ICP/Fine Registration tool, I see that normally it also tries to synchronize the Global Shift information:
- if the reference cloud has a Global Shift set, then it's transferred to the aligned cloud (and you should see a message in the console that says '[ICP] Aligned entity global shift has been updated to match the reference...'.
- otherwise if the reference cloud has no Global Shift set, and the aligned cloud has, the CC asks the question whether the aligned cloud Global Shift should be dropped

I would be curious to know if any of the messages appears in the Console? And at the end of the registration process, what the 'Global Shift' are for the two clouds (it appears in the cloud properties).