Align 2 clouds failed when exporting

[stefano.ponti]

Dear Daniel and All,
I am struggling a lot to solve the following issue. I need to co-register 2 clouds deriving from Metashape because they differ of about few meters (different GPS used to take marker coordinates). I save the clouds in LAS and import them in CC 2.14: one as reference, the other to align on the previous. No matter if I skip (not apply) the global shift transformation at the importation or apply it, when I export the aligned cloud into Metashape again to build the DEM, it does not match the DEM of the reference cloud into ArcGIS. I tried to:

a) not apply the global shif --> difficult to properly see the points in a UTM coordinte system and I obtained a squeezed cloud in metashape
b) apply the shift only on the reference cloud --> no aligment in ArcGIS
c) apply the shift only on the align cloud --> no aligment in ArcGIS
d) apply the shift on both --> no aligment in ArcGIS

Of course I also tried to reapply the CC shift when importing the aligned cloud in Metashape with all the above combinations, nothing changes.
I realy don't know what I am missing.

Thank you very much for your interest,
Stefano

[daniel]

And what happens when you reopen these files in CloudCompare?

Normally, you should always apply the Global Shift(s) so as to not lose any information/precision. And then the registration tools should automatically apply the Global Shift of the reference entity to the aligned one.

[stefano.ponti]

Hi Daniel,
thank you for your suggestion! The registered clouds, once imported in CC, are correctly aligned. It still is strange that, for a few meters of GCPs error, CC suggests a shift of thousands of meters for one cloud and only 1000 meters for the second. While exporting them from CC I also tried many combinations: "use current global shift", "original LAS offset" and custom shift that is the difference of the previous two. The resulting DEMs are always not overlapping into GIS.
I'm sorry,
Stefano

[daniel]

If the clouds are properly positioned when re-importing them in CC, then doesn't it mean that the coordinates and LAS offset are properly defined? And that the GIS tool fails to restore the global coordinates? Or did I miss something?

[stefano.ponti]

Hi Daniel,
apparently I found a solution. The problem is the LAS format. I exported everything in PLY and, once imported into CC, I noticed that the global shift applied was the same. After all the alignment and transformations, GIS properly visualized the overlapping coulds. So I would recommend to everybody to use PLY format instead of LAS when postprocessing in ArcGIS.
Thank you,
Stefano