CloudCompare forum

I am experiencing a crash when I try to "merge" 21 small scans in an imported .E57 file. The total file size for the .E57 is 2.5GB.

The crash occurs in the latest beta 2.13, the currently installed 2.12.4 (flatpak) and the older 2.11 (.deb i think?) which is in the Kubuntu software repo/Discover.

To reproduce, I simply select all scans, click "merge" and when prompted to regen scalar fields I select yes or no (makes no difference) and then the program immediately closes.

Launching CC from the console allows me to capture the below error when it crashes

Details of my system:

Hum, it's strange that you run into an 'assert' statement as it should only be active in Debug mode (I hope these versions are not compiled in Debug mode as it's super slow!).

Nevertheless, around this line number (which is not exactly the one currently used in the latest 2.13.beta?), I see it's either due to scalar fields or scan grids.

Can you maybe try to remove all the scalar fields (select all the clouds before merging, and use 'Edit > Scalar fields > Delete all'). And see if it still crashes?

And if yes, can you select all the clouds and use 'Edit > Grid > Delete scan grids' and try to merge again?

I'm not sure how these were compiled but I have no issues with performance, in fact I think it's faster than my work PC on windows.

I can confirm it still crashed (with the same error statement) if I delete all scalar fields before merging. I am not familiar with scan grids and have never used this feature before but I noticed the option is greyed out. Both options "Delete Scan Grids" and "Mesh Scan Grids" cannot be selected.

I found the crash unsurprisingly also occurs while attempting to merge the clouds with the command line interface.

However I manage to avoid the crash by sub-sampling each cloud, setting the method to "Random" and just removing 1 point from each cloud. They then merged without issue.

I'm not sure what this suggests, perhaps something with the e57 format created by my registration software (Leica Register 360)?
Saving the sub-sampled scans back out to .e57 with only one point removed somehow reduces the files size by over 25MB on a 120MB cloud...

Happy to share with you some sample scan data if it helps.

Do you reproduce the issue on Windows? Or is it only on Linux? (I must admit that the fact that it works after removing only one point is really disturbing).

If it crashes on Windows as well, I'll definitely be interested by having a look at the corresponding E57 files. Otherwise, I won't be able to help as I'm a pure Windows dev (sorry ;). But there are some Linux experts (especially on the project's github page).

Now that I'm back in front of a windows PC I have tested on Windows 10 and can confirm the crash seems to be on linux only. I will hop over to Github and make a post.