CloudCompare forum

Hi everyone,

I have been using the command line mode to register point clouds and had pretty good success thus far. When registering high density point cloud to low density point cloud, it took a while to calculate the matrix but everything prior worked fairly fast. However for this next stage of processing I am trying to register low density point cloud to high density point cloud. I'm locked up at the octree step. It's still using CPU in the Task Manager, but it's not responding. I haven't encountered a problem like this previously. Usually when CC isn't responding in Task Manager it's still working and it will come back after a few minutes. In this case it's been at 0% for over 12 hours.

Does anyone know what's going on here? I included the command I ran below and a screenshot.

CloudCompare -C_EXPORT_FMT LAS -O G:\gears\scratch\thartsook\als_test_registration\reproject_als_postfire_output\plot_clips\P401_ALS.las -O G:\gears\scratch\thartsook\als_test_registration\registration_outputs\prefire_tls\las\P401_E1_REGISTERED_2019-10-27_02h40_53_398.las -ICP -MIN_ERROR_DIFF 1e-8 -ADJUST_SCALE -RANDOM_SAMPLING_LIMIT 1000000000

According to the top message in the Console, it seems you have dealt with the large coordinates (you should see the -GLOBAL_SHIFT option). The gridding effect that the loss of accuracy could also create some strange ineffectiveness if it's "aligned" with the octree?

Hi Daniel,

I haven't been using the global shift option because the loss in accuracy is not a big problem to me. My working solution for now is to register a smaller subset and then apply that matrix to the entire cloud, as you mentioned in the wiki.