Page 1 of 1
Point cloude
Posted: Tue Mar 02, 2021 10:16 pm
by Gunn
Hello
I have a point cloude generated from pix4d. It is an fotegrammerty consisting og rufly 1000 picture. Of this I got 768 000 000 points that I segmented a little part aout of. So I only have around 180 000 000 points to work with.
My problem is that the point cloud is cut up in many planes insted of beeing a solid 3D model showing all of the points.
Is there any tric to fix this?
Re: Point cloude
Posted: Tue Mar 02, 2021 11:09 pm
by WargodHernandez
What format are you actually loading?
Are you exporting these in .LAS/GeoTiff from pix4d?
Re: Point cloude
Posted: Wed Mar 03, 2021 11:13 pm
by daniel
And it looks like an 'accuracy loss' issue. Have you accepted the 'Global Shift' as suggested by CloudCompare when loading the cloud?
Re: Point cloude
Posted: Thu Mar 04, 2021 10:09 am
by Gunn
WargodHernandez wrote: ↑Tue Mar 02, 2021 11:09 pm
What format are you actually loading?
Are you exporting these in .LAS/GeoTiff from pix4d?
I am loading <<las>> file and when Global shift appear I klick <<yes for all>>.
Right now I am running pix4d over again but with using; Image scale: <<1/2 (Half image size, Default)>>, Point Density: <<Optimal>> and Minimum Number of matches: <<5>>.
This will hopfefoully reduce the point clod and make a nicer surface. The probelm with sliched point clod maby dissapear after this process, if not I ask for advise again :)
Re: Point cloude
Posted: Thu Mar 04, 2021 10:26 am
by Gunn
Hello I use las format that is generated from pix4d.
Re: Point cloude
Posted: Thu Mar 04, 2021 3:10 pm
by daniel
And in the "Global Shift" dialog, are there any values in red maybe? (i.e. large "local" coordinates)
Re: Point cloude
Posted: Thu Mar 04, 2021 4:08 pm
by Gunn
Hello
Just going to tell that now everything is fine after processing the point cloud once more with new settings that gave me a mutch more nicer point cloud wihtout all the noice and less point then the previous one.
There are now no sliching of the point cloud so I am happy for that.
Thank you all for trying helping mee :)