complex cloud 2 mesh

Feel free to ask any question here
Post Reply
kro_pa
Posts: 5
Joined: Tue Mar 01, 2016 9:52 pm

complex cloud 2 mesh

Post by kro_pa »

Hello,
How to do mesh from *.las - big and complex point cloud? Poincloud is from cave - very complex wall's relief, c.a 80 mln points.
1) Cloud have normals (by triangulation), 2) PSR on lv 8 split cave for different parts (screen in attch.), 3) PSR on on level 17 is a one big green buble :-), 4) PSR on higher levels was never made - reconstruction is always failed.
What do I wrong? Thanks for advice in advance.
Attachments
cave_printsreen.jpg
cave_printsreen.jpg (76.95 KiB) Viewed 4642 times
mesh.jpg
mesh.jpg (179.86 KiB) Viewed 4642 times
daniel
Site Admin
Posts: 7713
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: complex cloud 2 mesh

Post by daniel »

Level 17 for PSR is already gigantic, so it's not a surprise that it fails above this level ;)

But the result is definitely not what I would have expected! Are the normals properly oriented? They seem to be a bit randomly oriented. How did you compute them exactly?
Daniel, CloudCompare admin
kro_pa
Posts: 5
Joined: Tue Mar 01, 2016 9:52 pm

Re: complex cloud 2 mesh

Post by kro_pa »

Thank you for reply.
I have just made Edit/normals/compute. Then I've choose 'triangulation' , rest I have left as default.
daniel
Site Admin
Posts: 7713
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: complex cloud 2 mesh

Post by daniel »

Ah ok, then it's probably the cause of the issue here. Only computing the normals (with triangulation or any other algorithm) won't be sufficient. You need to get well oriented normals (i.e. all normals pointing outside of the surface for instance), which is generally not the case at first as CC can't guess where is the inside and where is the outside. You'll have to play with the 'Edit > Normals > Orient' methods, but it can be very tricky on complex shapes such as the share of your cloud. If you struggle too much, you can send me your cloud, and I'll give it a try.
Daniel, CloudCompare admin
kro_pa
Posts: 5
Joined: Tue Mar 01, 2016 9:52 pm

Re: complex cloud 2 mesh

Post by kro_pa »

Thank you. 'Orient normals' options are not active, they are grey... ?
daniel
Site Admin
Posts: 7713
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: complex cloud 2 mesh

Post by daniel »

Have you selected the cloud with normals?
Daniel, CloudCompare admin
kro_pa
Posts: 5
Joined: Tue Mar 01, 2016 9:52 pm

Re: complex cloud 2 mesh

Post by kro_pa »

I did :-). After computing normals I have saved cloud as *.las again. Re-saving as *.las saves normals in cloud also or do I have to compute them again?
daniel
Site Admin
Posts: 7713
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: complex cloud 2 mesh

Post by daniel »

Sadly, LAS files don't store normals. You have to use either CloudCompare's BIN format, or the more generic PLY format.
Daniel, CloudCompare admin
paddywhells
Posts: 2
Joined: Tue Jun 04, 2019 5:59 pm

Re: complex cloud 2 mesh

Post by paddywhells »

I have a similar issue with a point cloud from within a cave. I'm having trouble orienting Normals effectively. I've tried Z, Barycenter, and spanning tree (knn=6) but get funky results. do you have any suggestions for a long narrow cave? Did a solution come out of this original post that I could use?

[/img]
daniel
Site Admin
Posts: 7713
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: complex cloud 2 mesh

Post by daniel »

Do you have a set of points describing the scanner path - in the middle of the cave / tunnels - maybe?
Daniel, CloudCompare admin
Post Reply