Definition Low

[realjohn]

Hie,
I am novice with the program and i have some problems with the definition:
- I have a cloud point (CC_1 Image).
- I use PCL Wrapper to estimate normal.
- I use Poisson Recon to define the mesh (CC_2 Image)
- In the properties i use FSF display to select (CC_3 Image)
- I use scalar fields/filter values to export (CC_4 Image)
The problem is that i start with a point cloud with high definition but at the end i get a mesh in low resolution and with empty spaces. I lost much information in the process and it looks like that there is a los of points that cloud compare not use (CC_4 Image).
I am sure that the problem is not the software, and i think that the point cloud hace enough definition then i suppose that the problems is the parameters of the operations.
Could anybody help me please?
Regards.

CC_1.JPG (189.91 KiB) Viewed 589 times

CC_2.JPG

CC_3.JPG

CC_4.JPG

[daniel]

I think the biggest issue is probably with the normals. You need properly oriented normals (i.e. all in the right direction). And that's something extremely hard to obtain on such a cloud. You may want to try with the 'Edit > Normals > Compute' method, and play with the 'preferred orientation' parameter. Maybe split the cloud in several parts to do it (and eventually merge all the parts together back again).

Once done, you can then increase the octree depth for the Poisson Surface Reconstruction algorithm to get a finer mesh. But without good normals, it's useless...

[realjohn]

Thanks a lot Daniel, I am going try to solution and comment it.
Regards