Left, reconstruction in Geomagic without smoothing, middle, after performing 300 iterations of smoothing on the point-cloud, right after performing 150 iterations on the mesh produced in the middle.
All can be done using this software.
Perform Multi GPU EGT smoothing On a Point Cloud or a Mesh
This is the code that perform multi-gpu smoothing on a PointCloud and also Normal Computation.
This code is based on the paper:
Alexander Agathos, Philip Azariadis,
Multi-GPU 3D k-nearest neighbors computation with application to ICP, point cloud smoothing and normals computation,
Parallel Computing,
Volume 121,
2024,
https://doi.org/10.1016/j.parco.2024.103093.
The license is MIT so you can use it for Academic and Commercial purposes.
Still I am perfecting it, the normal computation is still under development and it needs a lot of polishing up.
As with all open source software you may use it at your own risk.
It runs as:
KNNCUDA "file.ply" "integer:number of neighbors" "integer:number of Gabriel iterations" "float:alpha-value" "integer:0 if point-cloud, 1 if mesh" "integer:number of gpu cards"
Example:
For point cloud:
./KNNCUDA bunny.ply 40 150 0.65 0 1
For a mesh:
./KNNCUDA bunny.ply 40 150 1e-8 1 1
Copyringht Dr. Alexander Agathos.