- Structure from motion with visualsfm
- 3D surface model reconstruction
- File/ Open Project
- File/ Import Mesh
- Clear the vertices with the select and delete vertices.
- Filter / Point Set / Surface Reconstruction: Poisson
- Filter / Selection / Select faces with edges longer than
- Filters –> Selection –> Select Non-Manifold edges
- Filter –> Texture –> Parameterization from registered rasters
- Filter –> Texture –> Project active rasters color to current mesh, filling the texture
- Export Mesh as…
多一个想法和支持,心里就不会觉得绝望
Structure from motion with visualsfm
you can also find a 64-bit windows version with cuda as well as the pmvs with the code of uaxa
smart3D with the code of 7xrb
A sparse point cloud model can be generated by the visualsfm. Furthermore, a dense point cloud model can be genearted with the pmvs or cmvs program. However, these point cloud model are still less beautiful than the 3D surface model.
To this end, I will list some steps for construct the surface model with the results of the visualsfm and the dense point cloud model.
3D surface model reconstruction
File/ Open Project
choose the bundle.rd.out file first and choose the list.txt file.
File/ Import Mesh
choose the dense point cloud model .ply file
Clear the vertices with the select and delete vertices.
![]() |
Filter / Point Set / Surface Reconstruction: Poisson
with the octree Depth with 10 or larger.
Filter / Selection / Select faces with edges longer than
![]() |
Filters –> Selection –> Select Non-Manifold edges
delete the non-manifold edges
Filter –> Texture –> Parameterization from registered rasters
Filter –> Texture –> Project active rasters color to current mesh, filling the texture
Texture projection. Set the pixel size of the texture image, the larger the clearer.
we can set the value as 512, 1024, 2048, …
Export Mesh as…
![]() |
That’s all.