The meshing process, when referencing depth maps + laser scans for generation, must somehow reference and prioritize the laser scan geometry, right?
Perhaps generating a point cloud from the mesh vertices would generate a different result than generating a point cloud directly from depth maps? Am I making sense here?
The reason I ask is because I run into situations where the mesh geometry is clean (e.g., single wall plane) and the point cloud is erroneous (e.g., multiple intersecting walls planes generated).
If I’m able to open my model in Meshlab (or similar) to convert the mesh to a point cloud, I’ll report back.