Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.
I'm working on my dissertation project for university. I'm currently trying to create a mesh over point cloud data but i'm struggling to either create an accurate mesh or even create a mesh. My current laptop is just basic so wondered if that would have an impact on it? however, I will have access to a top of the range computer for processing soon!
I was hoping to be given some advice as to what are the best settings are for an accurate Poisson reconstruction (Octree Depth, Solver Divide, Samples per node & surface offsetting). If it helps, I'm currently processing a point cloud of a hillside that is prone to landslides. I'm hoping to create a mesh that will show the landslides.
Any help will be seriously appreciated!
In my experience, it depends of what degree of detail you are looking for.
I assume that you have a pulse-time scanner, and maybe some millions points dataset.
For very detailed mesh, the maximum "fidelity" I got is Octree 12 Solver 10. 10/8 brings good results, too. I'm meshing samples with 10+ million points, and patience ;-D. (OpenSUSE 12.2 + 16Gb RAM + Dell Precision M6400)
I've not found useful mesh with less than octree 8.(heritage)
Be aware that Poisson is an interpolation method, so if you have areas with low density of points or no points at all, it will create mesh anyway, so you have to verify the quality of the result. I use CloudCompare for that.
Hope this can help you.
I have a question: How can I create DSM from DTM? How can I filter out automatically the points of the buildings?