pointcloud to a Mesh, converting a pointcloud to a usable mesh Options

[JohnVV]

First ,I am by no means an expert on this so....

MeshLab is designed for working with pointclouds
from optical images or a laser scanner or a lidar or any tool that outputs a pointcloud
including ASP ( Ames Sterio Pipline) and "bundler"

Bundler is what i used for the 67p giff to pointcloud
-- see the 67p thread --
post 212
http://www.unmannedspaceflight.com/index.p...st&p=212227


the software is here

http://www.cs.cornell.edu/~snavely/bundler/
https://github.com/snavely/bundler_sfm
( there are a few different forks of this code including "OSM-Bundler" )

bundler outputs a "bundle.out" file
there IS a Blender import tool for them "import_bundler.py"
http://koelzk.crispy-cow.de/article/9/impo...enes-to-blender

remove everything but the pointcloud from the scene

----------- on to the pointcloud ----------

a view of a " just ok " PC of Comet 67P in Meshlab


http://meshlab.sourceforge.net/
The current version is 1.3.3
now this program is in rather slow development so most of the online guides are for 1.3.2 or 1.2
http://sourceforge.net/projects/meshlab/files/meshlab/
and some of the tools and menu items HAVE changed

unfortunately there is really nothing to see in meshlab for adding a "skin" to a PointCloud ( PC)
and they are a TON of tools and options and every mesh is going to be a bit different
a very orderly one that is GRIDDED DATA will need different settings that a disorderly one like the above image of 67p

see google , this is a good but brief post
http://gmv.cast.uark.edu/scanning/point-cl...esh-in-meshlab/

and this one from "makerbot" and 3d printing
http://en.wikiversity.org/wiki/MakerBot/Cl...For_3D_Printing

step one
import the mesh into meshlab
( file / import mesh )

step two
add normals
( "filters / Normals , curvatures and orientation / Compute normals for pointcloud" )

Step three
add a skin
--- there A LOT OF OPTIONS !!! --
just take a look at
( Filters / remeshing, simplification and reconstruction )
i used a poison noise option
- surface reconstruction : Poisson
( Filters / remeshing, simplification and reconstruction /surface reconstruction : Poisson )
-- octave tree 10
-- solver divide 8
-- surface offset 0.8


for other types of PC's it would be a good idea to read through some google search results
the number of options in MeshLab is rather DAUNTING

[Malmer]

poisson

I find the poisson reconstruction to be almost useless when handling a sphere-like surface. It is intended for surfaces that are mostly flat-ish.

My pointclouds are very dense, many millions of points, so I had to find a fast efficient algorithm. The VCG surface reconstruction works well for me. It is intuitive and fast.