CGAL users discussion list

[Laurent Saboret <>]

Charlie,

Yes, the demo contains the surface reconstruction example's code, plus code to load a point set without normals and compute them, plus opengl rendering to see the result.

The surface reconstruction package produces a polyhedron, whose triangles are guaranteed to be correctly oriented. You may have a bug in your X3D output.

Best regards,
Laurent Saboret


Le 20/02/2012 08:10, ckehoe a écrit :

Thanks Laurent, I'll take a look at the demo. Is the code basically the same as the surface reconstruction example?

I looked at my output in a 3D viewer, and I think I'm having a problem with the triangle orientations. I want the surface to be transparent, but this doesn't appear to work unless all the triangles are also single-sided. When I turn this on as well, some triangles appear to be lit from the back, while their neighbors are lit from the front. Multiple viewers produced this same effect.

Aren't all the vertices supposed to appear in clockwise order from one side of the surface, and counter-clockwise order from the other? Is there some flag I have to set to ensure this?

Again, here is my CGAL code. I can post the code that outputs to X3D, and the X3D output as well, if needed. But I wanted to check if I'm calling CGAL correctly.

http://pastebin.com/WThHKx6p

On Fri, Feb 17, 2012 at 2:21 AM, Laurent Saboret-3 [via cgal-discuss] <[hidden email]> wrote:

Hi Charlie,

Did you try the Surface Reconstruction demo? It is a GUI application (built over Qt) that lets you load a point set, estimate and orient normals, then reconstruct a surface from it.

If you have a Windows PC, a precompiled version is available at http://www.cgal.org/demo/3.9/surface_reconstruction_points_3.zip
On other platforms, you will have to compile it with cmake. It is located in .../CGAL/demo/Surface_reconstruction_points_3/ folder.

Best regards,
Laurent Saboret


Le 16/02/2012 13:30, Charlie Kehoe a écrit :

On Wed, Feb 15, 2012 at 11:10 PM, Sebastien Loriot (GeometryFactory) <[hidden email]> wrote:

1. I don't have normals for my points. My surface is pretty smooth, so
normal estimation should work, but I couldn't find the documentation that
actually described how to do this. Can anyone point me in the right
direction? 

See this package:
http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Point_set_processing_3/Chapter_main.html

Sebastien.

 

Thanks Alex and Sebastien for your help. I stitched the normal-estimation example together with the surface reconstruction one, but got a sphere with a pretty pock-marked surface. It looks as if some of the normals are getting passed through wrong. Here's my code:

http://pastebin.com/WThHKx6p

I'm happy to pass along the output vrml as well, but of course it's sizable. I also tried fiddling with the number of nearest neighbors to use. If I make it smaller, it either can't solve the poisson equation, or gives *really* weird output. If I make it bigger, the surface is smoother, but still has lots of artifacts. Is this just a weakness of the normal estimation algorithm? Shouldn't fitting planes to the neighbors produce a pretty smooth sphere?

- Charlie

---

If you reply to this email, your message will be added to the discussion below:

http://cgal-discuss.949826.n4.nabble.com/Surface-reconstruction-without-normals-tp4392786p4396853.html

To unsubscribe from Surface reconstruction without normals, click here.
NAML

---

View this message in context: Re: Surface reconstruction without normals
Sent from the cgal-discuss mailing list archive at Nabble.com.