We closed this forum 18 June 2010. It has served us well since 2005 as the ALPHA forum did before it from 2002 to 2005. New discussions are ongoing at the new URL http://forum.processing.org. You'll need to sign up and get a new user account. We're sorry about that inconvenience, but we think it's better in the long run. The content on this forum will remain online.
IndexProgramming Questions & HelpOpenGL and 3D Libraries › 3D mesh for point cloud
Page Index Toggle Pages: 1
3D mesh for point cloud? (Read 7379 times)
3D mesh for point cloud?
May 22nd, 2010, 2:27pm
 
Hi everyone,

I've been looking for awhile to find a convenient way to render surfaces to the 'outside' of a point cloud, much like these do.

http://meshlab.sourceforge.net/
http://blenderartists.org/forum/showthread.php?t=144504

But have yet to find it. Can someone help?
Re: 3D mesh for point cloud?
Reply #1 - May 22nd, 2010, 3:30pm
 
look for delaunay triangulation. also there is a Mesh lib for processing that includes that and might help you out

http://www.leebyron.com/else/mesh/#delaunay
Re: 3D mesh for point cloud?
Reply #2 - May 22nd, 2010, 4:31pm
 
Thanks, I've looked at it, and the triangulation hack, but can't get it to work in three dimensions..
Re: 3D mesh for point cloud?
Reply #3 - May 22nd, 2010, 5:56pm
 
ah yes didnt remember it was 2d only.
well you will have to go for your own implementation.

a quick search for "delaunay triangulation" gave me this

http://www.cse.unsw.edu.au/~lambert/java/3d/delaunay.html
Re: 3D mesh for point cloud?
Reply #4 - May 23rd, 2010, 3:51am
 
Thanks for the link, sounds nice.

I'll be happy to make my own implementation, but though the "gift wrapper" algorithm comes close, it's still a convex hull type, which only gets the most extreme of points and fills in the surfaces in between. It doesn't quite allow for inward dents in the surface.

IS there really no 3D mesh algorithm available for processing?! It seems like the most basic way to render a 3D model..?
Re: 3D mesh for point cloud?
Reply #5 - May 23rd, 2010, 5:57am
 
well, sorry,, i don't really know. Maybe you can find something in Java that is easy to use within processing.
Otherwise you will have to write your own.

have you thought about using Marching cubes and generate the "shell" out of your point cloud?

an ideia: Meshlab is open-source, check out it's methods and write your own.
Re: 3D mesh for point cloud?
Reply #6 - May 23rd, 2010, 9:01am
 
Ah well it seems like toxi has nailed the subject pretty well in his libs, so I'll just be extracting my simple needs out of his examples. Indeed, like you mentioned, probably marching cubes stuff. His work is IMPRESSIVE by the way. Love the MRI example.
Re: 3D mesh for point cloud?
Reply #7 - May 23rd, 2010, 9:07am
 
i would be interested in your results as this is a task i will face in one of my upcoming projects.
How many points do you have?
Reply #8 - May 23rd, 2010, 10:02am
 

How many points do you have in your cloud?

Can you post them?

What is your goal? Just display them orderly?

Where are they from? Can't you get back to the source and get the data-set of the connecting lines to?

Cheers!

Chris  Wink

Re: 3D mesh for point cloud?
Reply #9 - May 24th, 2010, 2:07pm
 
Aw crap. Still haven't succeeded..  Angry

edit: Brilliant emoticon, fits my expression perfectly.
Re: 3D mesh for point cloud?
Reply #10 - May 24th, 2010, 7:30pm
 
look what i found: http://www.heikman.de/index.html/tools/java/triangulation.html
doesn't look perfect but might be a good starting point if you don't have one already
Re: 3D mesh for point cloud?
Reply #11 - May 25th, 2010, 4:14am
 
Niiice.

Looks very nice indeed. Will take a look tonight!
Re: 3D mesh for point cloud?
Reply #12 - May 26th, 2010, 11:28am
 
One of the algorithms used in meshlab for reconstruction is the Poisson Reconstruction. You could find it here: http://www.cs.jhu.edu/~misha/Code/PoissonRecon/ (code and executable are provided in c++). For my experience, to get good smooth surfaces, Poisson Reconstruction is the most effective and is good at preserving details.

Delaunay triangulation should be slower and might be problematic for a large number of points.

Marching Cubes implementation seems to be a long standing request for processing.

Page Index Toggle Pages: 1