Re: Discuss Digest, Vol 38, Issue 22

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: Discuss Digest, Vol 38, Issue 22

jim_klessig
"Actually it is possible, but probably not inside OpenSCAD.  Many years ago I
was in the land surveying business and I wrote a routine to take randomly
spaced 3D points"

I would disagree.  Your routine would be making some (probably un stated) assumptions  about the shape. Such as that the points do not form a "cave", and probably what points represent a boundary.

For example how would your routine deal with the example I gave,  of 8 points representing the corners of a cube.
There are multiple "correct" representations of that set of points.

I would suspect that your routine would assume those points are either a cube shaped bump in a flat plane, or a cubical hole in an otherwise flat plane.  Both of which are making perfectly reasonable assumptions for making it topographical map.
But those assumptions would be/could be  incorrect for just that set of 8 points, "floating in space" with no other information.
Essentially there are too many degrees of freedom for the problem to be solvable as he posed it.

Sent from my U.S. Cellular® Smartphone

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Reply | Threaded
Open this post in threaded view
|

Re: Discuss Digest, Vol 38, Issue 22

jim_klessig
Also, for whatever reason, i can never access any of the pictures people attach here, so no, i can not see it.



Sent from my U.S. Cellular® Smartphone

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Reply | Threaded
Open this post in threaded view
|

Re: Discuss Digest, Vol 38, Issue 22

doug.moen
In reply to this post by jim_klessig
This idea has been discussed before. The CGAL "Alpha Shapes" module is one way to implement this. Note that we already use CGAL for boolean operations like union.


On 23 January 2018 at 13:09, jim.klessig <[hidden email]> wrote:
"Actually it is possible, but probably not inside OpenSCAD.  Many years ago I
was in the land surveying business and I wrote a routine to take randomly
spaced 3D points"

I would disagree.  Your routine would be making some (probably un stated) assumptions  about the shape. Such as that the points do not form a "cave", and probably what points represent a boundary.

For example how would your routine deal with the example I gave,  of 8 points representing the corners of a cube.
There are multiple "correct" representations of that set of points.

I would suspect that your routine would assume those points are either a cube shaped bump in a flat plane, or a cubical hole in an otherwise flat plane.  Both of which are making perfectly reasonable assumptions for making it topographical map.
But those assumptions would be/could be  incorrect for just that set of 8 points, "floating in space" with no other information.
Essentially there are too many degrees of freedom for the problem to be solvable as he posed it.

Sent from my U.S. Cellular® Smartphone

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org



_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Reply | Threaded
Open this post in threaded view
|

Re: Discuss Digest, Vol 38, Issue 22

cbernhardt
In reply to this post by jim_klessig
jim_klessig wrote

> I would disagree.  Your routine would be making some (probably un stated)
> assumptions  about the shape. Such as that the points do not form a
> "cave", and probably what points represent a boundary.
> For example how would your routine deal with the example I gave,  of 8
> points representing the corners of a cube.There are multiple "correct"
> representations of that set of points.
> I would suspect that your routine would assume those points are either a
> cube shaped bump in a flat plane, or a cubical hole in an otherwise flat
> plane.  Both of which are making perfectly reasonable assumptions for
> making it topographical map.But those assumptions would be/could be 
> incorrect for just that set of 8 points, "floating in space" with no other
> information.Essentially there are too many degrees of freedom for the
> problem to be solvable as he posed it.

You are correct.  My routine eliminates any duplicate XY points, so it would
not tessellate an exact cube or an overhang correctly; however, for purposes
of my headlight bezel, or 99% of our topo maps, it works fine.
Charles




--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Reply | Threaded
Open this post in threaded view
|

Re: Discuss Digest, Vol 38, Issue 22

NateTG
Sure, you can, for example, do a Delauney triangulation in the XY plane to
generate a mesh, and then do a quadratic or bezier fit.   That's technically
possible using OpenSCAD functions, but I would be looking at existing mesh
manipulation stuff like blender or meshlab.




--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org