[OpenSCAD] Discuss manifoldness, co-incident faces edges etc

Michael Marx michael at marx.id.au
Sun Nov 17 04:29:22 EST 2019

> I frequently ran into non-manifold conditions, and also due to (I believe) the small number
problem, producing self intersections.


The problem is OpenSCAD snaps to a grid and then stores the result as a polygon soup. That can
create non-manifold geometry whenever there are very close vertices. It is a massive bug to
truncate vertex coordinates and not fix the mesh.


Agreed, however with floats it will still be a problem, perhaps lesser, without the grid.

The two-cube-edge will be a problem regardless of grids/floats.




From: Discuss [mailto:discuss-bounces at lists.openscad.org] On Behalf Of nop head
Sent: Sun, 17 Nov 2019 19:26
To: OpenSCAD general discussion
Subject: Re: [OpenSCAD] Discuss manifoldness, co-incident faces edges etc



On Sun, 17 Nov 2019 at 07:29, <arnholm at arnholm.org> wrote:

On 2019-11-17 00:33, Michael Marx wrote:
> As Doug said 'STL can't represent curved surfaces, it can only
> represent a polyhedral approximation'.

That's true of anything that comes out of OpenSCAD or any other mesh 
based modeller, it is nothing particular to STL, OBJ, OFF, AMF or 
whatever format you choose. A mesh representation is a finite 
approximation, of course it is.

It is not the file format, is the core modeller.

If you took the CGAL library and replaced it with another library based 
on CAD BREP style datastructure, for example OpenCASCADE 
https://dev.opencascade.org/ the situation would be different, and you 
could output STEP files or other formats with accurate respresentation 
of curved geometry. But as long as the core model is a mesh 
representation, the output will be a polyhedral approximation.

Carsten Arnholm

OpenSCAD mailing list
Discuss at lists.openscad.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openscad.org/pipermail/discuss_lists.openscad.org/attachments/20191117/aa4e2f6a/attachment.html>

More information about the Discuss mailing list