Ok, I guess I made multiple errors. For some reason I assumed that (if I input a curve or a line) I'd get the whole volume from the y axis out to the points. Obviously, what I actually get is just a shell of zero thickness for that curve or line. Is it better for me to get a full volume by adding one point to the polygon [0.0001, 0] and then if I want a non-zero thickness shell doing subtract with an offset of the solid. Or is it better for me to make the polygon with twice as many points (each point has a corresponding double that is offset by the thickness)? I ask because elsewhere it seemed like difference() {} might be something one wants to avoid in OpenSCAD for performance reasons. <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free. www.avast.com <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link> <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2> On Wed, Jul 17, 2019 at 11:19 AM Parkinbot <rudolf@digitaldocument.de> wrote: > So your problem is clear. You try to rotate around a single vertex and get > a > singularity. > > The following modification avoids this problem by introducing another point > on the y axis and creating an edge on the y axis. > > function coshPolygon(multiplier, power, domain, flip, steps) = > let (flipValue = flip ? cosh( ( multiplier *(pow(domain,power))) ) : 0) > let (result = [for(i = [0.0001: (domain / steps) : domain]) [i, > flipValue - cosh( ( multiplier *(pow(i,power))) )] ]) > concat(result, [[0, result[len(result)-1][1]]]); > > > > > -- > Sent from: http://forum.openscad.org/ > > _______________________________________________ > OpenSCAD mailing list > Discuss@lists.openscad.org > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >

You don't get a shell, you get a 3D solid who's cross section is the 2D shape. The problem is if that 2D shape only touches the Y axis at a single point then the 3D shape has a pinch point there where it has zero thickness in the Z direction. If you want to make a shell then the 2D shape should be the cross section of a shell and yes that is faster made in 2D before the extrude than in 3D with a differences after. On Wed, 17 Jul 2019 at 16:58, Dan Shriver <tabbydan@gmail.com> wrote: > Ok, I guess I made multiple errors. > For some reason I assumed that (if I input a curve or a line) I'd get the > whole volume from the y axis out to the points. > Obviously, what I actually get is just a shell of zero thickness for that > curve or line. > > Is it better for me to get a full volume by adding one point to the > polygon [0.0001, 0] and then if I want a non-zero thickness shell doing > subtract with an offset of the solid. Or is it better for me to make the > polygon with twice as many points (each point has a corresponding double > that is offset by the thickness)? I ask because elsewhere it seemed like > difference() {} might be something one wants to avoid in OpenSCAD for > performance reasons. > > > <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free. > www.avast.com > <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link> > <#m_2910737154312055855_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2> > > On Wed, Jul 17, 2019 at 11:19 AM Parkinbot <rudolf@digitaldocument.de> > wrote: > >> So your problem is clear. You try to rotate around a single vertex and >> get a >> singularity. >> >> The following modification avoids this problem by introducing another >> point >> on the y axis and creating an edge on the y axis. >> >> function coshPolygon(multiplier, power, domain, flip, steps) = >> let (flipValue = flip ? cosh( ( multiplier *(pow(domain,power))) ) : >> 0) >> let (result = [for(i = [0.0001: (domain / steps) : domain]) [i, >> flipValue - cosh( ( multiplier *(pow(i,power))) )] ]) >> concat(result, [[0, result[len(result)-1][1]]]); >> >> >> >> >> -- >> Sent from: http://forum.openscad.org/ >> >> _______________________________________________ >> OpenSCAD mailing list >> Discuss@lists.openscad.org >> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >> > _______________________________________________ > OpenSCAD mailing list > Discuss@lists.openscad.org > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >

On 2019-07-17 17:32, Bruno Boettcher wrote: > heh happily swimming in my ignorance and incompetence...... > still..... you mention a tool polyfix that repairs the stl?? > some pointer please? > thanks! > Bruno It is bundled with AngelCAD, to get a binary use the downloads link at https://arnholm.github.io/angelcad-docs/ Regards Carsten Arnholm

Nop_head: when I did my original hypDome() {code given earlier in the thread} it looked hollow but that could be some weird artifact. I am using the 2018 release of OpenSCAD (not the latest one) and I see this: [image: image.png] For a shell with thickness I simply make a new function that assigns coshPolygon() to a vector, then I make a second vector taking the elements of the first vector (in reverse order) and subtracting the "thickness" from the y value (which isn't exactly correct as it should be on an angle with both x & y values altered). On Wed, Jul 17, 2019 at 12:20 PM nop head <nop.head@gmail.com> wrote: > You don't get a shell, you get a 3D solid who's cross section is the 2D > shape. The problem is if that 2D shape only touches the Y axis at a single > point then the 3D shape has a pinch point there where it has zero thickness > in the Z direction. > > If you want to make a shell then the 2D shape should be the cross section > of a shell and yes that is faster made in 2D before the extrude than in 3D > with a differences after. > > On Wed, 17 Jul 2019 at 16:58, Dan Shriver <tabbydan@gmail.com> wrote: > >> Ok, I guess I made multiple errors. >> For some reason I assumed that (if I input a curve or a line) I'd get the >> whole volume from the y axis out to the points. >> Obviously, what I actually get is just a shell of zero thickness for that >> curve or line. >> >> Is it better for me to get a full volume by adding one point to the >> polygon [0.0001, 0] and then if I want a non-zero thickness shell doing >> subtract with an offset of the solid. Or is it better for me to make the >> polygon with twice as many points (each point has a corresponding double >> that is offset by the thickness)? I ask because elsewhere it seemed like >> difference() {} might be something one wants to avoid in OpenSCAD for >> performance reasons. >> >> >> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free. >> www.avast.com >> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link> >> <#m_8551169469257401297_m_2910737154312055855_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2> >> >> On Wed, Jul 17, 2019 at 11:19 AM Parkinbot <rudolf@digitaldocument.de> >> wrote: >> >>> So your problem is clear. You try to rotate around a single vertex and >>> get a >>> singularity. >>> >>> The following modification avoids this problem by introducing another >>> point >>> on the y axis and creating an edge on the y axis. >>> >>> function coshPolygon(multiplier, power, domain, flip, steps) = >>> let (flipValue = flip ? cosh( ( multiplier *(pow(domain,power))) ) : >>> 0) >>> let (result = [for(i = [0.0001: (domain / steps) : domain]) [i, >>> flipValue - cosh( ( multiplier *(pow(i,power))) )] ]) >>> concat(result, [[0, result[len(result)-1][1]]]); >>> >>> >>> >>> >>> -- >>> Sent from: http://forum.openscad.org/ >>> >>> _______________________________________________ >>> OpenSCAD mailing list >>> Discuss@lists.openscad.org >>> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>> >> _______________________________________________ >> OpenSCAD mailing list >> Discuss@lists.openscad.org >> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >> > _______________________________________________ > OpenSCAD mailing list > Discuss@lists.openscad.org > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >

Em qua, 17 de jul de 2019 às 15:00, Carsten Arnholm <arnholm@arnholm.org> escreveu: > Therefore, the problem seen in OpenSCAD is not really in rotate_extrude, > but how the model is interpreted when subject to union using CGAL. The > same union using Carve works fine as should be expected. > CGAL seems to be very strict: it doesn't do boolean operations with a non-manifold. Considering your experiment, I am led to believe that Carve is complacent with some non-manifoldness when doing boolean operations. To confirm it, I have examined the OFF file exported by OpenSCAD for your rotate_extrude() example (without the cube(), of course). The topological information in the OFF file indicates that there are 12 faces incident to vertex index 0. Those faces are the triangles pointing to [0,0,0] from the top (6) and from the bottom (6). So no neighborhood of the vertex 0 is homeomorphic to a disc in the plane. So, the topology represented in the OFF file generated by OpenSCAD is not of a manifold. Summarizing, OpenSCAD may generate and export non-manifold objects. CGAL doesn't accept to do boolean operations involving non-manifold. Carve seems to be complacent with some non-manifoldness even in boolean operations. It is worth to observe that the union of the cube to the rotate_extrude is still a non-manifold object.