On May 18, 2015, at 06:51 AM, kaefert@gmail.com wrote:

The inner end of your screwprop isn’t closed properly; you have some reversed polygons.
If you put an exclamation mark in front of an object, you can render that object alone, e.g.:

!screwprop(65, mitte_innen_oben/2, 4, 10);

If you then turn on Throwntogether mode (F12), you can see reversed polygons colored magenta.

-Marius

Hi there Marius!

Thanks for your response! Okey, I've actually removed this part as
part of my simplifications of the original design since I thought it
would be unnecessary because this side is never visible because of the
implicit union with the central cylinder. But fixing this doesn't
solve my problem.

I've attached another version where you can't find any magenta
polygons when pressing F12 but still get this error. Any more ideas?

Thanks, Kind regards,
Thomas K.

2015-05-18 17:43 GMT+02:00 Marius Kintel marius@kintel.net:

On 05/18/2015 12:19 PM, kaefert@gmail.com wrote:

According to Meshlab, the blades are inside-out: the surface normals
point the wrong way. Selecting non-manifold edges lights up the blades.

Perhaps a polygon winding order error?

--
Ed
softsolder.com

Hi there Ed! Thanks for your reply!

2015-05-18 20:20 GMT+02:00 Ed Nisley ed.nisley@pobox.com:

I apologize for my ignorance, but sadly I don't know if you mean by
that, that there is something wrong in my code, or something wrong in
the way OpenSCAD renders my code.
Please elaborate.

Thanks, Regards,
Thomas K.

On May 18, 2015, at 12:19 PM, kaefert@gmail.com wrote:

I think it’s because your propeller blades touch part of the surface of the hub. This could create non-manifold geometry. In some cases, it’s just a warning since the non-manifoldness goes away when converting down to floating point coordinates, but our analysis is done with infinite resolution.
Analyzing and potentially repairing such issues after rendering is on the wishlist..

-Marius

On 05/18/2015 02:51 PM, kaefert@gmail.com wrote:

It's not exactly wrong, but the order in which you specify the points
of a polygon / polyhedron determine which way the surface normals point:

http://en.wikibooks.org/wiki/OpenSCAD_User_Manual/Primitive_Solids

"When looking at the face from the outside inwards, the points must be
clockwise. You can rearrange the order of the points or the order they
are referenced in each tuple. [...] Note that if your polygons are not
all oriented the same way OpenSCAD will either print an error or crash
completely"

Given the way that you're generating the points, at least some of the
polyhedrons may be wound backwards.

The reason this doesn't show up in thrown-together mode may be due to
the way the polyhedrons combine: everything looks fine from the outside,
but some inside (invisible) faces wind backwards. Perhaps assembling one
polyhedron at a time will be productive.

Using Meshlab (or something similar) to check your work may also help
pin this down.

It's a nice-looking prop, though!

--
Ed
softsolder.com

2015-05-18 22:30 GMT+02:00 Ed Nisley ed.nisley@pobox.com:

hmm, well the "if else if else" variant with 3 different polyhedrons
being drawn (1 for begin, 1 for middle 1 for end) does not have any
faces on the inside as far as I can tell.

Yes, I've found by trial and error that I can produce a STL that I can
slice into something useful with slic3r by opening the STL produced by
OpenSCAD in MeshLab and apply the Filter "Cleaning and Repairing" ->
"Remove Duplicate Faces". It still gives errors on some triangles
where the wings hit the center cylinder but who cares if I can slice
it ;)

I had to increasing the calculation resolutions of my loop from 15 to
30 (variables rad_res & chord_res) for those errors at the
intersection between wings and the center cylinder to be small enough
to not have a negative effect on the slicing process.

Yes! Thats what I thought as well ;) But any congratulations need to
go to user DMeyer from the thingiverse.com - he did the hard part of
designing that propeller wing.

Thanks & Regards,
Thomas K.