Checking if two Placements are parallel and creating one orthogonal.

[edwilliams16]

Two circles worked - but two spheres created a tessellated body - like it used the regular importer despite the message.

Code: Select all

07:36:21  Processing : /Users/ed/Downloads/Two-Circles-Hull.csg
07:36:21  Using Alternate OpenSCAD Importer
07:36:21  Circle Shape : True
07:36:21  Circle Shape : True
07:36:21  End processing CSG file
07:37:01  Processing : /Users/ed/Downloads/Two-Spheres-Hull.csg
07:37:01  Using Alternate OpenSCAD Importer
07:37:01  End processing CSG file

So now you are working on two parallel cylinders? The skew case will certainly be more complicated...

[keithsloan52]

Two circles worked - but two spheres created a tessellated body - like it used the regular importer despite the message.

If you use File | Open and then select the file you should get a prompt like

A2DD9B28-4865-468D-90B3-A024403FB57A_4_5005_c.jpeg (29.82 KiB) Viewed 879 times

The other reason it might not have worked is that you had not checked out branch newmayer.
Ypu need to go to .../Mod/OpenSCAD_Alt_Imporrt and run the command git checkout newmayer.

[edwilliams16]

Got your test file working (needed the git checkout), however if I translate a sphere in the z-direction, it fails.

Try:

Code: Select all

hull() {
	sphere($fn = 0, $fa = 12, $fs = 2, r = 5);
	multmatrix([[1, 0, 0, 20], [0, 1, 0, 0], [0, 0, 1, 20], [0, 0, 0, 1]]) {
		sphere($fn = 0, $fa = 12, $fs = 2, r = 10);
	}
}

as the csg

[keithsloan52]

Got your test file working (needed the git checkout), however if I translate a sphere in the z-direction, it fails.

Try:

Code: Select all

hull() {
	sphere($fn = 0, $fa = 12, $fs = 2, r = 5);
	multmatrix([[1, 0, 0, 20], [0, 1, 0, 0], [0, 0, 1, 20], [0, 0, 0, 1]]) {
		sphere($fn = 0, $fa = 12, $fs = 2, r = 10);
	}
}

as the csg

Same bug if two circles

[edwilliams16]

Same bug if two circles

In OpenSCAD hull() doesn't work in 3D on 2D objects, It projects the objects onto the xy-plane, so I didn't expect it to work on circles that were not co-planar. https://en.wikibooks.org/wiki/OpenSCAD_ ... tions#hull

On the other hand, it should be expected to work for two spheres regardless of their placement. Looking at your code (I can't debug it readily, see https://forum.freecadweb.org/viewtopic.php?f=4&t=57657), it appears you haven't ensured that all your four circles are in the same plane. They need to be all in some plane containing the line between their centers. Note that their Placement.Rotation.Axis are irrelevant. You can rotate a sphere around any axis and it remains unchanged. You need to define Part.Circle.Axis in a common direction orthogonal to v1 - v2. One way is to cross-product v1 - v2 with the three axis directions and choose one that is non-zero. You need to special case v1=v2 anyway - return the larger sphere.

The other thing that looks like an error to me is the definition of l2, surely it should be

• l2 = Part.makeLine(c2.value(t3), c1.value(t3 + Math.pi))

or it won't connect up with a2?

[keithsloan52]

In OpenSCAD hull() doesn't work in 3D on 2D objects, It projects the objects onto the xy-plane, so I didn't expect it to work on circles that were not co-planar. https://en.wikibooks.org/wiki/OpenSCAD_ ... tions#hull

So I see

08807832-2B2F-4351-B352-D498DBB5BE6F.jpeg (77.18 KiB) Viewed 813 times

Where as currently in FreeCAD

5116A1E4-271E-40C3-82CC-A8446F2106CD.jpeg (154.22 KiB) Viewed 813 times

Two spheres was based on treating as two circles then revolved so resolving situation with 2D might fix two spheres .

[edwilliams16]

The 2D case will require even more coding because a tilted circle projects down to an ellipse. In general, you'll have to construct the common tangent of two ellipses, which is more complicated than the two circle case. The 3D case of two spheres only requires the fixes I already gave.

Here's a generic 2D case:

Code: Select all

hull() {
      rotate(35, [1, 0 ,1]) circle($fn = 0, $fa = 12, $fs = 2, r = 25 );
      translate([40,0,40]) rotate(60, [1,1,1]) circle($fn = 0, $fa = 12, $fs = 2, r = 5);

Screen Shot 2021-04-11 at 10.58.09 PM.png (16.71 KiB) Viewed 797 times

[keithsloan52]

The 2D case will require even more coding because a tilted circle projects down to an ellipse. In general, you'll have to construct the common tangent of two ellipses, which is more complicated than the two circle case. The 3D case of two spheres only requires the fixes I already gave.

Here's a generic 2D case:

Code: Select all

hull() {
      rotate(35, [1, 0 ,1]) circle($fn = 0, $fa = 12, $fs = 2, r = 25 );
      translate([40,0,40]) rotate(60, [1,1,1]) circle($fn = 0, $fa = 12, $fs = 2, r = 5);

Screen Shot 2021-04-11 at 10.58.09 PM.png

Think it is going to be a case of check if they are planer and then fine, if not resort to OpenSCAD and Mesh/DXF

[edwilliams16]

I'm not sure what the use case is for the 2D version.
Anyhow, I fixed OpenSCADHull.py as I described above so it works for other placements of two spheres. It would take a while to figure out enough git to create a pull request, so here's the fixed file:

[keithsloan52]

I'm not sure what the use case is for the 2D version.
Anyhow, I fixed OpenSCADHull.py as I described above so it works for other placements of two spheres. It would take a while to figure out enough git to create a pull request, so here's the fixed file:

Thanks I added and merged.

Okay assume I have two cylinders, or might be a cylinder and a cone, arbitrary position and rotation other than collinear ( Not sure if that is correct term) but axis of cylinder/cone are the same.

Is there a good way to check if they are collinear from their Position.Rotations which would have to be equal and Position.Bases bases such that a vector from the Base of one to the other would also be collinear with the axis.