Here's a refinement using children.
Still only generates a thin shell object.
Aals the morphology operators only work on 2D shapes . Else coudl outset the
resulting shell and difference it with a cube.
I've highlighted the cube mod and used inset so it works on arbitrary stl
files.

eps=0.01;
use<scad-utils/morphology.scad>;

// sample shape
module shape()  {
hull() {
sphere(10);
translate([15,0,10]) sphere(10);
}
hull() {
translate([15,0,10]) sphere(10);
translate([45,0,0]) sphere(10);
}
}

// Generate parting line object
module parting_1() {
// Call with a child object of any kind
scale([2+eps,2+eps,1])color("red")
intersection(){
// move object below origin
translate([0,0,-50])
linear_extrude(convexity=6)
// create a thin border line (flat at origin)
#difference(){
square(size=[200,100], center=true);
inset(d=eps)
projection()
children(0);
}
// intersect with shape
children(0);
}
}

//
parting_1() shape();

--
View this message in context: http://forum.openscad.org/Parting-Line-tp13715p13740.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

Ideally the projection() function would somehow make a polygon list
available.
Then if we could test if intersection() of a point, projected along Z, hit
an object and the resulting XYZ location we could then construct a
polyhedron from resulting list of intersection points.
This could then be further adjusted using existing tools to make a two part
mold.

any chance of thi skind of interaction in OpenSCAD2 ?

--
View this message in context: http://forum.openscad.org/Parting-Line-tp13715p13741.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

...again we come to the dichotomy of OpenSCAD; currently OpenSCAD is a
geometry generator as a descriptive language, lately people want to make it
an iterative geometry processor, other languages do that.

(yes I have asked for stuff like bounding box in the past...)

Newly minted Admin - PM me if you need anything, or if I've done something stupid...

Unless specifically shown otherwise above, my contribution is in the Public Domain; To the extent possible under law, I have waived all copyright and related or neighbouring rights to this work. This work is published globally via the internet. :)  - Obviously inclusion of works of previous authors is not included in the above.

The TPP is no simple “trade agreement.”  Fight it! http://www.ourfairdeal.org/  time is running out!

View this message in context: http://forum.openscad.org/Parting-Line-tp13715p13743.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

yes - indeed.
However my point is to not go as far as that but to see if we can do this
kind of op with a sensible set of useful operations within OpenSCAD.
I.e. is there a way to use the core to construct (which I feel is
reasonable) a negative mold of a complex part.

My suggestion is to facilitate this within the design of OpenSCAD, not to
turn it into a more complex programming lang.

E.g.

• we have list complerehensions and we can construct polys from lists and
  polyhedrons from lists.
• we should also (IMHO) be able to get a list back from a projection() so we
  can use it in a list comprehension to further adjust it...
• I agree BBOX() is another super useful thing for exact same reasons.
  • potentially a query like bbox(modulename) would fit into the language
    (the geometry from the module being discarded from the CSG tree.

So:

• bbox(modulename); could return a list of min,max XYZ triplets.
• 2D(projection(modulename)); could return a nested polygon list. I.e.
  points and paths.
• hit_test(object1, object2); could return a true or false depending on the
  existance of an intersected result. or better an XYZ triplet of center of
  bbox of the intersected objects.
  In all three cases any resulting geometry would be discarded (or hidden(*)
  so caches would be kept) from the final tree.

In this way there is minimal adjustment to how OpenSCAD currently works.

--
View this message in context: http://forum.openscad.org/Parting-Line-tp13715p13744.html
Sent from the OpenSCAD mailing list archive at Nabble.com.