On Jan 12, 2025, at 7:53 AM, Jon Bondy jon@jonbondy.com wrote:

"learn to write models through points in space, create meshes ... The sooner you can do that the better."

Our goals are opposite.  Languages like OpenSCAD are useful to me EXACTLY because I do not have to create meshes by hand.

On 1/12/2025 4:32 AM, Sanjeev Prabhakar via Discuss wrote:

http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient Virus-free.www.avg.com http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient x-msg://1/#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2

I would try Bosl2's join_prism() to do the fillet.

On Fri, Jan 10, 2025, 8:03 PM Mark Erbaugh via Discuss <
discuss@lists.openscad.org> wrote:

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

On Jan 12, 2025, at 10:47 AM, David Phillip Oster via Discuss discuss@lists.openscad.org wrote:

Taking Todd Allen's suggestion, here is amended .scad, adding the fillet in BOSL2.  I've marked the additional source code lines with '// add' to make them easier to see.

On Fri, Jan 10, 2025 at 7:11 PM Todd Allen via Discuss <discuss@lists.openscad.org mailto:discuss@lists.openscad.org> wrote:

<Challenge 11-fillet.scad>_______________________________________________
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

First,

Thanks for the replies.

The method I used was to do as much in 2D and then rotate_extruded that
into 3D. In both modules top() (vertical piece) and side() (horizontal
piece) the outline array and polygon() creates a 2D outline of one side of
the cross-section of the pipe with flange. The fillet module adds a 2d
fillet under the flange. The polygons are created in the XY plane with the
flange on the X axis and will be translated and rotated into the final
position. Since the horizontal piece is symmetrical around the Y axis
(which is the X axis before rotation), the polygon only creates half and
that is mirrored.

After the rotate extrude, the hole for the other pipe is added as are the
bolt holes. The two pieces are then rotated and translated so that the
intersection is at the origin and the two pieces are joined() with a union.

I’m not sure I fully understand the comment of creating meshes from
polyhedra. I thought that was the purpose of OpenSCAD.

Mark

On Jan 12, 2025, at 7:53 AM, Jon Bondy jon@jonbondy.com wrote:

"learn to write models through points in space, create meshes ... The
sooner you can do that the better."

Our goals are opposite.  Languages like OpenSCAD are useful to me EXACTLY
because I do not have to create meshes by hand.

On 1/12/2025 4:32 AM, Sanjeev Prabhakar via Discuss wrote:

It's a little difficult to go line by line and understand the code written
by you (At Least for me).

I appreciate the final outcome from your code though.

I would suggest you learn to write models through points in space, create
meshes and use module polyhedron to render the final model.

The sooner you can do that the better.

For creating fillets, you need to to find intersection points between the
2 solids.That would need some work.

Getting the fillets from primitives in openSCAD is extremely difficult as
per my understanding.

Good luck

On Sat, 11 Jan 2025 at 07:33, Mark Erbaugh via Discuss <
discuss@lists.openscad.org> wrote:

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
Virus-free.www.avg.com
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

OK, just for a simple example of creating a cube
consider the following code. 'swp' is a module which creates the mesh
from the points.
p0: is a square in x-y plane
p1: is a square translated to 10 units in z-direction.
you can play around with the coordinates in p0 and p1 to see different
shapes

p0=[[0,0,0],[10,0,0],[10,10,0],[0,10,0]];
p1=[[0,0,10],[10,0,10],[10,10,10],[0,10,10]];

cube_1=[p0,p1];

swp(cube_1);

function faces(sol)=

//    calculate the faces for the vertices with shape l x m with first
and the last end closed
    let(
    l=len(sol),
    m=len(sol[0]),
    n1=[for(i=[0:m-1])i],
    n2=[for(i=[0:l-2]) each ([ for(j=[0:m-1])
    each
j<m-1?[[(j+1)+im,j+im,j+(i+1)m],[(j+1)+im,j+(i+1)m,(j+1)+(i+1)m]]:
    [[0+im,j+im,j+(i+1)m],[0+im,j+(i+1)*m,0+(i+1)*m]]
    ])],
    n3=[for(i=[0:m-1])i+(l-1)*m],
    n4=[for(i=[len(n3)-1:-1:0])n3[i]],
    n=[n1,each (n2),n4]
    )n;

function faces_1(sol)=

//    calculate the faces for the vertices with shape l x m with first
and the last end open
    let(
    l=len(sol),
    m=len(sol[0]),
    n2=[for(i=[0:l-2])each([ for(j=[0:m-1])
    each
j<m-1?[[(j+1)+im,j+im,j+(i+1)m],[(j+1)+im,j+(i+1)m,(j+1)+(i+1)m]]:
    [[0+im,j+im,j+(i+1)m],[0+im,j+(i+1)*m,0+(i+1)*m]]
    ])]

    )n2;

 function vertices(sol)=
[each for (p=sol)p];

// module for rendering the polyhedron with ends closed
module swp(sol){
let(
v1=vertices(sol),
f1=faces(sol)
)
polyhedron(v1,f1,convexity=10);

}

On Sun, 12 Jan 2025 at 21:51, Mark Erbaugh via Discuss
discuss@lists.openscad.org wrote:

 First,

 Thanks for the replies.

 The method I used was to do as much in 2D and then rotate_extruded
 that into 3D. In both modules top() (vertical piece) and side()
 (horizontal piece) the outline array and polygon() creates a 2D
 outline of one side of the cross-section of the pipe with flange.
 The fillet module adds a 2d fillet under the flange. The polygons
 are created in the XY plane with the flange on the X axis and will
 be translated and rotated into the final position. Since the
 horizontal piece is symmetrical around the Y axis (which is the X
 axis before rotation), the polygon only creates half and that is
 mirrored.

 After the rotate extrude, the hole for the other pipe is added as
 are the bolt holes. The two pieces are then rotated and translated
 so that the intersection is at the origin and the two pieces are
 joined() with a union.

 I’m not sure I fully understand the comment of creating meshes
 from polyhedra. I thought that was the purpose of OpenSCAD.

 Mark

 _______________________________________________
 OpenSCAD mailing list
 To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

Cube is just to explain the process.
Much complex shapes can be developed with a list of points in a
similar way.

Once you create models with the list of points , there are a lot of
operations which can be done on them like calculating the intersection
between 2 solids.

But this is only required if you need to do some complex things like
fillets.

Otherwise stock openSCAD is good enough.

On Sun, 12 Jan, 2025, 11:52 pm Mark Erbaugh via Discuss, <
discuss@lists.openscad.org> wrote:

This is my python few lines of code for this:

from openscad2 import *
sec=circle(9/2,s=100)
path=corner_radius(pts1([[0,0],[1.5,0],[0,19+1,1],[2,0],[0,2],[-2,0],
                        [0,-1],[-1.5,0],[0,-19]]),10)

sol1=translate([0,0,-9],prism(sec,path))

path1=corner_radius(pts1([[0,0],[1.5,0],[0,.35],[2,0],[0,2],[-3,0,1],[0,24,1],
 [3,0],[0,2],[-2,0],[0,.35],[-1.5,0],[0,-28.7]]),10)
sol2=translate([-14,0,0],rot('y90',prism(sec,path1)))
l1=ip_sol2sol(sol1,slice_sol(sol2[15:17],2))
l2=o_3d(l1,sol1,1)
l3=i_p_p(sol2,l1,-1)
f1=convert_3lines2fillet_closed(l3,l2,l1)
f2=mirror_surface(f1,[1,0,0],[0,0,0])

d1=o_solid([0,0,1],circle(9/2,s=100),20,-8)
d2=o_solid([1,0,0],circle(9/2,s=100),20,-10)
with open('trial.scad','w+') as f:
    f.write(f'''
    include<dependencies2.scad>
    difference(){{
    {swp(sol1)}
    {swp(d2)}
    }}
    difference(){{
    {swp_c(sol2)}
    {swp(d1)}
    }}
    {swp_c(f1)}
    {swp_c(f2)}

    ''')

Screenshot 2025-01-13 at 10.33.27 PM.png

scad output file is attached, but this is all just points in space

On Mon, 13 Jan 2025 at 07:09, Sanjeev Prabhakar
sprabhakar2006@gmail.com wrote:

 Cube is just to explain the process.
 Much complex shapes can be developed with a list of points in a
 similar way.

 Once you create models with the list of points , there are a lot
 of operations which can be done on them like calculating the
 intersection between 2 solids.

 But this is only required if you need to do some complex things
 like fillets.

 Otherwise stock openSCAD is good enough.


 On Sun, 12 Jan, 2025, 11:52 pm Mark Erbaugh via Discuss,
 <discuss@lists.openscad.org> wrote:

     I ran the code in OpenSCAD, but how is that different from a
     simple cube(10); which is a lot easier to type and understand?

     Is this what OpenSCAD’s cube() does under the hood? Is that
     how more complex shapes are generated by BOSL2?

     I asked a question on this forum a while back about using
     multmatrix to combine a series of rotate(), scale() and
      translate() calls into a single transformation, thinking that
     it might be quicker if those calls were done many times in a
     script and the answer was that OpenSCAD combines those calls,
     and computes one matrix and uses the matrix in its
     calculations. So I’m thinking that if OpenSCAD has a  module
     to do it, that module should be used.

     This raises a question: are the routines in OpenSCAD written
     in a lower level OpenSCAD or do they go to a lower level
     language, such as C(++).

     Mark

     _______________________________________________
     OpenSCAD mailing list
     To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

You can get the path of the plain tube joints in raw openscad (thin walls
and intersection). You can then Minkowski that path, using an extruded
square with the corners cut away by circles, which almost gives the
required rounded fillet. However the fillet radius needs to change as it
follows the path, and the shape needs to rotate, and Minkowski is too slow
in this instance. A flat fillet would be simpler, I think, since could get
the path for the edges of the fillet on each tube. IIrc, freecad can fillet
openscad script generated files, but changes everything to multi-matrix. If
it wasn't a requirement for it to be parametric, then I suppose a quarter
of one joint fillet could be manually or mathematically  generated in fine
enough steps, and that replicated 8 times.
On 13/01/2025 17:10, Sanjeev Prabhakar via Discuss wrote:

This is my python few lines of code for this:

from openscad2 import *
sec=circle(9/2,s=100)
path=corner_radius(pts1([[0,0],[1.5,0],[0,19+1,1],[2,0],[0,2],[-2,0],
[0,-1],[-1.5,0],[0,-19]]),10)

sol1=translate([0,0,-9],prism(sec,path))

path1=corner_radius(pts1([[0,0],[1.5,0],[0,.35],[2,0],[0,2],[-3,0,1],[0,24,1],

[3,0],[0,2],[-2,0],[0,.35],[-1.5,0],[0,-28.7]]),10)
sol2=translate([-14,0,0],rot('y90',prism(sec,path1)))
l1=ip_sol2sol(sol1,slice_sol(sol2[15:17],2))
l2=o_3d(l1,sol1,1)
l3=i_p_p(sol2,l1,-1)
f1=convert_3lines2fillet_closed(l3,l2,l1)
f2=mirror_surface(f1,[1,0,0],[0,0,0])

d1=o_solid([0,0,1],circle(9/2,s=100),20,-8)
d2=o_solid([1,0,0],circle(9/2,s=100),20,-10)
with open('trial.scad','w+') as f:
f.write(f'''
include<dependencies2.scad>
difference(){{
{swp(sol1)}
{swp(d2)}
}}
difference(){{
{swp_c(sol2)}
{swp(d1)}
}}
{swp_c(f1)}
{swp_c(f2)}

 ''')

[image: Screenshot 2025-01-13 at 10.33.27 PM.png]

scad output file is attached, but this is all just points in space

On Mon, 13 Jan 2025 at 07:09, Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

Hi Ray,

The point I want to make here is that if you learn to draw models with
points in space instead of openSCAD primitives, it gives a lot of added
flexibility to draw many complex models.

I am sure there could be ways to generate the same thing someway in stock
openscad.

Like you mentioned minkowski but it might take a very long time to render.

Finally, I wish python could be added in openscad as an alternate
language, which can make things much easier.

Regards

On Tue, 14 Jan, 2025, 12:59 am Raymond West via Discuss, <
discuss@lists.openscad.org> wrote:

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

re: python as an alternative language...

You might want to take a look at https://pythonscad.org/ and
https://github.com/SolidCode/SolidPython. I played with them a year or
two ago when working on some 3D parametric optimization using genetic
algorithms.  I have not finished that work, and it will be quite
awhile before I can get back to it.

On Mon, Jan 13, 2025 at 7:57 PM Sanjeev Prabhakar via Discuss
discuss@lists.openscad.org wrote:

 Hi Ray,

 The point I want to make here is that if you learn to draw models
 with points in space instead of openSCAD primitives, it gives a
 lot of added flexibility to draw many complex models.

 I am sure there could be ways to generate the same thing someway
 in stock openscad.

 Like you mentioned minkowski but it might take a very long time to
 render.

 Finally, I wish python could be added in openscad as an alternate
 language, which can make things much easier.

 Regards


 On Tue, 14 Jan, 2025, 12:59 am Raymond West via Discuss,
 <discuss@lists.openscad.org> wrote:

     You can get the path of the plain tube joints in raw openscad
     (thin walls and intersection). You can then Minkowski that
     path, using an extruded square with the corners cut away by
     circles, which almost gives the required rounded fillet.
     However the fillet radius needs to change as it follows the
     path, and the shape needs to rotate, and Minkowski is too slow
     in this instance. A flat fillet would be simpler, I think,
     since could get the path for the edges of the fillet on each
     tube. IIrc, freecad can fillet openscad script generated
     files, but changes everything to multi-matrix. If it wasn't a
     requirement for it to be parametric, then I suppose a quarter
     of one joint fillet could be manually or mathematically 
     generated in fine enough steps, and that replicated 8 times.

     On 13/01/2025 17:10, Sanjeev Prabhakar via Discuss wrote:

     _______________________________________________
     OpenSCAD mailing list
     To unsubscribe send an email to discuss-leave@lists.openscad.org

 _______________________________________________
 OpenSCAD mailing list
 To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

I've used cadquery, which is sort of openscad with python. However it
constructs its solids differently, and some aspects do not work well,
giving distorted images. iirc, things like a sphere has a seam in it,
and if you try to take a scallop out of a surface of a cube, say, then
the sphere has to be rotated so the seam is not in the that region.
https://cadquery.readthedocs.io/en/latest/vis.html shows the seam in
the illustrations. I found the only sparse support was on Reddit,  I
couldn't find a forum/email list as helpful or responsive as this.

But I thought that the idea in this 'puzzle' was to use openscad
script only, else you may as well use any other language/cad program.
My idea, although it failed in this case, gives a very good
representation of a beginner's welded joint...

But I  agree that the points in space is important, I have in the past
resorted to exporting an SVG, and in python (or any other language
would suffice) calculated intermediate points and manipulated them,
saved, and reimported to openscad. It would be nice if openscad could
programmatically export and import plain text files, but in order to
keep the functional paradigm, I expect it would have  to be two
separate compiles. Finding work-arounds is interesting.

On 14/01/2025 02:58, John David via Discuss wrote:

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

On Jan 14, 2025, at 7:34 AM, Raymond West via Discuss discuss@lists.openscad.org wrote:

This gives a flat fillet (which I've called a bevel - depends if you think it is internal or external, draughtsman or carpenter)

module tube(d){
linear_extrude(100)
difference(){
circle(d=d);
circle (d=d-0.0001);
};
}

hull(){
intersection(){
tube(30);  //upright tube
translate([-0,0,50])
rotate([0,90,0])
tube(20+bevel);
}
intersection(){
tube(30+bevel);  // side tube
translate([-0,0,50])
rotate([0,90,0])
tube(20);
}
}

$fn=100;
cylinder(d=30,h=100);
translate([-0,0,50])
rotate([0,90,0])
cylinder(d=20,h=50);

     bevel=6;  //  greater than 2.

On 14/01/2025 11:59, Raymond West via Discuss wrote:

OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

On 1/14/2025 6:26 AM, Jon Bondy via Discuss wrote:

There are several techniques for fillets and chamfers/roundoffs. They
are just rarely easy.  BOSL2 does offer a lot of options, which it
does the hard way (by constructing polyhedra) and then hides the
difficulty.

I could not see any chamfer or fillet here.
Anything missing?

On Tue, 14 Jan 2025 at 18:04, Raymond West via Discuss <
discuss@lists.openscad.org> wrote:

    circle (d=d-0.0001);

…