hi Sanjeev,

I know exactly what you mean, I also do green-sand-casting and demolding it
the key for successful metal casting.

I spent over 6 month on writing software to find the best split of a given
bust stl file programmatically. the tool is called "stltool"

Right now developing the tool is on hold as I need some time for more
inspiration.

Examples of successful splits are here:

https://twitter.com/gadget_3d/status/1581385519308890112/photo/1

was also able to split "batman" mold successfully, but by far not all bust
are splittable for molding.

when splitting your weel, which is very symmetrical, isn't it possible to
split the face at the symmetry line using "difference" or "intersection" ?

when you very easily split your triangles into two bins when watching if
the z coordinate of the normal vector is pointing up or down.

For my tool i spent

25% of the time on finding the best cut line around the model
75% of the time to filling the big hole in the model which arises when
splitting the model into 2 halves ....

On Fri, Feb 24, 2023 at 2:29 PM Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

Hi Guenther

Your picture of the split looks interesting.

Can you give some more information on the workflow and the logic.

In case it is possible to share the stltool developed by you, would be
great, although i have very limited understanding of softwares.

I started working on openscad as it initially looked simpler to work on.

Then learnt a little bit of python to circumvent the limitations of
openscad language.

Now it seems I need to learn another one. Not sure if I will do that as it
is taking a lot of my time.

Anyway, I am not using it for any professional purposes and is a hobby.

On Fri, 24 Feb, 2023, 7:15 pm Guenther Sohler, guenther.sohler@gmail.com
wrote:

Hi Sanjeev,

stltool is coded in C and its a requirement to achieve the speed

stltool - moldsplit  main task is to split a model around the biggest
lateral extent(this is where faces-up meet faces-down)

A) Find starting point
this is easy: find the lateral point, which has biggest lateral distance
from the center
B) find cutline around model
from the starting point build a tree, only valid directions are edges
where edge-up meet edge-down
finally hoping you can surround the model in CCW direction and hopefully
hit the start point again
C) find out which triangles are "above" and "below" the cutline. this is
not necessarily the triangles with the normvector show up and down

D) duplicate all points at the culine and use them for bottom and upper
half respectively
E) finally move the objects apart by simply raising z coordinate of upper
half
F) now fill the big hole which arises when pulling the halves apart

• simple polygon fill is not possible as the polygon might be concave
• polygon tesselation is not possible because the points are in 3d, not 2d
• finally i came up which an algoithm which offsets the polygon with
  negative values iteratively.
  at some point two corners melt into one. this is exaclty the required
  "helper point" to use to tesselate the polygon in a concave  - friendly way.
  (actually i believe this algorithm is very similar to openscad "roof"
  function.

Hope you got an idea, how the algorithm works, but i doubt you will make it
in python...

cheers

On Fri, Feb 24, 2023 at 3:07 PM Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

These are good tips to find way out.

I will try this in my free time.

Thanks for suggestions

On Fri, 24 Feb, 2023, 7:54 pm Guenther Sohler, guenther.sohler@gmail.com
wrote:

I haven't played with this problem myself, but is there a reason to not
have the algorithm decide a flat cut plane, and then find the triangles in
the stl that intersect this plane, and then cut these triangles into new
ones that are placed in the appropriate half? The intersection with a
triangular face in the model seems like it can only have three forms:

1. coincident with a side, which is sort of like a non-intersection
   (triangle goes in one half)
2. coincident with a corner, which then creates two triangles (one to each
   half)
3. the else... will produce a triangle to go into one half, and a
   trapezoid... the trapezoid can be broken into two (or more) co-planar
   triangles.

Or just have the program suggest the best cut plane and let openscad handle
things using difference or intersection appropriately...?
In the case of the wheel, I'm not sure I understand why this is a challenge
(intersection with a cube where one face is the center line) - can you
explain?

On Fri, Feb 24, 2023 at 7:53 AM Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

you have spokes also which are not perfectly in center and the difference
doesn't work, due to the profile of the spokes.

Also if we need to work on this, it should be a generic solution as far as
possible which can be used elsewhere.

On Fri, 24 Feb 2023 at 20:38, Daniel Harvey dwaharvey@gmail.com wrote:

Hi Daniel,

In ideal case, a model like a wheel, is actually symmetrical around a flat
XY plane with 0 elevation at a certain z coordinate and fulfilling sanjeevs
task is just to
difference and to intersect with a big cube.

Unfortunately many models are not as easy as that. they have the point of
maximum lateral extent in differenz z heights.
If you still split such a model with a flat XY cut face, you could end up
in an upper mold which has faces which point downwards.

Did you browse youtube videos on green sand casting ? its very interesting.
if you do, you might get an idea why I came up with such a complicated
picture cut like in my first post.

On Fri, Feb 24, 2023 at 4:08 PM Daniel Harvey dwaharvey@gmail.com wrote:

Hi Guenther,
In everything I have seen, an inprint is made in each of two halves, and
the halves are assembled together (with imprinting objects removed) and the
resulting mold is filled with a liquid that sets. But in all these cases
the two halves basically have a flat intersection face.  In the example of
your Gollum bust, how are the two halves with the non-flat faces turned
into a mold? Also it seems like there would be many objects that could not
be handled at all - objects where there is no cutline that doesn't create a
situation with  an upward pointing face that is shadowed by a downward
pointing one.

On Fri, Feb 24, 2023 at 8:22 AM Guenther Sohler guenther.sohler@gmail.com
wrote:

Hi Daniel,

Maybe you have  only seen mold halves with a flat cutline, because
the models werd specially designed to have a flat cutline ?
(or nobody has managed to split models with a non-planar cutline ? )
But you are right.
Cutting an object into 2  halves can be a challenge. Finding out that the
gollum bust has an issue to demold i only found out after  I managed to
retrieve the cutline.
(between the ears and the neck)
Also I doubt that the gollum bust can be demolded with a flat cutline.
Generally I believe it's quite hard to find busts on Thingiverse which can
easily be split for demolding.

But  I was successfully able to split batman.

https://cults3d.com/en/3d-model/art/batman-bust-ready-for-sandcast-molding

When you look at the splitted parts separated, you might understand, why a
flat cutline would not work out

Just for interest, do you  have some experience with sand casting ?

On Fri, Feb 24, 2023 at 4:48 PM Daniel Harvey dwaharvey@gmail.com wrote:

you can have various cases where the parting lines are very zigzag.

In die casting engineering products, I am aware of many such examples.

e.g. refer file attached
[image: download.png]

On Fri, 24 Feb 2023 at 21:17, Daniel Harvey dwaharvey@gmail.com wrote:

Not sure if this is what you want Sanjeev, it's pretty simplistic code-
I wrote it to create patterns for casting flywheels for small stationary
engines. It has draft so you can get it out of the sand, and you can set
oval or round spokes, any (reasonable) number of spokes, plus either a
circular hub or one with sides equal to the number of spokes. I just use
difference to get the half, and there are two holes in the flat back for
registration pins.

$fn = 128;
include <BOSL2/std.scad>

//Set up the dimensions

hubdiameter = 31;
shaftdiameter = 12;
wheeldiameter = 100;
hubthickness = 10;
rimthickness = 20;
spokethickness = 8;
numberofspokes = 5;

//Set draught so mold can be
//removed from sand. Be aware
//that too much draught will break
//the fillet at the base of the spoke.

draught = 1;//[.5:.1:10]

//Don't change the next 5 lines
wheelrimthickness = (wheeldiameter - rimthickness);
spokelength = 1+(((wheeldiameter-rimthickness)/2)- (hubdiameter/2.4));
rotation = 360/(numberofspokes);
spoke_rotation = rotation*0.5;
hubradius = hubdiameter/2;
circumradius = hubradius/cos(180/numberofspokes);

//Set the ovalness of the spokes (1=round,
//  less than one is oval in one plane,
//  more than one is oval in the other plane
ovality = 1.5;

//Make the wheel

difference(){
//Wheel rim
difference(){
  difference(){
    color ("violet")
cylinder(h=12,d1=wheeldiameter,d2=wheeldiameter-draught,center=true);
//Two 3mm holes so the wheel halves
//can be correctly registered in the mold
    translate([0, wheeldiameter/2-5, -5])
    cylinder(h=10,d=3);
    translate([0, -(wheeldiameter/2-5), -5])
    cylinder(h=10,d=3);
  }
//Cut out the annulus and make the spokes
difference(){
    color ("blue")
cyl(h=13,d1=wheelrimthickness-draught,d2=wheelrimthickness,center=true);
   for (i=[0:numberofspokes]){
    rotate([90, 0, rotation * i + spoke_rotation])
    translate([0,0,12])
    scale([ovality,1,1])
    color ("red")
//comment out cylinder line and uncomment
//cube line for rectangular spokes
    cyl(h=spokelength, d=spokethickness, rounding1=-draught,
rounding2=-draught, center=false);}
 //  cuboid([65,spokethickness, 5], rounding= 1);}
  }
  color ("lightblue");
cylinder(h=12,d=15,center=true);
}

//This cube differences out half the wheel
//to make a mold for casting
translate ([0, 0, -10])
cube ([wheeldiameter+10, wheeldiameter+10, 20],center = true);
}
//Draw half a hub center
difference(){
  color ("green")
  translate ([0, 0, 3])
  rotate ([0,0,90])
cyl(h=6,d1=circumradius2,d2=(circumradius2)-draught,center=true,
rounding2 =.5, $fn = numberofspokes);
  color ("navy")
  cylinder(h=13,d1=shaftdiameter,d2=shaftdiameter-draught, center=true);
}

On 2023-02-25 00:28, Sanjeev Prabhakar wrote:

--
Cheers, Ken
bats059@gmail.com
https://vk7krj.com/running.html
https://sstv.vk7krj.com/all_bands.html

A baby can be defined as an ego with a noise at one end and a smell at the other.
Your job as parents is to teach them to control all three.
My job as a retiree is to tell you how you are doing it all wrong!

Hi Ken
my model is a little complicated and difference will not work, I have tried
it
[image: wheel.png]

Ah yes, I see.
Looking at it, that would be a rather tricky moulding job,
at least 4 parts that I can see, and rather tricky to assemble.

Sooner you than me Sanjeev!

On 2023-02-25 09:50, Sanjeev Prabhakar wrote:

--
Cheers, Ken
bats059@gmail.com
https://vk7krj.com/running.html
https://sstv.vk7krj.com/all_bands.html

A baby can be defined as an ego with a noise at one end and a smell at the other.
Your job as parents is to teach them to control all three.
My job as a retiree is to tell you how you are doing it all wrong!

I think I have found a generic solution which doesn't need any difference
or intersection.

This will also show in case there are any undercuts.

It was simpler than I anticipated.

Will share it once I get some time

Thanks everyone

On Sat, 25 Feb, 2023, 12:35 pm Ken, bats059@gmail.com wrote:

results of the top and bottom surface extraction
there is some more work that needs to be done, to remove some unnecessary
things but i think it's quite workable now.
[image: Screenshot 2023-02-26 at 6.26.52 AM.png]

[image: Screenshot 2023-02-26 at 6.28.00 AM.png]

[image: Screenshot 2023-02-26 at 6.44.18 AM.png]

[image: Screenshot 2023-02-26 at 6.45.00 AM.png]

On Sat, 25 Feb 2023 at 12:50, Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

it looks very nice!
Whats your workflow like?
apparently you cannot pour molten metal onto the plastic

On Thu, Mar 2, 2023 at 6:45 PM Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

Guenther

I am not planning to do anything with this.

I was just checking if dies can be modelled with this approach.

In the process I found it is very complex right now to do something like
this. But lot of learning from this challenging work.

On Fri, 3 Mar, 2023, 3:38 pm Guenther Sohler, guenther.sohler@gmail.com
wrote:

You could use this as the start of a lost wax process.

Steve Schlaifer
On Mar 3, 2023 at 4:07 AM -0800, Sanjeev Prabhakar sprabhakar2006@gmail.com, wrote:

You can print a PLA pattern that is the same as the desired final object
but expanded by the amount the metal contracts and then build a sand or
plaster mould around it.

PLA can be burned out just as easily as wax.

On Sat, 4 Mar 2023 at 19:24, Steve Schlaifer steve@jetcafe.org wrote:

That's a good idea, will give it a try in our next project.

On Sun, 5 Mar, 2023, 12:59 am nop head, nop.head@gmail.com wrote:

My father was a pattern maker, long before 3D printing was a thing. He made
the patterns out of wood manually and had a set of rulers that were out by
the contraction ratios of different metals, which I still have.

On Sun, 5 Mar 2023 at 08:30, Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

Wow

Great to hear that

I have almost no experience of sand moulding, but we have made hundreds of
die casting moulds if not thousands.

In Engineering we have designers who mainly work on expensive softwares
like nx , Catia, creo, autocad etc.

It's very easy to make mould designs in these softwares, although I don't
work on that.

Many times I show our designers that openscad could be an alternative for
some type of designs.

But it seems they don't take me seriously and it's fine.

I like openscad although till date almost no professional work is done by
me on this, but I intend to make few functions which could be useful for
professional use.

On Sun, 5 Mar, 2023, 2:20 pm nop head, nop.head@gmail.com wrote:

I've been thinking about 3d printing molds, say for polyurethane foam
casting.

I never quite put in the effort to do it.

I'm a novice in this.

I'd seem making 3d printed molds may be difficult as it is after all
rather soft materials, and that the expanding foam may expand so much
the mold could deform.
And that it requires a mold release, which i'd guess things that could
be tried could include shoe polish, baby powder, silicone oil (spray)
etc, silicone spray is actually used in metal molds i'd guess.

that is actually the lesser of problems, but that i'm not sure how toxic
is the polyurethane materials after all MDI etc if it would necessitate
using a proper VOC respirator etc, quite expensive but may after all be
necessary. expanding foams are quite easy to buy say from construction
products businesses, e.g. for foam insulation etc.
no need for the speciality types, just that handling the chemicals I'd
guess would need quite some precautions, VOC respirator etc.

On 05/03/2023 20:37, Sanjeev Prabhakar wrote: