Yes but in this particular 4 / symetrics gem I believe that work ;)

Hi Pilou,

> Yes but in this particular 4 / symetrics gem I believe that work ;)

Actually no, I don't think so because there are different shaped facets in the gem, which are at varying angles to each other, each face is not of the same exact shape at the same angle from the center point.

For example, here is again the image that I posted earlier that shows what an offset of this shape should look like:



Notice how the offset result should have a vertex structure where there are 3 different vertices where pieces meet in the center?

If it was equivalent for scale and offset, it should have just a single vertex where the offset results meet up.

- Michael

Hi Marco,

> btw. are there any plans for the future to support the DXF-format natively?

Yes, I would like to support DXF and possibly DWG as well. But they are not really that easy of a format to work with, it may be a while before I will be able to get to that.

- Michael

So I must make a big zoom on these crossing facets :)
Homothety seems keep the same variable angle
I don't see problem
I use my previous post file
Maybe Scale 3D is different than Offset?

Hi Pilou,

you wrote:
> I don't see problem

If you take a look at my image above that shows what the offset should look like, you can see that although those 3 vertices in the middle are distinct, they are not terribly far apart from one another, they are fairly close together.

So at a smaller distance, the error will be kind of hard to tell just by looking at the result, you would need to measure distances between shapes instead of just looking at them to be able to detect the difference.


> Maybe Scale 3D is different than Offset?

Yes - Scale3D is definitely different than Offset, that's what I was trying to show with the illustrations here: http://moi3d.com/forum/index.php?webtag=MOI&msg=1939.20

Only in special cases like a regular polygon or a circle/sphere are they the same. This is not one of those cases - it is close but not quite.

- Michael

:)
If I understand the problem with the "Shell" function" I don't see why the use of Scale 3D will be forbidden in this case?
Thickness is not given?
Thickness can be anything, angles between each facets are kept, and there are no problem for the vertex
It's just that the Shelling is forbidden for this case, Scale 3D is adapted ;)

Hi Pilou,

> I don't see why the use of Scale 3D will be forbidden in this case?

Maybe it isn't, it depends on what Marco's specific design needs were.

Since he was talking about shelling originally it seemed like he had a design goal of making a shape with a constant thickness.

If that was an important design goal for him, then he can't use the Scale 3D as a replacement, since it will not produce a shape of constant thickness, so not adhering to the design goals....

Maybe his design goals were flexible enough in this case to be able to tolerate a deviation from uniform thickness, only he can really determine that.

Often times when shapes are being used for engineering and physical model building it is just not a very desirable option to replace pieces with similar ones that just look pretty close with a basic visual inspection - that's usually fine when you are only going to be rendering your geometry but manufacturing your geometry tends to have a lot different set of goals...

- Michael

So what function use? By hand ?

Hi Pilou,

> So what function use? By hand ?

It is possible to do it by hand by creating a curve structure like I showed above.

But actually an easier way I found to do it is explained earlier in the thread here: http://moi3d.com/forum/index.php?webtag=MOI&msg=1939.16

I have a shelled result there that was done by shelling individual surfaces and then booleaning those together, then doing a bit of trimming. That actually comes together really quickly, definitely easier than working at the curve level.

Then the other option is to export to a different CAD program that has a shell function that can handle that situation where there isn't a 1-to-1 matching between the original shape's vertices and the shelled result vertices.

- Michael

But there are imperfections inside face object! ??

Not make something like draw a same circle from each vertice then link them by the extrema radius?
Thickness will be constant and no problem at the vertex

Hi Pilou,

> But there are imperfections inside face object! ??

Which ones do you mean? There are some areas with multiple vertices - that is the proper result of a constant thickness shell just like I was showing in that previous example.

To maintain a constant thickness it can require a different vertex structure on the output shape than the input shape, again see that previous illustration I posted here: http://moi3d.com/forum/index.php?webtag=MOI&msg=1939.10 for an example.

You can't have both a matching vertex-for-vertex on this shape as well as constant thickness, you have to choose one or the other.

There is probably some way to have rounded type patches in those corner areas instead of having the planes extend to meet each other in a sharp corner - the one I show there has the straight extension method.

EDIT: actually no - since that one I shelled to the inside I think that is the proper result there isn't room to fill in rounded patches like there might be for the shell to the outside direction...

- Michael

Yes this one!
I suppose we can't consider that like a perfect result ;)

Hi Pilou, here is another example, say I have this curve:



If I want to produce a constant distance offset of it of a certain amount, I will get this:



Notice how the offset curve has a different number of vertices and just a different structure than the starting shape?

I know it seems like it should be logical for constant thickness to mean equal structure and equal number of vertices, but there is not really a connection there - a constant thickness in many cases requires creating a shape with a different structure in it from the original.

This is also again another good example of how offset is different from scaling as well...

- Michael

Hi Pilou,


> Yes this one!
> I suppose we can't consider that like a perfect result ;)

Actually if you are looking for constant thickness, yes that is completely perfect as far as I can tell!

It all depends on what you are looking for - if you want equal number of vertices, then you will have to create a shape that does not have constant thickness. If you want constant thickness, that is what that shape I have there does.

I know it seems intuitive that an equal number of vertices will produce and equal thickness, but that is not the case - it is only case for completely symmetrical things like a sphere, you are just thinking of this object too much like a sphere or regular polygon when it has different angled pieces which produce non-sphere like results...

- Michael

Thx : it very difficult to fight the intuitivity!
Intuitivity want same numbers, your logic result seems more complex to make for a jeweler!
I am curious of the Marco's result wanted and realised