>If you want to have a hole going all the way from the top to the bottom, then you could get that by selecting both the top and the bottom faces before doing the shell.
That's the trick :)

< One face + 2sides
Tricky indeed!

Seems shell a very poweful function in expert's hands :)

< The other thing is that it is possible to just select the whole solid object instead of any faces, and this will create a hollow cavity within the solid
Yes cavity is inside but not very easy to select something after that :)

it'a great tool!!!

> Yes cavity is inside but not very easy to select something after that :)

When you create a cavity in this manner (through shell or by boolean), then the cavity becomes just one portion of the entire object, so it is selected if the entire object is selected. So for instance if you want to do a boolean with the object that contains the cavity with another object, you don't have to do anything special to select the inside parts, it's all one object.

But it can be difficult if you want to do something that operates on a single face of the cavity, that's when you'll have to hide some of the outer faces.

- Michael

thx for the precisions!

Pas de problème!

- Michael

Thanks for the trick...
I was not very happy with the "Shell" command so far because I was trying to give depth to objects like the one on the left (see enclosed image). With such object wholly selected, it seems that the computer performs calculations for hours and nothing happens.
With the object on the right, the "Shell" operation was performed in a few seconds (on an old computer...)

Is it normal that MoI performs differently for both objects ? After all, they are very similar in shape.

- Schbeurd

> Is it normal that MoI performs differently for both objects ? After all, they are very similar in shape.

It's not normal to go into an endless loop, that's definitely a bug.

But it would be normal for it to work differently between those 2 objects, with it failing on the left one. I'll explain why.

If you do a "Separate" on that shape, and then offset each surface individually, you can see this is what happens:



And zoomed in:



You can see that those surfaces don't make a clean edge-to-edge intersection there. The surface on the left there (without the curved edge) needs to be extended to make it work properly. The extension mechanism normally works by extending the surface out and then intersecting it with all the other ones that neighbor it. But since one neighboring one is missing in this case, it prevents the extension from working along that side.

Since this is a simple case, if it was smarter it could know how far to extend it here without needing to intersect with the missing surface, but it doesn't work like that because that wouldn't work for the general case where the surfaces are all bendy and curved (it is actually difficult to extend a curvy surface by a constant amount in one direction, it is easier to extend it further and then trim it back).

So basically, having the other surface there helps make the extension-and-trim-back mechanism work properly, and extensions are frequently required when doing a shell. Does that make more sense?


And actually, even if that was able to extend properly, there is another reason why doing it with the solid is better - it creates a cleaner result.

Here you can see a box that was open on one side that was shelled. The side surface in this case is created by doing a bunch of individual lofts between surface edges. This creates several surfaces instead of a single surface, with more edges. These will tend to cause complexities with further operations such as filleting. When you do a shell with a solid box instead, it will trim that surface to make a hole in it, so you'll only have one surface there without the extra edges that are circled.



Does this all make sense? Let me know if you need more info.

- Michael

Morality : Shell prefers to work on "closed volumes" :)

Very practical for make wallls of pieces some complicated :)
Just select some faces & voila :)

Hi Michael,

Very clear and useful information.
It's almost certain that I will come back on this subject later when I will want to give some thickness to more complex objects (what are then the best strategies for "preparing" the model...) but at least I know where to start now !

Many thanks

- Schbeurd