I recently tried to hollow out a sphere by modeling the "void" volume from hand and "subtract" it from the sphere using a Boolean modifier in Difference mode. It turned out however that the "void" object that I created wasn't manifold, so my plan didn't work out (yet). Here is what I did (partly copied from the original post):

Because the void is radially symmetric, I thought maybe I could create it from a 2D mesh that traces the edge of the void and is then spun around 360°, similar to one of those folding lampions that you can buy. So I added a Circle mesh and scaled it to roughly fit the edge of the void. Then I deleted all the vertices of the left half of the circle and manually repositioned the lower group of the remaining vertices to roughly retrace the lower edge of the upper void segment:

enter image description here

Then, I duplicated the mesh, mirrored the duplicate on the z axis, joined the two meshes, and merged the two inner vertices at center to connect the two meshes. Finally, I used the Spin tool to spin the 2D mesh around 360° to create the 3D shape of the void.

enter image description here

But when I try to remove this object from a sphere via the Boolean modifier using the Difference mode, it doesn't work. I learned that this is because the "void" object that I created isn't manifold (I already had a hunch about the reason, but didn't know what it was called). So my first question is: Why is that? Why does the process I described not lead to an object that is manifold?

I managed to "solve" this by switching into Edit Mode and then going to Select -> Select All by Trait -> Non Manifold. This selects the very top and bottom vertices of the object and its central vertex ring (which resulted from merging the two meshes). The top and bottom vertices I can understand, since I assume that there must be a top and bottom vertex from every copy of the initial 2D mesh that were created by using the Spin tool, and which are now lying on top of each other. This issue I managed to solve by merging the respective vertices groups. But I have no idea what is wrong about the central vertex ring. So my second question is: Why is that ring not manifold and how can I make it so?

I tried to create new edges/faces from the vertices or filling them (and some other random stuff), but nothing worked.

So when I try to remove the void object from my sphere (which I created analogous to the void object from a half circle that was spun 235°) via the Boolean modifier using Difference mode, it (unsurprisingly) doesn't work either.

enter image description here

Here is my project file in case you want to play around with it:


1 Answer 1


(Using Blender 3.6.8)

Normals are inverted between the top and the bottom parts, because of the mirror. To visualize normals orientation, open the Viewport Overlays panel and check Face Orientation. To solve this issue, go to Mesh/Normals/Recalculate Outside in Edit Mode.

Inverted normals

There is also an interior face at the junction of the two parts. It is a disk joining the centre ring vertices (in red at the centre of the following figure). This must be deleted also. To track such faces, go to Select/Select All by Trait/Interior Faces in Edit Mode.

Extra face

All of that is solving the Boolean operation issue.

Hollow sphere


  • $\begingroup$ Hi, thanks a lot! By normals, I suppose you mean the vectors that define which direction is inside/outside? I didn't know 2D meshes had such a property! But the explanation makes a lot of sense. But unfortunately, I cannot find the Viewport Overlays panel. Where can I find it? $\endgroup$
    – mapf
    Commented Mar 27 at 16:48
  • $\begingroup$ Sorry, I did manage to do what you say, but for some reason, the Boolean operation still doesn't work properly. The sphere is not really hollowed out, but instead, the silhouette of the void object is cut from the two large faces, leaving the sphere non-manifold. What is wrong? $\endgroup$
    – mapf
    Commented Mar 27 at 19:41
  • $\begingroup$ I just found that if I invert the normals of the sphere to match the direction of the void object (from inward to outward), the sphere remains manifold after the Boolean operation, but only the top part of the void object is cut from the sphere. Really strange. $\endgroup$
    – mapf
    Commented Mar 27 at 19:47
  • $\begingroup$ I added explanation about a interior face at the bottle neck that must be deleted, as well as a link to the documentation about the Viewport Overlays panel (it is the circle and disk icon, in the top right corner of the 3DViewport). $\endgroup$ Commented Mar 27 at 21:09
  • $\begingroup$ This is great, thank you so much! I would have never caught that interior face. $\endgroup$
    – mapf
    Commented Mar 28 at 12:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .