I posted an earlier question about this issue (Boolean applied, but not changing geometry), and using the advice given to me, resolved most of the issues, except the following:

My object to cut: the object

The cutter object: cutter

The two objects overlaid on each other, plus my boolean settings:

boolean settings

Results of the boolean operation:

enter image description here

The geometry of both objects is manifold, per the 3D printing toolbox.


I was able to get this to work two ways:

Set the Overlap threshold to 0

I don't know exactly what this setting does or why it works. I suspect the setting might be related to this old carve bug, given that the bmesh method was recently added to replace carve, but that doesn't explain why it fixes anything in this case. You might want to report a bug with this file, as I'm pretty sure the bmesh method should work without messing with this setting.

Use the Carve boolean method

Your cutter object has a negative Z scale, which, consequently, means the normals are inverted despite appearing correct. This makes carve act up (bmesh is fine with it though). To fix it:

  1. Apply the scale with ⎈ CtrlA
  2. Switch to edit mode (↹ Tab)
  3. Select all (A)
  4. Recalculate normals (⎈ CtrlN)
| improve this answer | |

You say that the two objects are non-manifold, however, the Boolean modifier is only intended to work with manifold meshes. Using non-manifold meshes will generally result in unexpected or unintended behaviour. Make the meshes manifold and try again.

| improve this answer | |
  • $\begingroup$ However, the objects in the OP's .blend are manifold. $\endgroup$ – gandalf3 Dec 24 '17 at 6:04
  • $\begingroup$ @gandalf3 I was working to the OP’s description - specifically stated as non-manifold. Your answer mentions that the Z-scale is negative - doesn’t that in itself make it non-manifold as it would mean the mesh effectively forms the infinite volume outside the mesh since the normals would be pointing in rather than out? $\endgroup$ – Rich Sedman Dec 24 '17 at 9:15
  • 1
    $\begingroup$ Sure, I just figured that if the OP was familiar with the concept of manifolds they probably understand that manifold geometry is required. Hence why I decided to double check the file. As I understand it, manifoldness is (strictly speaking) a property of the surface and where the volume is doesn't matter. Indeed, boolean operations work just fine on inverted geometry, they are merely reversed. $\endgroup$ – gandalf3 Dec 24 '17 at 9:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.