--- Operating System, Graphics card ---
Win7 64
Intel pentium 2.3ghz dual core
Intel HD gbraphics (mb integrated)
--- Blender version with error, and version that worked ---
Blender 2.67B
Blender 2.66A worked
--- Short description of error ---
reversal of boolean
--- Steps for others to reproduce the error (preferably based on attached .blend file) ---
Create a complex non geometric shape, nonuniformed, such as a car body. Then create a cylinder, deform it like a teardrop, then use a boolean modifier on the body, set to difference of the cylinder. The operations have reversed to a difference performing as a union. The union will act as a difference.
Able to reproduce repeatedly with complex shapes. Not present in 2.66a or prior.
Description
Event Timeline
Hi Kevin,
this is most likely caused by your open mesh... just have a look here http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.62/Boolean_Modifier and you'll see that the carve booleans will not work reliably in this case.
I closed your mesh and everything was alright! But I'm sure Sergey can enlighten you a bit more..
Greetings, Thomas
Booleans are requiring volumes, aka closed manifolds. There're some workarounds in the code to support open manifolds, but in the example it's really unsupported usage. Math fails to figure out inner and outer volumes for the mesh.
It is known limitation, so thanks for the report, but it's not a bug.
the effect working perfectly fine in reverse application when outside a closed mesh is not a bug?
The difference modifier performing as a union when applied on an open mesh. The union modifier operates as a difference modifier when outside the closed meshing. The modifier still works, the operators have reversed is all.
I wont argue its a bug or a limitation, you guys are the masters of this stuff, I just use it.
Also to note: the same file I have attached, I closed the mesh, it worked fine. however, when I operated on a open mesh, and then closed the mesh the operators were again reversed. This was with the cylinder dead center on the object.
If its a limitation, ok. If its a bug, ok.
Thanks for assisting. This can only make it better.
It's not reversed, it's mathematically unpredictable result in case of open manifolds. At least with assumptions made in Carve library which makes it behave really fast and robust for closed manifolds.
And check on the normals of your mesh which you created from open one. They all shall point outside. Otherwise you've got inversed manifold and for sure boolean operations wouldn't behave the way you expect them to behave.