Blender mirror modifier bugs

Hey guys,

I would like to listen to your opinion about how mirror modifier should work in jme.

I have made a commit today that fixes several bugs with this modifier, mostly with using an external object as a mirror.
There is still an issue with mirror object rotation, but I hope to work on that soon.
But when I worked on the modifier’s bugs I found out that since blender 2.63 the Mirror modifier merge option works differently.

When you check the option and set a proper tolerance, the modifier will merge all vertices of the original and mirror shapes, that are close enough to the mirror plane.
The tolerance describes how close they should be.
In all blenders before 2.63, the vertices were merged in the place between their positions. Just in the center.
But then it changed and the verts of the original shape are dragged towards the vertices of its mirror.
In other words - the mirrored vertices do not move at all.

I belive it might be a bug and I have already reported it to the blender guys.
But now I do not know what to do with this bug ? :slight_smile:

Should I implement that incorrect behaviour and make the model look like it does in blender ? Or rather compute it correctly as it should be, even if in blender it looks differently ?

2 Likes

I just received an email that they had fixed the bug.
So I assume that since blender 2.70 the behaviuor will get back to normal :slight_smile:

2 Likes

Well, I guess they’ve answered your question then :slight_smile: I wouldn’t implement a workaround for their bug then.

Currently, I’m always applying all modifiers before using the file in JME. Have you ever had a chance at fixing the bug with the “disappearing” faces? In the mirrored half of the model the faces always turned invisible depending on the camera’s angle.

OK, then I will leave it as it is :slight_smile:

@arpagaus
As for that bug I have no idea what might be wrong. I thought that this is the problem of badly computed bounding box.
But this is being always updated.
I will double check it tomorrow, but the bug is cause euther by bad bounding box computing procedure or it has nothing to do with bounding box at all.