"Mirror Vertex Locations and Weights" Blender script

[QuiteTheTail]

I'm sure it's a powerful tool, but in my hands is completely useless. I have tried to use it on a mirrored bunch of vertices, but once I tested the armor in game the mirrored piece was literally flying above my character's head. In a different occasion, the mirrored piece was ok, and the original part was floating in space. Obviously I'm making a mess of the various options.

 

At least I hope I'm getting right the meaning of Left > Right (and Right > Left). I assume it's referring to the .L (Left) or .R (Right) suffix in the vertex groups. Could someone please shed some light?

 

P.S.

Oh, I forgot to mention the version I'm using: Blender 2.49b with True Normals patch, plus the Blender scripts from kgtools.org

[Symon]

Haven't ever had call to play with that script but from the descriptions of what you see, it sounds like the mirrored pieces might need re-weighting.

[Vaelorian]

Re-weight the mirrored object. If you've duplicated half an armor piece, parts of a body mesh or similar, you may as well just separate it and rename the vert groups, then join it with the main mesh again.

 

It is truly an exceptionally useful tool once you get the hang of it.

[movomo]

If you're talking about kg's mirror weight tool, it's mostly intuitive. Version matters, though. Different versions work differently. There have been multiple versions of the same mirror weights script.

 

You need to be able to tell if the weights are messed or not, in Blender, before testing in game.

 

About its "irregular" behavior... It depends on whether the target vert group has L/R counterpart or not, whether it's bleeding into the other side (crossing the y axis) or not.

 

If it's not a L/R group, such as Spine, it will just mirror it.

If it's a L/R group, the other side will be renamed properly. These groups are not hardcoded. Anything that has .L or .R in their name will work.

If, for example you are mirroring R to L, and R group bleeds into the L side, L group will be bleeding into R side as well in the end result. This is to help painting objects that are supposed to be smooth - like skirt. This was my request, so I'm kind of responsible if this is confusing...

So the problem in this case is, if you completed the R side (Thigh.R of a skirt - bleeding) and want to exactly replicate it on the L side (this means R and L are currently different), the script will kind of "swap" them. Haven't taken a look on the code but this may be a bug. To avoid this you better remove the L group before executing the script.

 

If you're not comfortable with this behavior, you can always do it the traditional way.

[QuiteTheTail]

Thanks Simon, Vaelorian and Movomo for replying.

 

About the last comment, It's not confusing, if by "bleeding" you mean the influence of a left

bone to the right

side of a mesh.

The opposite side, in short. That's how I get what you wrote, but I might be wrong.

So, If I intend to keep it simple, I can just delete the vertex groups with the wrong letter and forget about them (with possible consequences, of corse).

 

I hope to learn more about these concepts, that's why I'm experimenting.

 

P.S.

The traditional way, is it the one suggested by Vaelorian, I suppose?

[movomo]

> The traditional way, is it the one suggested by Vaelorian, I suppose?

 

I never know unless I can read others' mind but there are many ways to do a same thing, as usual.

For example,

 

Paint the Thigh.R as you please.

dupe and mirror the object from R to L.

now rename Thigh.R to Thigh.L in teh duped object, and copy it from the duped object to the original object.

Now you have both Thigh.R and Thigh.L  in the original object!

 

Oh, and, the "right" side refers to the positive x axis. Your 3d model's right arm is the right side and the left arm is the left side.

[QuiteTheTail]

Oh, and, the "right" side refers to the positive x axis. Your 3d model's right arm is the right side and the left arm is the left side.

Just as one would expect when pressing Numpad+1, the front view facing the user. Thanks for the remark.

Wait... I think I'm a bit confused right now