Getting different results from markers created from geometry for opposites sides of the body

Hi there,

just a question. At the moment I’m getting different results for the same body parts that are on the opposite side of the body. The markers are created from the low res geometry version of the rig. I’ve deleted the history, frozen the geometry, and the pivots seem to be in the same place. Everything looks to be identical except they differ massively in the behaviour.

thanks

Gang

Hey Gang!

Are the markers behaving themselves as capsules? It might be a grouping collision thing, left and right side have the same grouping ? Could select all markers and turn collisions off to see if it is collision related .

J

Hi Jason,

thanks, I did try turning off collisions which didn’t seem to have an effect. Marcus, suggested it might be some kind of pivot based issue due to working with the geometry and not joints. All the attribute editor values are the same though so it’s a little baffling. I think I’ll go back to working with joints which seem to work better. I was trying to see if generating the ragdoll setup from the proxy model might be a good way to work being that the import/export for geometry wasn’t working yet. Marcus has said that having the same geometry as in the original scene will maintain the connections. I think there are some python commands that need to be run to make this work that Marcus provided but have yet to test. If that doesn’t work I’ll use the replace mesh command in a small script to run through the markers.
thanks
G

Oh right if you’re going straight to the geo to make the makers ,

  • Make sure the pivot is in the right place .
  • Bake the pivots .
    image

You could also make a joint instead and replace mesh , That might solve any strange pivots the geo might have …

1 Like

Sorry for the late reply! From our private chat yesterday, I’m aware you are assigning directly to geometry. That can work, but has a few gotchas that are not immediately obvious, primarily related to the rotate pivot and scale.

Joints have the lovely advantage of showing you both pivot and hierarchy, and it only has 1 pivot. That’s it, the one you see is the one there is. But transforms, like meshes, have at least 3 all of which are invisible.

Here’s an example of plain meshes, with the screen-left side being scaled negatively to achieve symmetry. Capsules appear when I was expecting a Shape Type = Mesh, and they have the strangest default sizes, this is clearly not correct and something I’ll address for the next release. But manually setting their shapes to Mesh seems to work.

Until you hit play and notice that on one side, the limbs don’t seem connected? :thinking:

Visualising the hierarchy can give you some idea of what’s going on.

And this is what I suspect you are seeing. Pivots are inverted when the scale is negative. This would only affect meshes with scale directly on the transform, but not meshes in a group that’s been scaled, which is typically the case for rigs.

The other thing about Freeze Transform on meshes is that even if it doesn’t seem like there’s a rotate pivot, it actually moves the mesh to the origin and replaces your current visible pivot with a rotate pivot instead. Making the problem worse.

So the best solution (i.e. workaround) is to avoid negative scale on the thing you assign. I believe I can address this on my end also, and will investigate this. Clearly this isn’t working the way it should, sorry for the inconvenience!

Thank you for the explanation Marcus. This is indeed what was happening I suspect!

Thanks Jason.