Hi,

PML fails to export figures that have vertices which are multiply welded. That means a vertex is shared by 3 or more body parts and these body parts reference each other as weld goals. This is true for a few vertices in rex and roxies fingers, where they touch the hand actor. PML might emit multiple vertices in this situation.

The actor.WeldGoals() API returns multiple weld goals for these vertices. This makes or makes not any sense. I've begged SM for revealing some insight. But maybe we have some people here who applied subdivision to poser figures, maybe they know how to deal with vertices with multiple weld goals? Which is the dominating one?

Or is it necessary to collect all weld references and compute a transitive closure to know which ones belong together?

Thanks, col

Interesting.  How are the weld statements in the cr2 set up?  Are they all in the normal "weld child to parent" form, or are there parent to child welds, as well?

I wonder when they added this capability for welding.  I've always ended up with some unwelded vertices, when trying to handle a multiple actor join like this.

Its sibling welding I think. In my own experiments, like you say, I was unable to reproduce this with the setup room. But when hacking the weld commands in the cr2, poser accepted it. and you can do awful things there.  finding out which vertices are supposed to be the same, gets a quite tricky question.

    weld    lMid0:8  
             lHand:8
    weld    lMid0:8  
             lIndex0:8

this is from roxie. its cross-welding the hand child actors. (lMid and lIndex should be siblings)

    weld    lIndex0:8  
             lHand:8

lIndex is only welded to the hand, not welded to lMid. so it seems to be a directed relation. possibly one can even weld siblings without welding with the parent at all. maybe its impossible to answer to question what is the master vertex python-wise. maybe, one can only evaluate carefully which vertices are supposed to be the same, and hope python will return the same coordinate for each of them. for example, if actor A is welded to B and B is welded to C and C is welded to A and they have a shared vertex, then one has to find this out in python by computing the hull.

How is Python reporting the situation?  I don't recall whether it reports any and all weldGoal actors or just proper parent-child situations.

I don't think I've seen working sibling welds before.  I'm pretty sure that was the case I encountered, in which some vertices simply wouldn't weld.  :unsure:  It is surprising to see something like this now present in one of SM's own figures.

Attached Link: Welding illegitimate children

http://www.renderosity.com/mod/forumpro/showthread.php?message_id=3210468&ebot_calc_page#message_3210468

Oh yes. Phantom was a true explorer. Wondering where he has gone.

Me too

Quote - Oh yes. Phantom was a true explorer. Wondering where he has gone.

Phantom had serious health problems.

I have made a fix, PML is hopefully now Rex/Roxie capable. But I had to rework lots of the exporter code because the problems in the poser API could not be clarified, so I'll distribute 2012.1.39 only on request for the moment.