In my last post I quoted the statement: "The vertices on the group boundaries would have to match exactly for welding to work".

This is not the first time that I have heard this statement, and I quoted it assuming it was true, but not actually knowing this from personal experience. But something was nagging at the back of my mind. Just how exact is "exactly"?

I decided to do some tests to see if Poser welding allowed any tolerance. This relates to welding done via weld statements in a Poser 6 cr2. I constructed a figure made up of 5 actors (+ BODY) stacked vertically. 5 cubes stacked one on top of the other were used as the geometry. I did a number of tests varying the vertical separation between the cubes, to see which ones would weld and which would not. The 5 cubes were exported from Poser as a single unwelded obj.

In the tests the cubes were yTranslated to check if they were welded. The results of the first test were as follows. Cubes with a vertical separation of 0.000010 PNU or less welded, cubes with a vertical separation of 0.000011 or greater did not weld.

For the second test, I introduced an offset along the X axis of the same magnitude as the vertical separation. A cube with both 0.000010 vertical separation and the same amount of x offset partially welded (see image above), cubes with less separation exhibited a complete weld, and those with more exhibited no welding.