Wrong G2Blend result on Twisted Surfaces

 From:  Michael Gibson
7869.6 In reply to 7869.4 
Hi Marco,

> I suppose that the downscaling was necessary because of something about the tolerance,
> but i don't understand why.

Yes, that's correct. The problem was that the different blend surfaces were not fully crossing each other - in some places they would cross each other and in other areas they were a bit apart from one another. The trimming was about trying to get rid of the crossing each other areas. The "too far apart" areas were still out of join tolerance which is 0.005 units. By scaling it down the apart areas were then under 0.005 units apart from one another and so would join. It's possible that just the scaling down alone would do it, maybe the trimming was not needed after all.

This type of blend is a fairly difficult thing, it's possible to do something like force the surfaces to touch by altering the last row of control points so the pieces would touch but at the same time that breaks the continuity right at that same spot too.

Another thing I've thought about trying is to extend the blend surfaces sideways and then I might be able to intersect them better with each other. Someday I'd like to give that method a try.

- Michael