Hi Nick, thanks for posting the bug and the example file!
After some examination, it looks like another instance of the underlying problem that causes that 6-turner bug as well.
This one is rather amazing though that the conditions so perfectly align to make the bug happen, since it does check in a lot more than 7 places along the helix curve. But it just works out that all the places that it happens to check are all in tolerance with that unwanted result surface. Here you can see the different stations that are checked in yellow:
In theory you should be able to work around the bug by making an additional copy of the profile up vertically 0.2 units (the helix pitch) to force a tolerance check of the shape at that additional profile location. But that does not work properly right now since it hits a recursion limit due to the pretty long rail. I've raised the recursion limit for the next v2 beta so that this one (with the 2nd added profile) will work properly in the next v2 beta.
The main bug should get fixed when I get a chance to work on sweep some more, which I think won't happen until after the scene browser stuff is complete. Anything that messes with the tolerance checking I need to test rather carefully so that I don't cause the sweeper to get way slower than it already is.
Thanks for the report, I have now got it bundled for testing along with some other sweep stuff.
- Michael |
