# Height of the RailRevolve lower than the rails

Hi,
Why does RailRevolve result lower than the rail curve, the provile curve or the axis? And how can I solve this in a generic way?
ProblemRailRevolve.gh (34.1 KB)

Cheers,
Bas

Because your section profile’s end is lower than the rail…

ProblemRailRevolve_re.gh (39.7 KB)

Mmmm, the profile curve is a planar curve with all z values zero, so the end point has a z of zero too. But if you mean that the end point of the profile and the rail curve should coincide, than I need to know how I can change the end point of the profile curve. Any idea?

Oh sorry, I see your solution now. Thanks!

I found the gh command seam, which changes the end point of a closed curve. With this command I can coincide the end points of the rail curve and the profile curve. It works when the scaling S is set to false.
RailRevolveWithCoincidingCurves.gh (65.3 KB)

Thanks Joseph,
I think a loft is in principle a simplification of the wanted shape

Both Loft and RailRev are remarkably slow using this section curve, especially when you increase the Divide Curve ‘N’ input to make more loft sections, which affects the result.

Going back to the original code you posted, here is a different way to align the section curve with the rail before using RailRev:

RailRevolve_2021Jan30a.gh (41.9 KB)

I suppose it could also be used to align the rail with the section curve…

P.S. Yes, this one rotates the rail to match the section curve instead:

RailRevolve_2021Jan30b.gh (42.4 KB)

Thanks. If you are interested, here is my complete script: Why this difference with Solid Difference?. But with a problem with solid difference…

Well, that is a much more complex model so I’ll pass on wading into it. Too many components that seem unnecessary at a glance. And redundant. If the two gears are mirror images of each other(?), why not create just one and mirror it? So much code is difficult to follow.

Rail revolution with python is much faster

RailRevolveWithCoincidingCurves_.gh (39.9 KB)

In this case it is a mirror, but in many cases it will not.

Dear Seghier,
Thanks for the tip. Indeed it is. Only 30 ms instead of 6.1 seconds.