Hi there.
I am following a tutorial to build a ionic capital with Nurbs.
However, when I come to the rail revolve step, MoI gives me a different result than Rhino.
Any ideas on how to make it in MoI?
I am attaching the file. I have named the objects (profile, rail, axis) to make the Rail Revolve with.

Thanks.
M.

Rebuild curve

Hi Max, the geometry library that MoI uses originally assumed that the rail curve was a flat curve in the XY plane and ignored z coordinate values, that's why you're getting a flattened result. I made some adjustments to that in MoI v3 to allow a non-planar rail curve but it looks like it's not getting applied in your case here. I'll investigate it.

Thanks for posting the file,
- Michael

Hi
You can use a Sweep with a rectangle and Boolean Diff
See my screenshot

Hi Max, so I investigated this rail revolve issue for quite a while yesterday and today.

Like I wrote before, the geometry library that MoI uses had assumed that the revolve profile curve is planar and in the same plane as the revolve axis, and that the rail curve is planar and normal to the revolve axis, so it ignores z values for the rail curve.

In MoI v3 I tuned this up somewhat by making it recognize z values for a non-planar rail curve but only for the case where you're making a dome like result where the top end of the profile curve touches the revolve axis. So that did not cover your case here.

I've been able to tune this up the rest of the way for v4 so that a non-planar rail like you have here will work the same as Rhino and also profiles that are not in the same plane as the revolve axis will work ok as well. So in v4 this case should work like you were expecting.

Thanks for reporting the bug and posting the example file!

- Michael

it's supposed to make the entiere volume ?

Hi Pilou, no it's not supposed to make a solid in this case - it's supposed to make a surface that follows the rail curve as it rises in the z direction, rather than making it all flat with z = 0 everywhere in the surface as it does currently.

- Michael

No problem Michael, thanks to you.
M.

So like a Double Sweep if the second internal spiral was known ?

Hi Pilou, here's the best way to explain it -

Current result:




Fixed result for v4:




- Michael

but... we can do the same with a simple Sweep :
1• rebuild the curve
2• Sweep , Twist : Flat

Is this the very first sign of v4 life I see before me?


Martin.

@Finema:

> but... we can do the same with a simple Sweep :
> 1• rebuild the curve
> 2• Sweep , Twist : Flat

Yes, it's not a very common use for RailRevolve which is why it hasn't come up before now. But it's not bad for RailRevolve to be able to do this as well. RailRevolve will usually build lighter surfaces than Sweep because it's an exact construction from the input curves and doesn't go through an iterative refinement process like sweep does. Also I guess the results are very slightly different because sweep with twist = flat will rotate around the z axis with the rotation origin point being the point on the rail curve, while rail revolve rotates around the revolve axis only. So although the rotation directions are the same, the rotation origin points are different.


@Martin:

> Is this the very first sign of v4 life I see before me?

I wouldn't read too much into it - I've got it fixed in the old code base right now so that when rail revolve is ported over it will be fixed in v4 when it arrives there.

- Michael