Spiral Stairs and Physics

[Anna Nova]

I'm making a very detailed spiral staircase in Blender 2.8.   I use the array modifier for the steps using a circle as the Object Offset, with a Constant Offset in Z.

All well and good.

But now we get to the physics.

I could make a simplified step and edge, and use the same array modifier, but that will end up with 3 planes per step, and there are 19 of them!  Okay, I can apply the modifier and simplify, but maybe I am missing a trick here.

What would you suggest?

 

[Rolig Loon]

,Maybe just make one plane per step, forget the vertical ones, and then link your final structure in world to a pair of tall, transparent  concentric prim cylinders -- one with a small diameter as the center post and the other as a hollow barrier to keep people from falling off the outside.

[Anna Nova]

25 minutes ago, Rolig Loon said:

,Maybe just make one plane per step, forget the vertical ones, and then link your final structure in world to a pair of tall, transparent  concentric prim cylinders -- one with a small diameter as the center post and the other as a hollow barrier to keep people from falling off the outside.

The outside cylinder would have to start above head-height though....

Otherwise I'd never be able to get in.

 

[Rolig Loon]

Of course.  But then, nobody is going to fall off at the bottom anyway, and you can put a short, cut, hollowed cylinder segment there if you are worried about it.

[Aquila Kytori]

 

5 hours ago, anna2358 said:

What would you suggest? 

Something like the following :

 

[Wulfie Reanimator]

Is it even important to have that detailed of a physics model? (Not sure what the current LI or LI target is.)

You could have fewer triangles than there are steps, and just approximate the overall shape of the spiral. A good example is the outside cylinder @Aquila Kytori made. Extrude the bottom edge to create new faces and merge them at the center pole. You can see that it would cover about 2 steps per plane but is still well within walkable ranges. A simpler, continuous plane like that may resize better to different scales.

Edited June 11, 2019 by Wulfie Reanimator

[Coffee Pancake]

To be honest ,,, if this is just for avatars to walk up and down. This is perhaps one of the times when a a couple of prim spirals might be your best friend.

It's not technically perfect, and for lowest Li might be better unlinked ... or better yet just script the stairs to rez there own physics model and it to move with them, and die when the main goes away. No messy coalesced objects

[ChinRey]

I haven't done a spiral staircase in five years and this isn't really the best solution. If I was to make one today, I would have done the railing physics similar to what Aquila suggested and I might have put the ramp a little bit higher. But the physics weight is still only 1.1 and with a 1.4 download weight there isn't really any point in going lower than that.

The green here is the physics model and if you ignore that mess of disjointed rail physics triangles, you'll see I used a ramp for the steps as Wulfie suggested:

Here's the physics only with the ramp in green. It's only 13 tris and that's actually too much. Looking at it now, I can see how it could easily be done with 11.

You can't usually stand on a staircase with ramp physics, except - in this particular case at some specific points:

But you don't really want to stand still in it anyway, at least not when it's a spiral staircase.

For the center post three triangles is all you need:

 

Edited June 12, 2019 by ChinRey

[Kyrah Abattoir]

That's what I got. But it "is" a spiral staircase after all. If you want collisions you're gonna have to pay the price for it.

[ChinRey]

3 hours ago, Kyrah Abattoir said:

That's what I got. But it "is" a spiral staircase after all. If you want collisions you're gonna have to pay the price for it.

Spiral staircases are exactly the same in SL as in RL: gorgeous looking, compact and horrendously impractical.

In this case there shouldn't be any price to pay though, at least not when it comes to land impact. Look at the two examples above.

Aquila chose to use step phyisics with separate surfaces for each and every step. It's still 1 LI.

Mine was made by a newbie. It's a prim build, converted with MeshStudio. The only thing I did in Blender, was join some vertices on the handrail - that was as far as my Blender skills went back then. It's still 1 LI.

It is possible to make a more efficient physics model but obviously you can't reduce the land imapct any further and the performance improvements will be too minute to bother with even for somebody as obsessed with micro optimisation as I am. We've had a lot of discussions about optimising physics here and maybe we've overdone it. HAVOK isn't nearly as fragile as we may think and usually we can make very good physics models with no performance or LI issues worth mentioning. (It's different on opensim btw. When you're dealing with ubODE  you really have to be careful. There's a reason why so many sim and grid owners there still prefer the more primitive Bullet and ODE physics engines.)

---

With stairs there is another physics issue though, the difference between step physics and ramp physics (I suppose everybody's figured out what those words mean by now). Both have their pros and cons and for a regalr staircase it can be hard to choose. For a spiral staircase however, I'd say the advantages of a continuous smooth walking surface clearly outweighs the disadvantages.

Edited June 12, 2019 by ChinRey
Typos

[Kyrah Abattoir]

Mind you I haven't actually looked at the phys price of mine.

[Aquila Kytori]

@anna2358Combining the wisdom of the forum ...........  VOILA !

The Physics mesh for a Spiral stairway  😉 :

 

 

 

[Anna Nova]

38 minutes ago, Aquila Kytori said:

@anna2358Combining the wisdom of the forum ...........  VOILA !

Thank you everyone.  I am so glad I asked.  Your replies have been most instructive.