Big prims

Would there be any reason why we couldn't have prims larger than 10m with the added catch that such prims would not not cross sim borders? They could essentially be stopped in their tracks as you try to move them, or simply returned to inventory.

I am interested in this for 2 reasons:

1. Some large builds that require curved shapes simply don't look as nice when they need to be constructed out of smaller prims. Large spheres and cylinders are good examples.

2. In some cases large prims would reduce lag. One example is when you want to "pave" a very large area with prims. A 50m x 50m area would require 25 prims, each sitting side by side. The problem is each one of those prims has 6 sides to render even if four of those sides are hidden as they are pressed against each other. Using one large prim would reduce the unnecessary overhead.

Sorry if this has been discussed. I scanned this section and didn't see it.

I don't think the problem is the size of the prims per se, but rather, the amount of collision data the server needs to process.

Basically, larger prims = larger chance of multiple collisions = more physics bandwidth needed.

Given there are still lurking errors with collisions that require a rollback of a sim to fix, I'd much rather see them addressed.

Fortunately, they're arcane enough that the standard griefer doesn't *ahem* know about them.



Still, there has been mention of tweaking the physics system for Havok 2 and the next salvo of patches, so something like this might be possible.

That, and you don't want me to be in possesion of any device that allows me to build even *larger*.

Would be nice, Aimee, but I guess it is just the physics side.

the far and few times this subject comes up ive knoticed its becuase of different reasons, last time is was becuase of sim crossings or something....

could we have an offical LL stamp on this?

my bad it was becuase of physics ignore my insanity

The historical reason for only 10 meters was to reduce the ghosting with physics on sim boundries.

And the fact collisions don't exactly work on 100m x 100m prims in general. Once upon a time I found a few set out in Gibson.

Let's just say: fun bugs.

I can get 15x10 prim floors by making a 10x10x.01 prim and setting its top shear to .5 (either X or Y, + or - ), but then I get multiple faces to mess with for texturing, and not only is the connection not perfect to other prims, but sometimes the prim doesn't draw properly by being partly off screen.

I figure a 20x20 prim will drop off screen far too easily. Much though I would like to have bigger prims, I'd rather have more of them instead if it meant better drawing.

Meanwhile, though, it would be nice to have some sort of way to do larger cylinder sections (90° in 10x10)...

Aimee, you have been here long enough to know why prims are set at 10 meters.

The overlap from one sim to another is approximately 10 meters thus explained to me by a linden back in beta.

Creating Larger Prims is like what Jeffery said about collisions.

In beta at one point when a prim crossed over two sim lines it created a Phantom effect where you in essence fell through the prim.

Unfortunately they have yet to overcome this as it requires the sims to not only keep track of whats in them but whats in the 8 surrounding sims up to 15 meters into that sim.

So tell me are we looking to add Lag or decrease it.

While Prims are wonderful tools they are still ancient in their inception.

Combining Prims might be a better option ...Linking is one thing yet the server still has to keep up with each face on every prim. However, Combining would allow the elimination of those additional faces thus creating smoother surfaces and less faces for the server to keep up with.

This might be a better option for them to pursue than to pursue making a laggy prim more laggy.


Think Boolean with a Kick Aimee...

Sincerely, Shadow

Who cares about the physics engine?
Force them to be phantom and I'll build my own collision mask out of invisible cubes if I want any,
If I just want to make movies, or make something that looks good but isnt necessarily going to be walked on, I dont need it to be collidable.
Besides, why should the client-side view necessarily equal the server-side data model?
Split it into 10m prims automatically...
A good system must always include the necessary layers of abstraction so that users needn't be bothered with the system's little technical details.
Sometimes I think LL has no concept of MVC-centric software design.

Shadow, While I didn't know about the collisions issue I did know about the sim border crossing issue as noted in my original post:



While I'm not entirely sure I UNDERSTAND why the collisions on a single 100m x 100m prim would be worse than on 100 10m x 10m prims, I do believe the experts if they say so.