Physic shape don't work properly

[MaxTux Wonder]

i try to upload a really simple mesh, and create a physic shape with surface analysis "high" and semplification "retain" as in the picture belove, but when i walk over my mesh, i step in middle air...

 

 

using SecondLife-i686-2.7.6.234701

the object is in secondlife://Aditi/secondlife/Mesh Sandbox 1/155/48/23 

[Ashasekayi Ra]

After rezzing, did you change the physics shape to "prim" in the edit window?

[MaxTux Wonder]

i did it, and do not works

[Drongle McMahon]

Strange. It looks like you are standing on the convex hull of the whole object. Can you set it to Prim, then back to Convex Hull, looking in Develop->Render Metadata->Phyisics Shapes in between, and describe what differences you see?

[MaxTux Wonder]

convex hull:

Prim:

[Drongle McMahon]

Hmm. Very odd.  If that uploader picture you show is the actual decomposition you uploaded, then the Prim shape is not following it properly. On the other hand, if that is before the simplify, then maybe the simplify may have changed the shape to what you see inworld. I suggst you start again and upload with and without Simplify. Compare standing on both and compare the physics shape display with what you saw in the uploader as you pressed the uplod button. If you get a physics Prim-type shape that doesn't match the hulls shown on the preview when you upload, then it would indicate a bug and would need to be submitted as a jira.

[MaxTux Wonder]

thank you now works!

now i didn't analyzed and simplified!

 

[Drongle McMahon]

But be careful! That red colour means the physics cost is high. That's because there are lots of small triangles in the spherical part. That should be reduced if you use analyse but not simplify. Even that is not ideal. because the convex hull for the spherical part will still have many points. If you want to make an accurate shape, but not so expensive, then you should make a special mesh just for that. I think your sphere is a subsurfed or multires'd cube? In that case, the physics mesh can use one with one less level. Even better might be a UV sphere with, say, six segments and six rings. Make the same object with the simplified sphere, then use Analyze but not simplify.

[MaxTux Wonder]

here the bug report:

https://jira.secondlife.com/browse/CTS-682

 

[Falcon Linden]

I can't repro this problem and have commented on CTS-682. However, I just wanted to throw in a tip here:

The best way to represent a sphere in the physics engine is to actually use a sphere prim. I'd recommend using a physics mesh without the sphere part, doing a decomposition, uploading, and then linked to a 100% transparent sphere to get the physics right.

[Drongle McMahon]

But, Falcon,  since linking a standard prim to a mesh now makes it get a mesh-like PE, that's not practical for objects that sort of size. For instance, I just tried llinking a 3m diameter standard prim sphere to a mesh and the PE went from 2 to 4. Make that 6 m and the PE goes to 7. You have to reduce the mesh and prim spheres to less than 1.7m before there is no PE penalty. (Cubes have small enough streaming cost to be ok though, as are cylinders up to a much larger size.) So as long as they are in a situation where PE > physics cost nobody is going to use this method with spheres. This physics shape method was already compromised when the minimum PE per linked prim went from 0.1 to 0.5 (CTS-563). Now it is worse. That's another unfortunate result of the decision to punish the linking of standard prims to meshes. I did a jira (CTS-662) about other ways it stops us using nice physics shapes, but it was closed.

Note: the numbers here were from observations on MS 21 which is running the experimental server where the diameter-instead-of-radius bug (sorry no link, jira disppeared) has been fixed.

PS. The other thing with this approach is that the physics mesh gets stretched so that it's bounding box fits the bounding box of the high LOD mesh. So if taking the sphere away leaves that part of the physics mesh empty, you have to put something there to make it have the right bounding box. This is not diificult, but is easy to miss. In this case, just stretch the conical part a bit.