Position of PhysicsType vs. position of geometry?

Troubleshooting physics
1

I built a moonlander like model for my physics tests and have the problem that the DynamicPhysicsObject created from it seem not to behave properly.

I positioned the model to have its center at the position which should touch the ground eg built it up from z=0 in local space.

Now i’m experiencing strange behavior based on which way i create the phys Obj i use.


  1. load with bounding property “box” - the model never actually touhes the ground (terrain block) but hoovers like halfe it’s height in air when touched down but when i throw it over it gets stuck like half in the ground.

    setting the localTranslation of the node that holds my model doesn’t help here.


  2. load with bounding property "obb" - the model touches down correctly but the simulation acts strange as on slopes it penetrates the ground rolls around some and then comes to a stand at the correct position.

    But when i throw it over it also sinks some into the ground.


  3. Physics type box - same as 1.


  4. Physics type cylinder - hey the cylinder is oriented wrong eg my lander rolls along on the ground - looks funny



    so i think:

    a) the phys Obj is only computed correctly for complex models when i use an oriented bounding box which gives me trouble at simulation time

    and b) PhysicsTypes should have a way of specifying how the type is oriented and positioned to the geometry/node containing the geometry
2
1) load with bounding property "box" - the model never actually touhes the ground (terrain block) but hoovers like halfe it's height in air when touched down but when i throw it over it gets stuck like half in the ground.
setting the localTranslation of the node that holds my model doesn't help here.

Sounds a bit like shochu's problem. Do you by any chance scale your model through jME?
3

Indeed i do. Which post do You refer to ?

I’ll check it with an unscaled model at home.



edit: ah, think i found it You refer to the updateGeometricsState after the scaling right? - testing that asap

4

Also try rendering the boundings to visually see what the physics is using for edges. That could help explain things.

5

Ok, that did the trick :slight_smile:

Best choice for me so far is obb which behaves like i expected it to, except that on rare ocassions my model seem to sink into the terrain block by a very small amount and get stuck there (maybe the same problem as with the missing triangles) - the thruster force isn’t able to get it loose until some fiddling with the steering thrusters to shake it, it lifts of finally 8-O impressive simulation of swamp terrain at least :wink:



I know that jme physics is in it’s early stages but i want to congrat You guys for what You did so far.

6

try increasing the step iterations and also increasing the step size. This will check the collisions more frequently per update. However, it is slower :?



Thanks for your remark, we appreciate it :slight_smile:



DP