Hi.

I'm working on a game that requires unique usage of the physics engine in order to work. Here is one example:

Normally, the physics engine makes one simulation step per frame. But the project I'm working on demands more flexibility than this. I need to have the physics engine save states (a state is a complete description of the position, velocity, etc. of every physics object), restore states, and perform any number of physics simulation steps per frame, all of which according to a script or DLL.

Here is a simple scenario:
Suppose you wanted to simulate a physics object. You'd have a green cube, and a blue cube. The green cube is a normal physics object, but the blue cube represents where the green cube will be N frames into the future.

When programmed successfully, here are some things you'd see:
- The green cube's movement would generally match that of the blue cube.
- If the blue cube came to a stop, the green cube would come to a stop and land in the exact same position as the blue cube.
- If the user altered the green cube's trajectory, perhaps by pushing it in mid-flight, the blue cube would suddenly move to compensate for the new trajectory.

Here is how it would be programmed:
For each frame:
The physics engine would simulate the green cube, at state X, for N steps and then go back to it's original state X. The position of the green cube after N steps would be where the blue cube will be rendered.

Lite-C is evaluated on the fly, which makes it slower than a pre-compiled DLL. Therefore I'd need this sort of flexibility for DLL-based behaviors, not just for Lite-C.

Is this possible? Please let me know anything you'd think is helpful.

Thanks,

Steve

i think you can do this with A8, because you can calculate your physics with PhysX
i don't know how, but it should work



i don't think so, Lite-C is also compiled and no interpreter language

So it is compiled into binary instructions, in other words? I need clarity on this issue since this is a fairly intensive project programming wise, and I need to make sure I start working with the right language. I'll look through the forums myself, but I'd appreciate input if anyone has any.

Let me answer my own question.

physX_run() lets you tell the physics engine to run whenever you want it to according to the script. When you use physX_open() the physics engine runs once per frame.

As far as state capturing and restoring, I haven't got that part figured out just yet.

A general suggestion: You could also use the C++ SDK to
run the engine.
This way you know excactly how much control you have
about the engine and data.

---


Anyhow, the command "physX_run" might be interesting for your purpose, to run the simulation outside of "framerendering"

Maybe you can combine this with the engine load/save commands, to
save and restore physics states.

I'm planning on using Visual Studio 08 for this. I'm still not 100% sure on how I'd go about doing this but so far it seems fairly clear.

But how would you do this? I mean there are endless possibilites with a physic object. I mean, if you drag the object that is existing in the real time, not the future one: How should the future one move? It doesn't know if you drag it around for a few seconds or if you let it drop, so there are endless ways, how should you future object should now what is going to happen?

Or maybe I dont get the point

its for the user to see whats going to happen if he leaves everything as is, I guess. propably some sort of prediction stuff