Physics test

[BlueCat]

Sorry about the openal error. My fault again ops:

I've put a copy of openal32.dll here and it's in the main zip file now too.

dUmAsS: I'm limiting the fps so that I can use more cpu time on the physics updates. Can you tell me how you've implemented suspension, steering etc. My car has 11 bodies and 10 joints, an F1 car must be pretty difficult to simulate :?:

It is easy to flip the car but the only way I've managed to make it stable is to make the centre of gravity impossibly low (so far below the car it's under the ground!) and this, unsurprisingly, gives you the effect of a weight 'swinging' from the car. It keeps the car upright but I've gone to the trouble of simulating real physics, I'd rather not make it artificially less realistic. If you change the suspension height/power and the ground friction levels you get different car handling. Like in the real world, if a car keeps falling over you make a better car, you don't have the option of changing the laws of physics

Thanks to everyone for testing, the frame rate shouldn't affect the physics quite so much now as I'm using a larger timestep.

[tux]

im not using ode to simulate the suspension in the f1 car (basicly because it hardly moves )

i use ode physics for soft suspension. try changing the slip of the tires

i just use 4 hinge 2 joints for suspension and sphears for tires and 1 box for the chassis.

have you got the bendy wheel problem yet?

[Snape]

p4 2.0 ghz and Geforce4Ti4400: ~30 fps @ 1280@1024@32 (with mcafee and firewall running).
Pretty cool thing. I've also tried to implement DelphiODE a few weeks ago into a small sample but I stopped it after some hours because it was too complicated for me (okay, maybe I wasn't motivated).
May you post the source code or a quick tutorial?

[BlueCat]

dUmAsS: I guess you're using a hinge2 joint? I did get the bendy wheel problem with those joints especially when using CFM to set suspension 'softness'. Now I'm using (for the front wheels), a slider for the suspension springs (need damping!), a hinge for steering and a hinge for the wheel axis. There are small bodies between each of these joints, I've used small boxes with low density just to connect them.

It works nicely but does stress the CPU with more than one car and large timesteps just make it all go crazy

Snape: Thanks for the results. I asked for the fps to see at what level it starts to drop below 30fps, it won't go much faster than that on any machine because I've limited the rendering. I know exactly what you mean about ODE being complicated, one of the hardest things was extracting a body's matrix and converting to use in glMultMatrix, not nice. OK if you're a mathematical genius, or if you at least know one!!! Unfortunately I'm not so I just stayed up late for a few nights until I stumbled across the solution through trial and error :lol:

The best way to get started is to play around with the demos and change them. I used the terrain demo and the buggy demo and combined some of the code to work with my terrain engine.

I have been thinking about writing a tutorial and will whenever I have the time (if ever :? ).

The only thing I have trouble with now is updating the physics at 150+ updates per second while rendering several hundred thousand triangles at an acceptable rate. OK if you have a high enough spec but below a certain speed it seems that what I'm trying to do just isn't practical. I only have a Geforce2 mx though, hardly the latest thing

I'm working on turning this into a game now and will just have to decide on the minimum requirements to run it. Hopefully I won't exclude too many people.

[Harry Hunt]

Athlon 1 GHz, Radeon 9000 Pro 128 MB here....
I get 24 FPS most of the time.

Looking pretty good. Kinda hard to play but then again you wouldn't expect a smooth ride on that kind of terrain. Good job!

[Crisp_N_Dry]

Athlon 2800+, 512mb RAM, GeForce FX5800 and I only manage 18-19 fps at 1024x768x32 (Desktop Resolution). I'm pretty sure it can do better than that. Any ideas what's wrong?

[tux]

dUmAsS: I guess you're using a hinge2 joint? I did get the bendy wheel problem with those joints especially when using CFM to set suspension 'softness'. Now I'm using (for the front wheels), a slider for the suspension springs (need damping!), a hinge for steering and a hinge for the wheel axis. There are small bodies between each of these joints, I've used small boxes with low density just to connect them.

It works nicely but does stress the CPU with more than one car and large timesteps just make it all go crazy

yeah, but thats no option for me. with maximum 54 cars on track at any 1 time (im making lemans not f1 )

you using tri collider?

[tux]

works now. but no sound (is tere ment to be).


ure physics work pretty well

[BlueCat]

The sound is not so good at the moment, I've fixed some bugs so that it at least works.
You've probably got the version with the strangely rotating sky, that's fixed now too.

I've simplified the suspension a little now too and removed 2 bodies and 2 joints to speed things up.

I'm working on fixing the frame rate and physics update rate to work on slower machines. Will upload a new version when it's worth doing so

And no, I'm not using tri-colliders. Have a look at the 'terrain demo' that comes with DelphiODE. It uses a function to get the height of the terrain. I think it's one of those nice GLScene terrain thingies. I don't use GLScene so I've just modified the function to work with my engine.

Thanks for trying it again

[tux]

i went to the openal site and dled the openal installer. sound works.

nice engine




to unlock the physics from fps (so u could have 1 or 1000000000000 fps and physics the same ) use the main program loop to do the gfx and set a tthread at interval 10 to do physics update thats how im doing it and running pretty well