9 replies to this topic

[SpookyCat]

Description
Test of a fully procedural tyre system for our racing games. The tyre mesh is generated from a cross section and various parameters including pressure, tyre wall stiffness, rim diameter, width and ratio etc. The tyre is then capable in realtime of being deformed by contact with a surface and from various forces, so the contact path can move as can the whole tyre. Pressure variations can also be shown eg running totally flat with a puncture. All the deformations are done on the GPU if desired or CPU. A video can be seen Here

[geon]

Cool, but its't that *way* too much effort spent on a vey minor effect?

[PaloDeQueso]

I disagree. If this is a racing game, there are a lot of people who pay very close attention to those sorts of details. Look at the Gran Turrismo Series. I for one am not a fan of racers, however I applaud anyone who pushes simulation to the next level! These games do not have as large of a scope as other games sometimes... like fallout3 or something like that, so they can pay close attention to these details. I'm also not a sports game fan, but look at the latest basketball titles. They all seem to have cloth simulation for uniforms. It just adds so much. Once again, great work duder!

[tobeythorn]

Is this intended to be just eye candy or does this make vehicles drive more realistic since you can take into account the varying surface contact with the road?

[rouncer]

cool!!!!

[Mihail121]

I don't find very much sense in this project.

[phresnel]

Very cool stuff. This is the type of effort that is also put into the next generation of racing simulators like e.g. rFactor 2.

I guess ppl disliking this are not into race-sims or motorsports (Mihail, geon). To give an extreme example, think of drag racing. Drag Racing is very much a tyre-sport, where the tyre-wall-stiffness is usually void and where the tyres get there stability from the radial force only.

Also, think of how world detail improves. Soon enough there will be small stones and other stuff on the tracks, and a punctured, flexible tyre will show a completely different reaction than a stiffly pumped one when driving over such things. This is especially, but not exclusively relevant in rallying games.


In general, *if* this is just a very minor effect, then why do *you* care about the tyre-pressure in *your* (real-life) car?


Surely, games like Codemaster's Formula 1 2010 or DIRT wouldn't profit much, but then they are arcade racers and not sims like rFactor or netKar Pro.

[geon]

> In general, *if* this is just a very minor effect, then why do *you* care about the tyre-pressure in *your* (real-life) car?

Fuel consumption, comfort and safety.

That the effect is done on GPU implies that it is purely visual, and thus won't even be seen (from inside the car) in a racing simulator.

If you actually want to simulate the tire *physics* you don't need the mesh deformation.

[phresnel]

geon said:

> In general, *if* this is just a very minor effect, then why do *you* care about the tyre-pressure in *your* (real-life) car?

Fuel consumption, comfort and safety.

Safety. Exactly what I say, in that it is just another word for human-predictable-car-physics ;)

geon said:

That the effect is done on GPU implies that it is purely visual

Of course I don't know implementation specifics. Maybe SpookyCat batch-processes physics on GPGPU?

Quote

and thus won't even be seen (from inside the car) in a racing simulator.

Surely one can see tyres in every open-wheel discipline, on some motorcycles, in some top-fuel cars, and many more.

Quote

If you actually want to simulate the tire *physics* you don't need the mesh deformation.

Quote

The tyre is then capable in realtime of being deformed by contact with a surface and from various forces

I think you need very well the mesh-deformation for exact mesh-mesh-intersection.


And even if all this does change physics by just a fraction, this can make a significant difference in sales and target-consumer-group. Also, some games like rFactor can't be exact enough, in that they also sell their product to professional racing teams (see e.g. Adrian Neweys [Red Bull Racing] quote on http://rfactor-pro.com/).

Apart from that all, SpookyCat is not alone: Live for Speed article about new tyre model -> http://www.lfs.net/?...=report_dec2010 .

[geon]

> And even if all this does change physics by just a fraction

I think an analytical solution (a set of equations) would be a better solution, since physics simulations are difficult enough to get stable. Relying on the simulated mesh deformation for friction and spring constant calculation seems like a bad idea to me.

But I could be wrong. (Happens all the time...) It would be interesting to see an actual vehicle simulation taking wheel deformation on the mesh level into account.