[GT5 physics] Internal friction and rolling friction

[KinLM]

There is no "first". It'll get done in parallel with everything else, if they plan to do it, that is.

Looking at the evolution of past iterations, I'm not so sure about that. But okay.

 

[Griffith500]

Looking at the evolution of past iterations, I'm not so sure about that. But okay.

I have no idea what evidence you think you might be drawing from, but there is no way the physics guys will be sitting around twiddling their thumbs whist everyone else is fixing other stuff "first".

 

[KinLM]

I have no idea what evidence you think you might be drawing from, but there is no way the physics guys will be sitting around twiddling their thumbs whist everyone else is fixing other stuff "first".

The physics have hardly evolved in so long. GT5 was a major step forward and there still is no real tire model. What Im saying is, PD as a group needs to fix stuff like AI and standard cars before they start worrying about wheel inertia. The physics people need to work on crash physics before they think about wheel inertia. They need to get a better suspension simulation before they worry about wheel inertia. What I'm saying is, PD need to prioritize well or else we'll have good wheel inertia but 800 standard cars still.

 

[Samus]

But what he is saying is the people doing physics are not the same people doing AI or car modelling.

 

[Griffith500]

The physics have hardly evolved in so long. GT5 was a major step forward and there still is no real tire model. What Im saying is, PD as a group needs to fix stuff like AI and standard cars before they start worrying about wheel inertia. The physics people need to work on crash physics before they think about wheel inertia. They need to get a better suspension simulation before they worry about wheel inertia. What I'm saying is, PD need to prioritize well or else we'll have good wheel inertia but 800 standard cars still.

The physics have been significantly different in every single game, even the "demos".

Their approach to physics has been roughly the same since GT3, with things added on here and there, but the underlying engine handles all the inputs and outputs in roughly the same way. Some of the updates to GT5 have shown interesting developments, like the one that improved the suspension and steering geometry modeling (easy things to plug-in, I guess).

It still needs a different approach from the ground-up, though.

Also, what Simon said. What needs doing "first" is a new physics engine that better represents the distributed (rather than centralised or aggregated, as appears to be the case in GT5) nature of the forces that act on a vehicle; then it needs testing and development; then the cars need to be converted to work with it; then it needs further balancing and tweaking; then it can be included in a release version. It's hard to know where, if anywhere, PD are in that process, but the plug-in upgrades to GT5 seem to suggest that they're at least at the prototyping stage of something...

Oh, and wheel inertia is a part of the suspension modeling, because without it the dynamics are wrong - you've probably heard of "unsprung weight", and inertialess in this context implies massless.

 

[KinLM]

But what he is saying is the people doing physics are not the same people doing AI or car modelling.

Yes I know. I'm also saying that there's other, more major flaws with the physics that need fixing first.

Wheel inertia specifically falls behind accurate torque steer and others is what I'm saying. Yes they could fix both of these things by doing an actual tire and suspension model, but this is PD. Lately known for not doing 100% of everything.

My point is, if they're not going to completely fix everything with the physics, if they're not going to do a proper tire model, then there's certain things that I feel deserve a higher priority than the wheel inertia.

 

[Exorcet]

The physics people need to work on crash physics before they think about wheel inertia.

I would put wheel intertia first, it would be part of the crash physics anyway. Crashing is all about energy going this way and that, and the wheels, because of their inertia, serve as energy banks.

My point is, if they're not going to completely fix everything with the physics, if they're not going to do a proper tire model, then there's certain things that I feel deserve a higher priority than the wheel inertia.

Wheel inertia isn't a little thing.

 

[KinLM]

I would put wheel intertia first, it would be part of the crash physics anyway. Crashing is all about energy going this way and that, and the wheels, because of their inertia, serve as energy banks.

Part of the crash physics, yes.

Wheel inertia isn't a little thing.

I stated earlier on that I feel it's quite a major thing actually. It's far from a little thing, it takes away from the realism in many ways.

 

[Griffith500]

Yes I know. I'm also saying that there's other, more major flaws with the physics that need fixing first.

Wheel inertia specifically falls behind accurate torque steer and others is what I'm saying. Yes they could fix both of these things by doing an actual tire and suspension model, but this is PD. Lately known for not doing 100% of everything.

My point is, if they're not going to completely fix everything with the physics, if they're not going to do a proper tire model, then there's certain things that I feel deserve a higher priority than the wheel inertia.

What I think you may be failing to realise is that this is all interconnected. A tyre model will bolt onto whatever dynamic simulation method / implementation you're using, and is independent of whether that's for a solid, rigid box with no suspension, or if it's for a car made of jelly (or jello, whatever).

Subtle things like accurate torque steer fall naturally out of a proper, distributed simulation - where instead of all the properties to be simulated being lumped together in a block, they are placed where they are on the car and are allowed to behave as independent objects, with the proper constraints and inter-connections that are present in the real thing. Then the whole thing is simulated as a system, and it self-interacts and does interesting stuff all on its own - that "soft-body" physics demo that was derived from the Rigs of Rods code is effectively an abstract, low-level (i.e. highly distributed) form of this general approach.

If the underlying dynamic simulation of the objects you are in direct control of (hint: it's not the tyres) is fundamentally flawed, that is the first thing you should fix. Even if you then go on to use the same tyre model (retuned to suit), unless it's really bad, it will still result in a better driving experience.

That is, unless, realism is not the aim.

 

[MrWednesday]

What Im saying is, PD as a group needs to fix stuff like AI and standard cars before they start worrying about wheel inertia. The physics people need to work on crash physics before they think about wheel inertia.

AI, OK. Standard cars and crash physics, I couldn't disagree more. Standard cars are an aesthetic argument that has nothing to do with how they actually drive (unless you want to argue about how they actually drive, which would be a novel argument as far as I can recall). Since, ideally, crash physics will never actually be exercised (because, ideally, we never crash), I want the normal driving physics to be as realistic as possible before they even think about doing anything beyond the bare minimum with crashes.

You're not convinced that altering a global friction component in the physics model has wide ranging effects? I'm not saying it's smack you in the face obvious, I said it was subtle. Many such subtle omissions all add up to create what I called the "uncanny" effect; physics that are not convincingly real, but your brain still accepts it because it's a game. If the drag is not being applied to the wheels, how is that in any way a correct simulation?

It's arguably correct because the resistance in a well-adjusted wheel bearing is arguably negligible compared with other drag and resistance forces on the wheels and in the drivetrain. I'm by no means arguing that PD gets all of those correct, only that I think that the specific case of wheel bearings is negligible and potentially subject to being rolled into some kind of aggregate resistance.

For the wheels to stop immediately, either inertia is zero, or tyre grip is infinite - so it's pretty obvious what's happening. Modeling the inertia in wheels creates all sorts of subtle handling (and braking...) effects, especially through the force feedback (if it's being done properly).

I agree that there's a real need to get various kinds of rotating mass right, and not just in the wheels. There are several upgrades that are supposed to reduce rotating mass in the drivetrain; has anyone tested their actual impact to see if they work the way they should?