Physics of Kunos made cars

[elloLeo Kinnunen]

KS cars are off because the dev cycle is incredibly short and nobody has time to do research in that time.

Hi,

From time to time I read variations on the above comment and have often found myself wondering which Kunos cars are better or worse in terms of aligning with the data that others have come across. With apologies for lacking the effort and time to learn how collect and interpret the data myself and thanks to those that can and have, has anyone ever compiled this as a list?

I'm not across all the different websites and discord channels these days, but if this discussion has been had somewhere else, would be grateful if anyone could point me in the right direction.

Many thanks

 

[Kyuubeey]

For the long answer, as a rule of thumb, you should stay away from any racecars, especially GT, Cup and Formula types. Japanese pack cars are very bad. The Yellowbird is a quite serious offender as are seemingly all of the historic racecars, especially the cigar shapes.

Every one I've looked at is quite bad with various implementation errors or lack of research, but I mostly do Japanese cars, so I can't testify for all of them. It is possible some of the cars that I haven't looked at are good, but I can't confirm either way.

Probably the best one that comes to mind would be the NA Miata, but that is based on PURE's work so it is quite ironic. It's also not completely correctly implemented.

The tires in all KS cars are quite bad for 2021 standards, with most of the parameters being outside of likely bounds and some parameters are used incorrectly, like the 20x~ too high rolling resistance slip gain, which makes large slipratio slides very inaccurate and generally bogs down the car in wheelspin. Speed sensitivity is also significantly too high, around 0.035 when you'd probably want closer to 0.020. The effect is not linear at all so it is very strong by default. I also believe the flex_gain and relaxation length values for most road tires are somewhat too low.

Aris also didn't seem to bother to factor in installation stiffness for the stabilizers in any of the cars, so many of them have an incredibly front biased roll stiffness that likely isn't correct.

The short answer is that use what you like, but none of them are particularly accurately implemented.

Lack of good heating is also a quite serious issue in vanilla cars, which seems to still be the case in ACC. Not sure you can fault the car physics maker for that, though.

 

[Cote Dazur]

the above comment and have often found myself wondering which Kunos cars are better or worse in terms of aligning with the data that others have come across.

First, to answer your question, very few physic modders can claim to be able to produce "better" or more to the point more accurate physic than what we have from Kunos in AC.
Not more than 5, I can only think of 3. So beside for the amazing work of a very few, the Kunos physic is the best we can get, and when comparing to other developers work, we are very fortunate to have access to Kunos work in AC.

As per the latest Youtube from Neils, among other interesting comments from that esteemed physic developers, data is very hard to get, accurate and relevant data even harder when at all available,.
All the data? Not possible, which leads to interpretation and "making" the car drive how it should, or at least how the developers think it should.

To me, accurate data is very important for a SIM, without it, there is no SIM, just a game.
But SIM is not a pure science either, more like a craft, pure data as to be complemented by how the car should react when driven. In a perfect engineered world, data would suffice, but in this imperfect SIM world, data is one part of the equation, essential, but just a part.

The search for perfection does not lead to perfection but can get us to a better place, with more accurate car behavior, as long as we still have fun driving them, no harm done.

A lot of my favorite sim cars are in AC and have been developed by Kunos/Aris and that is all that matter to me at the end.

 

[Michael Krone]

Heh car go vroom...

 

[AccAkut]

These inaccuracies are rather well known. But I wonder if there's any racing sim actually offering more accurate content? I think with RF2 it's that the engine potentially allows for better accuracy, but especially for the needed tire data you'd need entry to the R&D departments of Michelin, Goodyear or whatnot?

 

[Kyuubeey]

These inaccuracies are rather well known. But I wonder if there's any racing sim actually offering more accurate content? I think with RF2 it's that the engine potentially allows for better accuracy, but especially for the needed tire data you'd need entry to the R&D departments of Michelin, Goodyear or whatnot?

Hah, good one. There's nothing inherently better about rF2's engine, in many ways the physical tire model is a hinderance. You'll get a far better result using any empiric tire model. All the physical tire model does is generate luts so that you don't "need" to test tires, in theory.

In practice you will test tires anyway and painfully attempt to keep the tire model within the parameters, which is effectively impossible.

I don't think any of the content in any of the consumer sims is super good, all of them have some kind of significant issue or another. Probably some cars in AMS1 and suspension wise in iRacing are good, though.