Physic discussion thread

The Procar has a very low diff preload that cannot be changes plus the current LSD implementation is too open on corner exit, which combined with the general tightness of the car mean it doesn't really allow power oversteer induced spins atm.

The 911 Cup can be turned into a drift machine by increasing the preload to ridiculous values, at which point it will spin out in third gear under mild cornering if you floor it.

Supposedly differential fixes are coming, so fingers crossed.

 

I agree! I wanted to add the higher speed tests in case the diff starts working well at low speed, but still locks itself above a certain speed. I’m as interested in that from a car control perspective.

Though even Londrina can test the mysterious high speed locking in the 120km/h sections so kart tracks can indeed be demanding and thorough tests.

The differential allows for so much stabilising inside tire slip right now that it’s no wonder it doesn’t spin. But I have no idea about its tires and such things. Just this one detail that ends up being very important.

You can indeed do this in some cars and it might even be a good idea to increase preload to mitigate oversteer at slow speed when the differential opens. But no way to achieve balanced diff setup atm.

 

Hey great job with the GT1 Mclaren F1, I love it!!

 

In the following videos you can see the main issue with some tires (most of the slicks) in AMS2 that makes the diff issue even worse (making this test before would have saved me a lot of play time...).

As evident to anyone with a bit of (real) driving experience with sporty vehicles, reasonably powered cars should be able to perform donuts (or at least start to rotate a bit) no matter the kind of diff implemented. With an open one it will be much more difficult, the radius you can rotate on will be quite high, but "something" will happen anyway.

Same car: RWD open diff

AMS1 implementation.

Perfect.
Initial promotion of rotation, loss of yaw velocity gain when the engine starts to go toward the limiter but playable with some modulation that will lead to desired rotation.



AMS2 implementation.

Disaster.
No rotation at all, no matter modulation, clutch kick, promotion by inertia...
The car rotates exactly on the same radius no matter if one wheel is on fire or not.

This is exactly what you feel while driving (most of) the cars.
They won't rotate on throttle unless the diff is completely closed.
In that case they start to rotate but in a quite weird way and only because the torque is totally (this is an impression, but it may be less) transferred to the outer wheel causing a yaw momentum.

P.S.
It may be an issue depending also on other parameters: wheel inertia is a possible good candidate as well as chassis inertia but it's much more difficult to go wrong on them...

P.P.S.
As a direct consequence, the same car is a pure joy to drive in AMS1 and horrible in AMS2

 

I wouldn’t write off wheel inertia just yet. The Madness driveline system is complicated.

I don’t know how it should be exactly, but do you think the stabilising effect of longitudinal inside tire slip is too effective in Trainer and Vee? If you get a bit too much body slip angle when lifting, you can straighten then pretty effectively with full throttle. Almost too easily...

Does the unloaded inside tire have too much longitudinal grip even when slipping a lot longitudinally? That would keep the car stable in a donut situation as well. But then why does the differential prefer the inside so much if it still grips?

My logic might easily be false of course. A shot in the dark . However I bet this kind of issue will get sorted out with time and beta testing. Devil is in the details.

 

finally someone who is sharing the feel of the physics "issues", there is a lot of ams1 cars thar just feel odd in ams2 and unfortunately that includes things like differentials and "boat like" suspension.

to the poster above - time and beta testing of what exactly? Game is in a released version already and these things are one of the most important in a driving simulation. A lot of people dont seem to care so luckily (or sadly) there isnt as much noise around this.

 

There’s still a beta test program where they work on physics improvements with Reiza in between public updates. A lot of the noise is hidden there.

But there is definitely noise about the physics. Have you been on YouTube ?

 

@oez No, I would call out too much longi grip on the inner tire. At least not for the cars I’m talking about. As you can see they spin as if they have no grip at all...

 

Oh yeah, that also makes sense to me. Anyway not even remotely my area. I just wondered about the open diff open wheelers.

 

oh right the backer room i've not watched much ams2 on youtube, there isnt really much reason to watch simracing if I can do it myself.

 

I do feel like the Opalas got the short end of the stick in one of the recent tyre updates, along with some other "vintage" cars (Fusca, Mini 65). The Opala 86 used to drive beautifully, almost as well as in AMS1, although I can't remember if it was possible to do donuts with it. Now it feels stiff and unresponsive. And the less said about the Old Stock the better.

I guess we can add the "Donut Test" to the basic checklist of race car physics. Even works for FWD cars, just gotta stick it in reverse first .

 

I can do donuts with the 86 impala with the default settings just fine.

 

Can you show an example? It’s impossible on my build (as evident from the video).

 

I can do very lazy donuts with the Vee. I understand and could verify the difference when trying it in AMS1. Where it starts the donut more aggressively. No data to show.

 

Some fun from your (un)friendly neighborhood a**hole

The cars in the videos below are perfectly identical apart from one parameter.
Both open diff, no additional friction and hard slicks.

Test one (AMS2-Pcars2 default)

Test two (single point modification)

(I exagerated things a bit just for show...)

 

@Avoletta1977 How and what did you do ? Is that the mod car?

 

Yep, mod car.
I played a bit with inertia.
As I said above, it was a good candidate but I'm not fully satisfied with results and I'm not so sure about how the numbers are used by the physics engine so I won't draw any kind of conclusion for the moment.

In any case it would also explain some other oddities so it's worth a few additional tests, maybe reading some numbers from telemetries.

 

It did also cross my mind that maybe the open differential is too optimistic as in lacks mechanical friction. Maybe it's a mix of everything: longi tire slip falloff, wheel inertia, open differential internal friction.

Back when the MCR2000 had a botched open differential (pre 1.0.4), it had very little internal friction. It would just fire up the inside tire against the limiter the moment you leaned a bit on the outside tire with some throttle applied. Obviously the MCR2000 tires have moved on since (I think at least?), but it still makes me wonder if the differential is the only thing that stops it from being a complete one tire fire show.

To me it seems really close now in the Vee and Trainer. A few small compounding tweaks away. But they're still awesome, cause I'm not as picky (that is also justified). I don't know the Stock Car as well as these two so can't comment how well it handles.

 

This is a lap with modified tire inertia and clutch diff as open as possible (0 preload 2 clutches, 90 deg angles) so it should simulate something close to an open diff.



This is the video:

The car is nice to drive but it's like adding almost 200kg (only in rotational inertia) so it can't be considered as a solution.

Point is: by increasing inertia it is simulated some kind of benign differential lock working discretely only when the tire is accelerating.

The reason why some of the cars with slicks at the moment are not able to rotate on throttle seems to be confirmed as excessive reduction of longitudinal grip leading to complete loss of torque in input to the diff. When the slip happens, in addition, tires seems to loose too little lateral grip and the overall result is a car unable to spin.

Some combined intervention should allow to address the issues:
- increase longitudinal grip after peak
- increase lateral slip with longitudinal force
- increase (a bit) wheel inertia
- introduce a small friction between axles even on open diffs (this has been tried but it resulted in a locked diff... I would try again with different parameters).

If you want to try, change the following lines in driveline.rg with the car installed.

# Medium-sized road car wheels. 245/45-18(ish)
RECORD wheel_FR_ROAD_medium shaft
name "wheel_FR"
sim:mass 7.0
RECORD wheel_FL_ROAD_medium shaft
name "wheel_FL"
sim:mass 7.0
RECORD wheel_RR_ROAD_medium shaft
name "wheel_RR"
sim:mass 7.0
RECORD wheel_RL_ROAD_medium shaft
name "wheel_RL"
sim:mass 7.0

 

Looking at the video that Porsche looks like it would become my new favourite car to drive in AMS2! It looks to be doing everything right.
Thanks for the code to change to try it myself.

I'm a complete noob about physics and such, but I hope this is all useful to Reiza, I imagine they are watching here and also testing this and other possible solutions to the way many cars drive being unable to rotate.