Tyre friction/interaction with road surface

[T1specialist]

This post is not about ffb but has to do with the tires. The way I see it rfactor based sims output worse visual feedback than some others. Rf2 ffb is pretty good. But everytime I drive any rfactor based sim, be it rfactor, raceroom, rf2, ams the visual feedback is always missing something. It is much harder to feel the tire in rf based sims. I don't want to mention any other sims because it seems to trigger people into ignoring everything else I say. But I just find it odd that for me this is true for all rf based sims while all other sims I've ever played seem fine providing good visual feedback. I know there are other people like me who have this same feeling about rf based sims.

I don't really know why that is. I can set up ffb fine. I'm not clipping and I know my way around wheelcheck, minforces and dampings. Ffb is fine even though in rf2 light cars suffer from spiky ffb. It is not about fov or cockpit motion. It is about the raw data that comes through the screen which you can use to sense the car as it leans, rolls, tires deflect in all directions. Basically how the cockpit view moves as it is rigidly attached to the car.

Because rf code base is really old I've always thought this is because rf probably doesn't take everything into consideration when it draws the cockpit view so it can save a few fps. For example rf2 has tire deflection but is that taken into consideration when the car (and the cockpit view) visually rolls in corners? Or the nose dives under braking, or the rear squats? How the driven tires deform when going over the peak slip angle? Those minute things that seem really small but are essential when trying to feel the car at the limit. This could mean the visual car behaviour is simply missing the cues that come from the tires for example.

Or it could be the way the car slides. Or glides at times. The moment when the car is about to grip it is really difficult to detect without ffb. Same thing with accelerating out of corners and getting wheelspin or under hard braking. For me without ffb it is super difficult to drive rf based sims at the limit. It is a lot of guesswork and you never really know how hard you can really get on throttle or yaw and rotate the car into a corner.

 

[T1specialist]

I actually always set all helmet movement to 0 because I find it distracting. What I meant by visual feedback is the car itself and how it leans, rolls and etc. (and how the cockpitview then does the same as it is rigidly mounted on the car). To me it feels something is missing.

Also I disagree with you completely about gran turismo cockpitview. It is incredibly laggy (I think you get 30fps in cockpit view and 60 in bonnet view) and just odd .

As for steering geometry I'd find it extremely odd if ackerman was not simulated in the steering.

 

[Emery]

Hopefully someone can provide some info regarding the strange behaviour change when reaching 200 kph.

Offhand, I would think aerodynamic downforce begins influencing results at around 200 kph? We know, for instance, that aerodynamics won't affect drag until 100 kph and it takes a car shaped like a brick to notice much drag before 200 kph.

 

[Korva7]

For what I have seen, the simulation of tires themselves seems to be correctly done. However, it would seem that there is no proper steering mechanism linkage simulation. Apparently rF2 calculates tire positions based on reduction of the steering system and controller.json modifiers.

The fact that suggests me this, is that cars with no convergence trend to maintain similar slip angles for both tires even at in turn situation. This would imply a "perfect"steering system where both wheels would be permanently pointing at the right direction.

Yep, And with convergence MAX the slip angle difference between front and left front tire seems to stay approximately the same in cornering and on straight. I think that in formula there should be some anti ackerman which would lead to smaller slip angles for the inner front tire in cornering.

 

[Korva7]

Offhand, I would think aerodynamic downforce begins influencing results at around 200 kph? We know, for instance, that aerodynamics won't affect drag until 100 kph and it takes a car shaped like a brick to notice much drag before 200 kph.

That weirdness was because speed sensitivity wasn't at 0. So steering lock increased as speed droped under 200 kph

 

[Korva7]

This post is not about ffb but has to do with the tires. The way I see it rfactor based sims output worse visual feedback than some others. Rf2 ffb is pretty good. But everytime I drive any rfactor based sim, be it rfactor, raceroom, rf2, ams the visual feedback is always missing something. It is much harder to feel the tire in rf based sims. I don't want to mention any other sims because it seems to trigger people into ignoring everything else I say. But I just find it odd that for me this is true for all rf based sims while all other sims I've ever played seem fine providing good visual feedback. I know there are other people like me who have this same feeling about rf based sims.
About rf2 being more difficult to drive without ffb: I think that there is just more stuff being simulated in rf2 and you


I don't really know why that is. I can set up ffb fine. I'm not clipping and I know my way around wheelcheck, minforces and dampings. Ffb is fine even though in rf2 light cars suffer from spiky ffb. It is not about fov or cockpit motion. It is about the raw data that comes through the screen which you can use to sense the car as it leans, rolls, tires deflect in all directions. Basically how the cockpit view moves as it is rigidly attached to the car.

Because rf code base is really old I've always thought this is because rf probably doesn't take everything into consideration when it draws the cockpit view so it can save a few fps. For example rf2 has tire deflection but is that taken into consideration when the car (and the cockpit view) visually rolls in corners? Or the nose dives under braking, or the rear squats? How the driven tires deform when going over the peak slip angle? Those minute things that seem really small but are essential when trying to feel the car at the limit. This could mean the visual car behaviour is simply missing the cues that come from the tires for example.

Or it could be the way the car slides. Or glides at times. The moment when the car is about to grip it is really difficult to detect without ffb. Same thing with accelerating out of corners and getting wheelspin or under hard braking. For me without ffb it is super difficult to drive rf based sims at the limit. It is a lot of guesswork and you never really know how hard you can really get on throttle or yaw and rotate the car into a corner.

If you were speaking about ac i would agree that there is more camera movement in ac overal, but i think that it is probably because of differeces in road bumpiness and tire, suspension and car frame flex physics.
About rf2 being more difficult to drive without ffb: I think that in rf2 physics simulation goes more "deep" and there is stuff happening beyond what you can see and feel without g-forces and seat of the pants feeling.

 

[peterchen]

This post is not about ffb but has to do with the tires. The way I see it rfactor based sims output worse visual feedback than some others. Rf2 ffb is pretty good. But everytime I drive any rfactor based sim, be it rfactor, raceroom, rf2, ams the visual feedback is always missing something. It is much harder to feel the tire in rf based sims. I don't want to mention any other sims because it seems to trigger people into ignoring everything else I say. But I just find it odd that for me this is true for all rf based sims while all other sims I've ever played seem fine providing good visual feedback. I know there are other people like me who have this same feeling about rf based sims.

I don't really know why that is. I can set up ffb fine. I'm not clipping and I know my way around wheelcheck, minforces and dampings. Ffb is fine even though in rf2 light cars suffer from spiky ffb. It is not about fov or cockpit motion. It is about the raw data that comes through the screen which you can use to sense the car as it leans, rolls, tires deflect in all directions. Basically how the cockpit view moves as it is rigidly attached to the car.

Because rf code base is really old I've always thought this is because rf probably doesn't take everything into consideration when it draws the cockpit view so it can save a few fps. For example rf2 has tire deflection but is that taken into consideration when the car (and the cockpit view) visually rolls in corners? Or the nose dives under braking, or the rear squats? How the driven tires deform when going over the peak slip angle? Those minute things that seem really small but are essential when trying to feel the car at the limit. This could mean the visual car behaviour is simply missing the cues that come from the tires for example.

Or it could be the way the car slides. Or glides at times. The moment when the car is about to grip it is really difficult to detect without ffb. Same thing with accelerating out of corners and getting wheelspin or under hard braking. For me without ffb it is super difficult to drive rf based sims at the limit. It is a lot of guesswork and you never really know how hard you can really get on throttle or yaw and rotate the car into a corner.

Try in PLAYER.JSON

"Graphic Options":{
"Cockpit Vibration Freq1":5,
"Cockpit Vibration Freq1#":"Primary rate of vibration affects eyepoint position (higher framerates allow higher rates)",
"Cockpit Vibration Freq2":5,
"Cockpit Vibration Freq2#":"Secondary rate of vibration affects eyepoint orientation",
"Cockpit Vibration Mult1":0.1,
"Cockpit Vibration Mult1#":"Primary aerodynamic vibration multiplier affects eyepoint position (base magnitude is in VEH or cockpit file)",
"Cockpit Vibration Mult2":0.1,
"Cockpit Vibration Mult2#":"Secondary aerodynamic vibration multiplier affects eyepoint orientation (base magnitude is in VEH or cockpit file)",

Exaggerate Yaw=0.45000


Then report!

 

[green serpent]

This post is not about ffb but has to do with the tires. The way I see it rfactor based sims output worse visual feedback than some others. Rf2 ffb is pretty good. But everytime I drive any rfactor based sim, be it rfactor, raceroom, rf2, ams the visual feedback is always missing something. It is much harder to feel the tire in rf based sims. I don't want to mention any other sims because it seems to trigger people into ignoring everything else I say. But I just find it odd that for me this is true for all rf based sims while all other sims I've ever played seem fine providing good visual feedback. I know there are other people like me who have this same feeling about rf based sims.

I tend to agree with what you're saying here. I drive more from the ffb than the visuals. For example if the car is beginning to rotate, in other sims I would immediately correct based on visuals, however in rf2 sometimes it's better to not go into opposite lock and to do more of a power slide with the front tires pointing in the direction of the turn. So I don't go off visuals so much as ffb. Also when transitioning weight from one side of the car to the other at high speed, I purely rely on the ffb to tell me when it's safe to transition the weight. This is why I think a better wheel is so vital to get the best experience from rf2.

I also agree that from the cockpit with head movement set to zero, the car doesn't appear to be moving around as much as other sims. For example, I don't really get that sensation of going over a bump/crest, and then the car bobbing up and down a few times as the springs/shocks absorb the bump. There seems to be less vertical oscillation in the body.

If you've ever set the camera in iracing to be fixed, it's literally insane the amount of vertical/longitudinal movement in the mx-5. The car is bouncing all over the place. I think if it was set like that by default, people would be getting motion sickness. Not saying one is right and the other is wrong, and of course set up would greatly influence these things.

 

[Euskotracks]

I have repeated the tests with steering speed sensituvity set to 0 and reducing steering lock from 90 to 60.
The document compiling all the results has been updated with those.
No strange behaviour is observed for the different parameters.

Caster proves to have a huge effect on alignment torque. However, it would seem that abusing caster would yield a higher tire degradation, providing as well a smaller yaw rate. Working with mid to low caster values should apparently be preferred since it also provides a more constant value throughout the full speed range. If the feeling is too light FFB multiplier should be used with no apparent risk of clipping at high speeds.

https://www.dropbox.com/s/sx5b6gd792tiizd/Tires and FFB.odt?dl=0

 

[T1specialist]

Generally the biggest effect for steering is imho the scrub radius. Probably not adjustable unless you have different wheels with different offsets or different front uprights.

Try in PLAYER.JSON

"Graphic Options":{
"Cockpit Vibration Freq1":5,
"Cockpit Vibration Freq1#":"Primary rate of vibration affects eyepoint position (higher framerates allow higher rates)",
"Cockpit Vibration Freq2":5,
"Cockpit Vibration Freq2#":"Secondary rate of vibration affects eyepoint orientation",
"Cockpit Vibration Mult1":0.1,
"Cockpit Vibration Mult1#":"Primary aerodynamic vibration multiplier affects eyepoint position (base magnitude is in VEH or cockpit file)",
"Cockpit Vibration Mult2":0.1,
"Cockpit Vibration Mult2#":"Secondary aerodynamic vibration multiplier affects eyepoint orientation (base magnitude is in VEH or cockpit file)",

Exaggerate Yaw=0.45000


Then report!

Done ages ago but thanks anyways . I did set everything to 0 because I don't like camera movement.

I think as far as the visual behavior of the cars in rF2, it seems pretty accurate to me. It's probably not exactly what the physics are doing, but it's probably very close. Do you use a realistic FOV? If you use a higher than realistic FOV, then the movement won't look realistic.

Fov makes no difference about it. It is the car cockpit view (and all onboard views) motion itself that just feels like it doesn't convey all what the car is doing in rf based sims. I'm not saying there is anything wrong with physics. I think physics are great!

My guess is that the part of the code that handles the cockpitview orientation doesn't take all physics stuff into consideration. The physics are still calculated of course and used for the car handling but when calculating the pitch, roll, yaw, vertical, lateral and longitudinal orientation for the cockpitview - my thinking is that some of the less noticable stuff could be left out (on purpose) because it can save few fps. For the car rolling in corner for example the tire deflection is pretty small effect compared to the springs deflecting.

As for gt6 I can't really comment about it that much. I did play gt5 quite a bit (it was imho fun simcade racer-ish game). And at the time it was the only game with very accurate nordschleife. I was wrong about the 30fps in cockpitview though. I checked some digitalfoundry youtube vids and it did seem to achieve pretty consistant 60fps. But driving it was just too laggy for me.

Also great post @green serpent. I agree 100%.

 

[peterchen]

I did set everything to 0 because I don't like camera movement..

Then don´t complain about too less camera movement!!

 

[T1specialist]

Then don´t complain about too less camera movement!!

Without adding any additional vibrations or shakes to the camera you are getting just the raw car motions through the view. But this raw information in rf based sims is what I find lacking. It doesn't give as much information as other sims when you set all motions to 0. I'm not talking about cameras shaking. The raw motion itself doesn't seem to carry the full physical motions of the car. But something less.

Of course the FOV makes a difference.

What I meant was not that fov is meaningless. Fov is important. What I wanted to say is that this issue I'm describing is not about fov. Like I said rf based sims only suffer from this. If it was something as simple as fov I'd had fixed this on my 10 years ago.

I'm not sure what you're saying about the cockpit view orientation. The view is locked to the car itself and when you turn off head movement you're seeing exactly what the car is doing.

And what I'm trying to say is that the view doesn't show exactly what the car is doing. There is something missing. My guess is that there could be some dampening to it perhaps that you can't adjust (the whole car movement is slowed a bit maybe), it could be that smaller details are left out from the equations (like tire deflection). Or something else.

 

[Euskotracks]

Without adding any additional vibrations or shakes to the camera you are getting just the raw car motions through the view. But this raw information in rf based sims is what I find lacking. It doesn't give as much information as other sims when you set all motions to 0. I'm not talking about cameras shaking. The raw motion itself doesn't seem to carry the full physical motions of the car. But something less.



What I meant was not that fov is meaningless. Fov is important. What I wanted to say is that this issue I'm describing is not about fov. Like I said rf based sims only suffer from this. If it was something as simple as fov I'd had fixed this on my 10 years ago.



And what I'm trying to say is that the view doesn't show exactly what the car is doing. There is something missing. My guess is that there could be some dampening to it perhaps that you can't adjust (the whole car movement is slowed a bit maybe), it could be that smaller details are left out from the equations (like tire deflection). Or something else.

I could check that convergence isn't being represented graphically. Surprising since camber is indeed represented and changes can be appreciated.

Considering this, it cannot be discarded that some other visual representations are not in agreement with the physics running behind.

Ackermann isn't properly visualized either. I checked ISI's karts where you can select three different steering geometries and it was a simple animation being quite generous. In Formula ISI both wheels are turning the same angle visually but using zero convergence settings the slip angle is similar in both front tires which proves that in the sim both wheels have rotated differently. Otherwise, they would show different slip angles.

Testing things is the only way to prove things since all the doubts settled regarding simulation aspects are rarely clarified by the devs.

 

[Euskotracks]

Regarding the ackermann geometry in Formula ISI.
I have made a test that proves that both steered wheels are rotating the same angle. I made the folowing test and represented its results graphically:

Case 1:
Steering Lock 8.2 deg
Speed: 32 kph

Case 2:
Steering Lock 8.2 deg
Speed: 61 kph

In both cases, turning radius is quite similar but slip angles are quite different as it can be checked in the telemetry.
I found it interesting to prove that it is compatible to turn at an almost constant turning radius while increasing slip angle due to speed.
The reason is because the centre of rotation displaces forward as slip angle increases.

In the image, the lines representing slip angles have been accurately drawn matching exact slip angles as reported in the telemetry aside. Case 1 is represented in red and case 2 in dark grey.
The intersection between the four slip line axes is almost perfect for both studied cases. Beware that before rotating them front axis were parallel. The obtained radius differ in about 0,25 m with respect to telemetry calculated radius.

However, the zero slip angle axes represented in blue, clearly show an offset. between front wheel intersection points with rear axis. If ackerman was uesd, the angular difference to be applied to each wheel so that they would intersect at the rear axis would cause that 4 slip angle axes wouldn't intersect altogether. This proves that in internal physics both wheels are pointing at the same direction when steering Formula ISI.

 

[Euskotracks]

I just drove the DW12 with max caster and the steering does go lighter during understeer, or at least that's what it appears from observation and feel. I can see the FFB dropping slightly in the Pedal overlay plugin and i can feel it as well.

You should still be able to notice this yourself with the same plugin even if you can't feel the difference from your wheel. Also, like i said before, the forces don't go away completely, it drops maybe 20% or less usually, depending on the exact situation.

The sort of understeer you describe in other sims happens in rF2, but like you said the car will eventually start to oversteer if the loads on the car change. Usually by applying more throttle you would actually prevent oversteer, not induce it, just as long as you're not breaking rear tire traction, so what you're saying actually makes sense compared to reality. If you want it to oversteer, then lifting the throttle, more often than not would be what you want to do, since that would shift the weight forwards. The only way applying more throttle would make the car oversteer is if you're breaking rear traction, which isn't going to happen instantly, or at all, depending on the car and the situation.

Here's a video showing the car understeering mid corner through most corners (on the last lap, starting around 16:36), this is partly due to the setup, but mostly from my driving. I'm oversteering into the corners by braking as late as possible, then letting the weight shift to make it more neutral in order to use all 4 tires as much as possible to rotate the car, and then using the rear tires as much as possible to get the best acceleration, which is making the rear tires lose traction causing it to oversteer, but it wouldn't oversteer if i applied throttle too early, i still have to make the car rotate first and shift the weight around, which is why you'll usually see me straighten the steering at the same moment that i start applying throttle. However, if you look at the exit of the corner at 17:32, you'll see that i'm applying throttle and the car is understeering, that understeer actually started 2 corners back because of the way i entered that earlier left hander. I took the other corners after it as fast as i thought was possible given that mistake which resulted in understeer in the third corner, but had i taken that first corner differently, i wouldn't have understeered on that third corner.

So things aren't, and shouldn't be as simple as the car either understeering or oversteering when you steer more or less or whatever.



This is pretty interesting if completely accurate. It could explain why the default FFB feels so unrealistic, and maybe even why the cars sometimes seem as if they want to travel in the forward facing direction during slides.

Is it possible that, that particular car is actually supposed to have parallel steering? Have you tested with other cars?

I also tested flat6 and it showed similar results to formula ISI in the other test. However that test had very small steering lock so it is nit accurate.

I have to test if under low speed it behaves the same but I would bet it also does.

 

[Korva7]

Regarding the ackermann geometry in Formula ISI.
I have made a test that proves that both steered wheels are rotating the same angle. I made the folowing test and represented its results graphically:

Case 1:
Steering Lock 8.2 deg
Speed: 32 kph

Case 2:
Steering Lock 8.2 deg
Speed: 61 kph

In both cases, turning radius is quite similar but slip angles are quite different as it can be checked in the telemetry.
I found it interesting to prove that it is compatible to turn at an almost constant turning radius while increasing slip angle due to speed.
The reason is because the centre of rotation displaces forward as slip angle increases.

In the image, the lines representing slip angles have been accurately drawn matching exact slip angles as reported in the telemetry aside. Case 1 is represented in red and case 2 in dark grey.
The intersection between the four slip line axes is almost perfect for both studied cases. Beware that before rotating them front axis were parallel. The obtained radius differ in about 0,25 m with respect to telemetry calculated radius.

However, the zero slip angle axes represented in blue, clearly show an offset. between front wheel intersection points with rear axis. If ackerman was uesd, the angular difference to be applied to each wheel so that they would intersect at the rear axis would cause that 4 slip angle axes wouldn't intersect altogether. This proves that in internal physics both wheels are pointing at the same direction when steering Formula ISI.
View attachment 4057

Does this mean that the car have "perfect" ackerman, so that with 0 toe you would get about the same slip angle for both front tyres?
If it is, you could try different kind of cars to find out if there is cars with anti ackerman ( smaller slip angle for inner tire in corner with 0 toe ), or the opposite of that ( higher slip angle for inner tire in corner with 0 toe ). If not, then the steering linkage simulation is lacking a little.

 

[Euskotracks]

Does this mean that the car have "perfect" ackerman, so that with 0 toe you would get about the same slip angle for both front tyres?
If it is, you could try different kind of cars to find out if there is cars with anti ackerman ( smaller slip angle for inner tire in corner with 0 toe ), or the opposite of that ( higher slip angle for inner tire in corner with 0 toe ). If not, then the steering linkage simulation is lacking a little.

No. It actually means it would be using rectangular steering geometry where both wheels rotates the same angle.
As I said I will try Flat6 first. Probably later I will check how kart steering geometry behaves since it has ackermann. In that one, clearly wheels rotate differently.

If my intuition is right Flat 6 will show the same behaviour as Formula ISI. For karts I expect to get a constant relationship between both wheel rotation which is what I perceive in the visuals.

 

[Comante]

Karts are very wide in relation to turn radius, so, ackerman impact on wheels should be pretty evident. Cars with larger turn radius have front tires relatively near eachother, so unless ackermann is not extreme, could be difficult to see, you should check at maximum steering angle to see it best.

 

[Euskotracks]

Karts are very wide in relation to turn radius, so, ackerman impact on wheels should be pretty evident. Cars with larger turn radius have front tires relatively near eachother, so unless ackermann is not extreme, could be difficult to see, you should check at maximum steering angle to see it best.

I know. That's why I did the test for Formula ISI at low speed and low turning radius. I used the highest steering lock I had set for the "modified for testing" vehicle which was at 210 degrees for the wheel and 8.2 degrees for the tires.

I may increase this value and check again to make it more evident. However, the intersection of the different lines using slip angles was already much better compared to the lack of intersection of the lines with zero slip angle which was the initial reference before applying slip angles.

As you suggest, increasing steering lock will increase the misalignment of the initial reference and probably the precision of the aligned slip angle lines will get better.

What I want to test with karts is how is ackerman concept implemented. Oppositely to Formula ISI, the initial quick tests, showed very different slip angles for front tires which would suggest a significant angle difference between both front tires. I am presuming that I wiil find a very poor Ackerman implementetion. I suspect that ackermann is being simplified to a different steering ratio for inner and outer wheel which would remain constant in half steering range.

The following picture tries to explain how I think ackermann is being implemented compared to formula ISI.

 

[Korva7]

This proves that in internal physics both wheels are pointing at the same direction when steering Formula ISI.

You mean that front wheels are parallel to each other in cornering? I think this would lead to very big difference in slip angle between front tyres, and i think your measurements doesn't support this. Especially the earlier ones.
Without ackerman the inner tire would propably get slipangle to opposite direction than outer tire in tight low speed cornering.