“ are rfactor2 physics broken” video

Discussion in 'General Discussion' started by GTClub_wajdi, Dec 29, 2020.

  1. Nieubermesch

    Nieubermesch Registered

    Joined:
    May 18, 2019
    Messages:
    635
    Likes Received:
    257
    Ahah, but you mean real life too? The M8 I really like the behavior, GTE while the Aston Martin I don't quite like so much.

    I was thinking of the Norma, but honestly, since then and I am with it in my front (I wished real life.. ahah) I am thinking that this car isn't that powerful and maybe those big tank slappers I was expecting aren't there because the downforce hasn't already risen enough and I am in between mechanical and aero grip. Oreca LMP2 is quite unforgiving when it goes, it really goes...
     
  2. Yzangard

    Yzangard Registered

    Joined:
    Dec 6, 2018
    Messages:
    351
    Likes Received:
    173
    Well, actually it isn't that obvious after all, I've searched many "scientific" research on the matter and it seems we all have a misleading assumption about the grip loss after a slide, even the temperature isn't THAT compromising the grip, not as much as we think anyway.

    When looking at resulting grip curves, it appears that rFactor 2 isn't that bad, many results tend to show that there is indeed an "ideal" slip angle and you lose grip if you go beyond it but you still grip, the curve tends to flatten after some point, it is not like in iRacing where you instantly lose 95% of the grip, you may lose something like 30% with maybe 3° more slip angle and then, you will lose way less the more you add slip angle.


    Something like this :

    upload_2021-1-21_19-9-8.png

    upload_2021-1-21_19-8-39.png


    Actual grip curves aren't exactly like those but you get the idea.
     

    Attached Files:

    Last edited: Jan 21, 2021
    mantasisg and Nieubermesch like this.
  3. Bruno Gil

    Bruno Gil Registered

    Joined:
    Oct 12, 2020
    Messages:
    115
    Likes Received:
    73
    I just did a Q session this afternoon with the Norma (and I've driven it before), and in my experience, as soon as you start going as fast as the car allows you too, a bit too much slippage and you lose tons of time or the rear end. I don't think it's nearly as forgiving at the limit as the GTs. And what I said was specifically for this games GT cars, not real life. Although gt cars in real life are indeed easy mode race cars.
    Yeah I agree with everything, the point I'm making is that, at the other thread, we're starting to arrive to the conclusion that a) the ideal slip angle for the GT cars might be too big and b) the drop off curve after that seems seems very flat

    This might explain why Michelin said the tire wear was too big: if the tire allows more slip angle, and because of that, probably more performance than it should, it will also scrub more, having more wear due to Friction

    This might even explain why the GT cars lap spa a few seconds faster than real life counterparts, or acc for example
     
    Last edited: Jan 21, 2021
    Nieubermesch likes this.
  4. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,418
    Likes Received:
    3,279
    It is possible to drive fast with vacuum in tires in rF2. Atmosferic pressure is 100kpa. IIRIC minimum in S397 cars is restricted to be 140kpa, not much for heavy high downforce cars, but it is something, I don't remeber how big are hot tire pressures. Seems to be as declared by the rules of these racing series that I just accidently googled: https://www.gt-world-challenge-asia...lancpain GT World Challenge Asia GT3 2019.pdf

    Thats what happens with vacuum and buckling structures
     
    Nieubermesch likes this.
  5. Stefan_L_01

    Stefan_L_01 Registered

    Joined:
    Nov 18, 2012
    Messages:
    586
    Likes Received:
    377
    Hi
    I think the 2 videos alltogether proove that there is no grip peak. You turn the wheel more, you get more grip overall. Or let me say: You create more lateral forces by more steering. We should not forget rf2 is a physical based sim. Its all about forces.

    Note also that for 2nd video, I could go 7km/h faster (higher centrifugal forces) by applying much more power/torque on the rear to overcome the much higher resistance sliding by the front tires. In first trial I said I used about 60% throttle. In 2nd trial after getting the car into a kind of steady state (not so easy) I applied 90-100% throttle! The front wheels get lighter with more power/torque, less grip available, keep this in mind!

    All together we must conclude it does not make sense to go faster the way I did in 2nd vid.

    So we should ask why tire slip angles far higher than known for race tires allow for higher lateral forces keeping the car turning.

    There are 2 options imo:
    1) The slip angle/ grip model of tire it completly unreal. There is no peak in grip at a given slip angle, quickly dropping off beyond. This would be a big disappointment to say the least. I just dont want to believe it, but its a valid theory

    2) An error in the physical engine which creates forces in a non-realistic way.
    I thought about possible error cause and came to a theory what could be the reason (just a hypothesis!):
    First of all I demonstrated the effect on the front wheels. And steering. So we have a common problem: coordinate transformation. You have your car coordinate system, and your wheel coordinate system, both are rotated to each each other in case of the front wheels during turning.
    Next I said allready that there is happening a huge resistance on the front tire turning the steering too much like I did in 2nd trial going faster. For me this resistance can be the only source for forces that make your car turning as they might be the source for WRONG forces overcoming the grip loss of the tire (which might be modelled correctly).
    So when the sliding resistance is opposite to car movement, and someone transforms the forces into tire coordinate system and applies the transformed force accidently in the car coordinate system, you can create forces which will keep the car turning and overcome the grip loss. Why should someone transform the force into the tire coordinate system? Well they might need it for tire deformation calcs, whatever.
    In the following image I tried to demonstrate the effect of such a hyphothetical error, you can see that at the end a part of the original vector (lila) is helping the car to turn after a coordinate transformation and an apply in the wrong system.
    Just a theory of course.
    BUT: Every kind of drag or so due to sliding will usually create UNDERSTEER and definitly will NOT help the car to turn more

    BR
     

    Attached Files:

    Last edited: Jan 21, 2021
    Pawel44 and Nieubermesch like this.
  6. Slip_Angel

    Slip_Angel Registered

    Joined:
    Jun 15, 2019
    Messages:
    588
    Likes Received:
    297
    I have a good video to show how edgy a GT3 car can be and how quick driver hands moves to catch the slides.


    I will probably get hate for saying this but from watching some iracing lamborghini onboards it looks quite close.
     
  7. Comante

    Comante Registered

    Joined:
    Nov 20, 2013
    Messages:
    1,409
    Likes Received:
    919
    Ok now I throw a rock in the pond, then fly away: I describe this scenario, made a quick sketch to demonstrate it.
    The car is making a right turn, it is settled in the turn, so, we have left tires more loaded than inner tires, before going ahead, filling the picture, this mean that in the rear axle, the tires are able to put into the ground a different amount of force, if the car had no other tires touching, this would create a torque force, just a simple vectorial sum.
    If we analyze the front axle, I drawed the drag forces only, that happen when the tires are turned too much and act more like a rubber block than a rubber tire, in this scenario, if the car, again was only 2 front tires, the torque would turn the car away from the turn.
    If the sum both axes and keep into consideration the slip angle, then we notice, especially in the outer tires, that we have opposing forces that are offset of a certain amount, depending car lenght and slip angle. Those vector create again a torque that help the car turn inside the turn.
    Of course this is an oversemplification, because front tires keep the capacity to revolve, thus the vector direction is not just opposite to the direction of movement, and is of course not as long as the throttle vector, but this could help explain how a car can still turn inside even if with drag forces at play.
     

    Attached Files:

  8. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,418
    Likes Received:
    3,279
    Awesome onboard, I love thsoe quick sharp steering inputs, catching all those many little slides quickly.

    Speakign of iRacing, I think most cars there works pretty much right, they do look pretty awesome in slow motion replays, all the subtle slides are there, car kinematics looks real... They do fail at something else that can't be seen, which makes many of their cars there overly demanding.
     
  9. Nieubermesch

    Nieubermesch Registered

    Joined:
    May 18, 2019
    Messages:
    635
    Likes Received:
    257
    This again, the second theory, reminds me of the theories of Spinelli, a former banned member here. Saying there was a certain bug with "yaw physics" present since early iterations of Isi physics engine.
     
    Pawel44 likes this.
  10. Nieubermesch

    Nieubermesch Registered

    Joined:
    May 18, 2019
    Messages:
    635
    Likes Received:
    257
    https://forum.studio-397.com/index....ng-fr3-5-atm-one-of-the-worst-sim-cars.47243/

    "First, I believe the following to be a list of issues that plague either the ISI physics engine as a whole (going back to even before RF1, but also, slightly continually improved ever since, all the way up to and including rFactor 2), or just specifically the FR3.5 itself:



    1.
    More of the vehicle's momentum/inertia should continue forward in the original direction of travel while the rear-end swings out, rather than so much of the vehicle's direction of travel changing so drastically as if you had a ton of grip and literally turned the car sharply left or right.

    It's like the front-end just turns-in sharper and sharper with too much of the vehicle's original direction of travel changing to, and therefore following, the direction that the front-end is pointing towards. You can even easily see it when watching from a 3rd person, or trackside/TV cam; the car hardly looks like it's drifting, but rather just following the nose."

    Just the first words. Read this thread, maybe there is something to look at in there.

    Edit:

    Just another version of what he means:

    "I'm embarrassed that I made my friend buy rFactor 2. Everytime we go online, no one's on. The cars drive atrociously and slip and slide around as if there is a huge hole in the physics engine.

    All the typical ISI engine holes from the 1990s still exist like vehicles behaving like there is no weight in the physics engine, like there is no mass being pressed into the ground. Like there is no forward momentum. Or, as my friend put it: "Like you're being pulled by a rope from another car in front of you".

    You can even cleary see all that from the way the vehicle behaves. You can go around corners using like 20% the amount of steering lock that a real life driver needs to go around the corner because of what I explained above.

    Low speed slip/grip behaviour is still terrible and it's like the tyres turn into cement.

    Unless you're understeering, then you're just trying to prevent the car from turning in more rather than steering it yourself since it's often like there's no forward momentum. You're often trying to prevent the car from turning (or turning too much), rather than actually trying to make it turn, because of this massive weight/moment/being-pulled-by-a-string issue that's been around in the ISI engine for 20 years.

    If you drive a soft/old car or a road-car then it's like the physics run in slow motion or something. Everything is imprecise and like it's from a different planet. It's like the physics are of a different universe. In Assetto Corsa, you can tell the physics are trying to simulate actual vehicles, and on planet Earth at that.

    The ISI engine has tremendous potential, but it needs lots of fixes which surprisingly still exist after 15 or 20 years and complete ruin the driving experience for me personally, my friend, and every single person who's ever come over and driven at my place. The glaring holes in the physics engine need to be seriously fixed. You can clearly see and sense these issues in any game with any ISI physics engine coding in it including R3E and PCars."
     
    Last edited: Jan 21, 2021
    Pawel44 likes this.
  11. Slip_Angel

    Slip_Angel Registered

    Joined:
    Jun 15, 2019
    Messages:
    588
    Likes Received:
    297
    I sort of agree to this. what i observe in other sims and IRL is that whenever driver want to turn the car body itself looks like it wants to go ahead but wheels are generating force to create roatation. I see too much VISUAL yaw in both onboard and exterior cam in RF2 comapred to any sim and IRL footage.
    The feel that whole mass of vehical wants to go straight but the forces generated at tyres is "forcing" the car to turn, they seemed "detached " IRL and other sim. RF2 car's body is very willing to follow the tyres. it looks quite unrealistic IMO but driving wise cars feels believable. but again we got problem in driving end also so...yea RF2 seems in tough spot.
     
    Last edited: Jan 21, 2021
    R.Noctua and Nieubermesch like this.
  12. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,418
    Likes Received:
    3,279
    Then you should drive some car physics I make, I could prove you its not rF2. Or some particular other rF2 cars that does not have "must drift to grip" handling. They exist. But you are not using mods bla bla bla... also apparently you are too fancy to drive some old content... :/

    I agree any car on solid non deformable surface should have good grip and steady firm rotation under the limit, or even near the limit. It should be expected to start moving around near the peak, and obviously past the peak slip curves. Actually even drifting way over the limit should sometimes in particular circumstances end well if enough angle and speed can be scrubbed of by sliding to return to non sliding sate, there seems to be one sim that fails quite a bit at that.... though should not gain pace that way with modern stuff...



    If that does not happen, it means only one thing - insufficient static friction being generated at the contact patch to make the car bite in, instead of slipping in excessive way. It also means car is depending too much on sliding friction.

    One way to make car a lot easier is to reduce static vs sliding friction ratio (and it is a bit more complex than just altering sliding and static friction values that are at the top of realtme section in TGM). However, it also makes cars prone to too much "diagonal" movement. Which can be perceived visually, it is good for us as a way to evaluate such potential issue. Easy to see when it becomes very unnatural. Obviously increasing static vs sliding friction ratio brings the opposite effects. This is basic principle any simulation physics artist will adjust to make some or even maybe all cars less or more hardcore to drive. And I have read in some iRacing devs blog that it seems to be unknown what exact parameters the tires has IRL at really extreme conditions, because they just don't test that far, as they aren't supposed to operate there in the first place.

    And finally it all comes down to slip angles of which everyone is talking about. Slip curve is a result of tire contact patch loosing static contact area. It becomes quite clear that car will be the easier at abusing slip, the less significant will be change of friction once the last bit of static contact patch area disappears, and tire will become fully sliding...
     
    Last edited: Jan 21, 2021
  13. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,418
    Likes Received:
    3,279
    Thats interesting, but I don't see how these shows anything much more than just slip angle itself and slip ratio in other graph... I don't really see how this truly reflects temperature effect, also don't really see how it reflects the velocity effect. And of course the real friction curve will be sum of slip ratio, and slip angle, so it will drop harder then. It also can drop harder if sudden unloading happens (due to weight transfer or change in aerodynamic efficiency). With all variables in check the grip curve might happen to be rather flat, or it can drop down quite a bit. What will happen entirely depends on so many variables, or factors, race factors, or rFactors :D, thats why it is so interesting to be racing, driving on the limit - there always are surprises as so many different factors are tied together.

    Furthermore, these curves are for single tires. But real cars performance is a system of four tires, and it makes things slightly more complex. In addition the tire load sensitivity must be not forgotten.
     
    Last edited: Jan 21, 2021
  14. Nieubermesch

    Nieubermesch Registered

    Joined:
    May 18, 2019
    Messages:
    635
    Likes Received:
    257
    Could you provide me with that track, so I and maybe others, could do some testing with other cars? Thanks. Or anyone who knows the track.
     
  15. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    9,595
    Likes Received:
    4,606
    Nieubermesch likes this.
  16. ATQ

    ATQ Registered

    Joined:
    Jan 15, 2021
    Messages:
    52
    Likes Received:
    33
    Are those curves from rFactor 2?
     
  17. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    9,595
    Likes Received:
    4,606
    I looked into this and couldn't get things to add up, though I'm sure tyre deflection isn't completely ignored. Anyway, in reading up on the ETT I found this post which I think confirms the reported deflection isn't correct:

    https://forum.studio-397.com/index.php?threads/euskotesttrack.58457/page-4#post-928133
     
  18. Yzangard

    Yzangard Registered

    Joined:
    Dec 6, 2018
    Messages:
    351
    Likes Received:
    173
    I don't have graphs dealing with temperature from real life, sorry...will try to find some if you need.

    Here is a graph from 2 laps on Nurburgring with serious forced understeer and very soft springs. Note that without the forced understeer "technique" my usual lap time with same parameters (track, temp, fuel, etc) is 1:54 to 1:56, here it is 1:59 (but I'm not trying to make the best possible lap time, I'm just trying to force understeer (color channel is the oversteer one, the more it is blue, the more we are understeering and the more it is orange, the more we are oversteering) :

    upload_2021-1-22_9-32-10.png

    Same graphs but changes the full Vector G for only Lateral G (absolute value) :

    upload_2021-1-22_9-54-27.png

    Since I don't have actual lateral force (maybe I could calculate it from suspension force, will try....the most useful channel is locked by S397, maybe someone know how to unlock it ?) I only use G forces but the graph should be almost the same (except when tyres are sliding since lateral force should lessen when this happens). On these graphs we can see that there is an "ideal" slip angle and beyond this point, we are undesteering like crazy (and losing a lot of time, not to mention ffb vibrations induced). I don't know if it is due to not using actual lateral forces but only G forces but it seems we don"t lose that much "grip" beyond "ideal" slip angle and that we can still get a lot of "grip" despite we are sliding a lot (once again, those graphs only show G forces, not actual FORCE which depends on mass that is balanced between 4 wheels differently once we slide, so...), maybe @Slip_Angel is not that wrong, maybe it is a bit too much grip and should be more punishing, hard to tell but those graphs don't deny his theory (they don't confirm it neither because we are missing important data to be sure).

    All I can say is that not only I was losing a lot of time but also, it is clearly not easy to drive like this, a lot more dangerous and complicated that just driving as you should.

    PS : we can also see why we are not oversteering, front tires are sliding, thus the weight is more balanced between front and rear, you can see at the cursor position (worst undesteering recorded) that rear tyres aren't even close to slide...
     

    Attached Files:

    Last edited: Jan 22, 2021
  19. Yzangard

    Yzangard Registered

    Joined:
    Dec 6, 2018
    Messages:
    351
    Likes Received:
    173
    Nope, they are from real tyre tested in laboratory.
     
  20. Kevin van Dooren

    Kevin van Dooren Registered

    Joined:
    Sep 21, 2017
    Messages:
    99
    Likes Received:
    66
    Are you basing this "ideal" slip angle on the raw data or the curvefit?
     

Share This Page