1. Intel iGPU (6xx series) crashes
    Fixed drivers available!
    Instructions here

    Dismiss Notice

Tire model potential

Discussion in 'General Discussion' started by Flex, Apr 5, 2017.

  1. Flex

    Flex
    Expand Collapse

    Joined:
    Aug 6, 2012
    Messages:
    22
    I begin to be really interested in the tire model of this game. It seems to be very unique and I feel it could be the major breakthrough that this game provides.

    I'll explain what I mean by that. I'm a sim racer and I saw a lot of videos about the tire model of different recent sim games, and this tire model seems to be the central part, and what makes them apart. Its all about the most precise friction curves and heat behaviours and such. And yet most of them still use rigid tire models. Even the most advanced model in this regard, the one from rfactor 2, has tires that only compress a little, and seem to go ¨into the ground¨.

    On the other hand, the tires from beamNG deform a lot, like the real deal, and can even pop under pressure. Friction is probably not simulated as well as in racing sims, but it is still taken seriously. Heat is missing though, AFAIK.

    If a tire model based on the beamNG pressurewheels was to be developped for a racing sim, it would probably be the best and most realistic one ever.

    What do you think ? Am I wrong thinking that racing sims seem to still be a little primitive ? Would a racing sim based on beamnNG's physics engine be possible ? Is this the future of sims ?

    BTW, nice work, Goosah ;)
     
    • Agree Agree x 2
  2. DOUGL4S1

    DOUGL4S1
    Expand Collapse

    Joined:
    Sep 27, 2016
    Messages:
    187
    The tire model for Live For Speed is pretty good too. They also deform a lot, they can be heated to have better performance and if you use them too much you can wear them and they won't have any grip. However, I can't remember if the tires can actually blow up in that game.
     
    • Agree Agree x 1
  3. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    7,046
    With rF2 you have to remember, what you see is not what you get, it does a lot of math and stuff that they don't really bring up as a visual effect, but I don't know if they do have really much of flex even in reality, I guess one should try with low tire pressures and see how it goes.

    BeamNG tires are really good though, when things get light, there I guess it's bit less grip than one could except, but maybe that is because lack of heat then?

    What I would like to see is tire thread effect on how it goes on mud and softer stuff, street tires on grass are quite bad, while tractor tires are quite good on softer stuff.

    Anyway, tires are always some kind of compromise, they can be so complex and gains going to so deep might be not very much, but even with limitless resources, I doubt that they could be simulated perfectly in every situation.
     
  4. Fuzzwad

    Fuzzwad
    Expand Collapse

    Joined:
    Feb 8, 2017
    Messages:
    87
    Honestly, Beamng's actual DRIVING physics seem to be put to the back in favor of all the crashing that you can do, and while I fully understand why, Beamng isn't marketed as a racing sim, nor are people buying it as a racing sim, and because it is just nearly impossible to drive like a racing sim without a wheel. (I'm lazy and I didn't feel like going over that paragraph and fixing it. ALSO, refrain from wheel/controller/keyboard arguements, please. We've all seen plenty of them.) My point is, if you get a proper car (FR17 hype anyone?) It's actually a VERY good sim, and I don't think there needs to be a separate game with the same physics for it to be a "proper" sim.

    /rant

    TL:DR
    Beam's physics are already racing sim-ey enough, you just need to actually drive it like one.
     
  5. CaptanW

    CaptanW
    Expand Collapse

    Joined:
    Aug 18, 2014
    Messages:
    1,141
    Tire heat+tire wear is something I have always wanted. This could be useful in drag racing so it would be more realistic.
     
    • Agree Agree x 4
  6. wrinkle345

    wrinkle345
    Expand Collapse

    Joined:
    Jul 2, 2015
    Messages:
    553
    +1

    I'm also still holding onto hope that we'll eventually get backfires and flame spitting. It doesn't add much sim-wise, but god it would look good.
     
    • Agree Agree x 2
  7. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    7,046
    I guess heat needs some kind of abrasion, which again might be useful for general behavior when there is less weight?
     
    • Agree Agree x 2
  8. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    785
    Hey guys thanks for the good words :) Because the tire model stuff is not much to see I don't tend to post about it much. It is very far from taking a back seat, it only took a bit of a break while I helped to write the powertrain system. I have been constantly testing and trying ideas and Estama is always writing experimental features in the physics core for me to try. I have been meaning to make a blog post because we have made a lot of improvements lately but I can highlight some:

    Node friction load sensitivity (WIP-Not all tires yet)
    Now the ultimate grip of the tire is reduced when the tire nodes are loaded heavily
    • Lightly loaded tires have comparably more traction. This means less inside wheel spin with throttle in corners.
    • The cars behavior and balance at the limit is more progressive, cars do not "tank slap" as much when trying to recover from a slide
    • Cars can brake/accelerate with some 5-10% more force than they corner, because the wheelbase is longer than track, weight transfer reduces grip less in the longitudinal direction.
    • Camber effects ultimate grip more correctly because loading the outside tire nodes excessively reduces grip through load sensitivity.
    • Vehicles, especially tall ones, are less likely to roll after hitting curbs or off camber situations, because the extreme camber reduces the grip through load sensitivity on the outside tire nodes.
    Rolling resistance reduction
    • Rolling resistance reduced another 10-30% by improving the node to ground collision code, compensated for with more aero drag (a step in the right direction)
    Tire "softness" friction property
    • Some new feature in the friction model can make nodes more sticky in transient situations. The softer the compound would be, the more "softness" we add, this really improving steering response and makes the tire less likely to be shocked loose from the road by sudden input or a bump etc. It can also reduce loss of grip at high speeds (250+km/h) to a basically negligible amount.
    Tire "treadCoef" + ground model "roughness" friction properties
    Basically choosing the correct tire for the surface is more important now, as off road tires will perform well off road but not so much on asphalt, and slicks will be great on asphalt but practically useless in mud.
    • How it works: Tires have a tread coefficient that represents how well treaded the tire is (Slick = 0, mud tire = 1) we multiply this against a roughness coefficient in each ground type (mud has a high roughness, asphalt has very little) this means that a mud tire now works well in mud but mediocre on pavement, a slick works well on pavement but horribly on mud. Similar for other ground models.
     
    • Informative Informative x 8
    • Like Like x 3
  9. CreasingCurve

    CreasingCurve
    Expand Collapse

    Joined:
    Jan 19, 2016
    Messages:
    976
    What i would like is something along the lines of what @fufsgen has said above. Basically a Tread depth or tyre tread type variable/multiplier in the tyre jbeam. Basically drag slicks would either have a multiplier of 0.1 or Tread Type: "Slick" and stuff like the full on 6 lug offroad tyres would have a multiplier of 1 or Tread Type: "Offroad".
    Seeing how there are already so many variables, this shouldn't even be very hard to implement


    Edit: i got ninjad by a dev with my idea.
     
    #9 CreasingCurve, Apr 6, 2017
    Last edited: Apr 6, 2017
  10. Dummiesman

    Dummiesman
    Expand Collapse

    Joined:
    Sep 17, 2013
    Messages:
    4,503
    BeamNG's tire model in a sim game would not be great.

    Most major sims have quite complex tire models that very closely simulate a real tire. BeamNG's tires are simulated at individual points , which are comparably low resolution compared to sims.
     
    • Agree Agree x 1
  11. wrinkle345

    wrinkle345
    Expand Collapse

    Joined:
    Jul 2, 2015
    Messages:
    553
    You devs are the heroes we don't deserve.

    Not to mention one of the only dev teams who gives indepth info on wip projects.
     
    • Agree Agree x 1
  12. Spaceballs the Username

    Spaceballs the Username
    Expand Collapse

    Joined:
    Oct 13, 2015
    Messages:
    581
    All of this is really wonderful! Especially the last part about tires working differently depending on terrain. A question is, will this system be extended for wet / snow performance as well? a "treaded" tire that is good in mud is not necessarily better in rain than a slick.
     
  13. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    785
    I don't want to get into specifics about other sims that I am not qualified to talk about, but most racing sims have a tire model that is a point contact, or perhaps a few points, which skip along the ground at a few hundred hz with some connection upwards towards the axle. The complicated part of their tire model is the lookup tables/equations to determine how much force to apply at the contact point based on all the dimensions of the tire model (slip angle, slip ratio, normal force, camber etc etc). If they have visual deformation, it would probably be done in a similar approach, a lookup or equation method that morphs a visual mesh to match the camber, rolling radius that the model sees.

    Whereas for us, this whole multi-dimensional tire problem is solved physically and visually by a physical representation of the tire. I would say our model is more "physically honest" than the standard methods, it also has the ability to show emergent behavior whereas with an empirical model, one can only get out what they put into it. For example, instead of skipping along the ground, our tire structure is actually rotating, giving birth to all sorts of inertial effects which affect the rolling radius and ground contact length, turning response. Because it is a flexible rotating structure with collision, it is as comfortable working under a rock crawler, wedging between rocks, as on a race car. Our wheel will also continue to roll up to insanely high speeds without skipping over obstacles due to simulation frequency limits. Open wheeled race cars would exhibit the correctly disastrous result when running into each others tires. So for us the benefits are the versatility and emergence of physical behavior that we don't have to patch in, but the challenge for us is calibration, and in getting as much as we can out of as few nodes and beams as possible. As technology progresses our physical approach will converge closer and closer to reality with more nodes, higher frequency, and that's why I think it's the way forward.
     
    • Agree Agree x 4
    • Like Like x 3
    • Informative Informative x 1
  14. Dummiesman

    Dummiesman
    Expand Collapse

    Joined:
    Sep 17, 2013
    Messages:
    4,503
    rFactor 2 has this, which is where my "quite complex tire model" comes from. It even simulates flat spots on individual pieces of the tire, including in FFB.
     
    • Informative Informative x 2
    • Like Like x 1
  15. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    785
    To my knowledge it is running an offline simulation that it compiles into a really really deep lookup table but for the same type of simulation. I could be wrong though, nevertheless it is impressive!
     
    • Agree Agree x 2
  16. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    7,046
    I don't know, these devs have quite crazy fortune telling machines, whatever community comes up they probably have already working on. That is this incredible game where mostly everything is done like it should for first time, is going to have even more of the stuff implemented like it should, which is really incredible.
     
    • Agree Agree x 1
  17. Diamondback

    Diamondback
    Expand Collapse
    Moderator
    BeamNG Team

    Joined:
    Apr 8, 2014
    Messages:
    1,640
    hehe, I like that framerate display in the video :)
    Definitely cool to see though.
     
    • Agree Agree x 3
  18. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    7,046
    It's pita to make tires to that game and they still work only at limited situations, but might be something to do with their engine being limited to driving on rails, more or less. When tires lift off or driving on two wheels or getting bit of wrong side shown things start to go really weird.

    That is why I like to call it spreadsheet simulator, they compute ton of things behind the scenes, but applying that to believable visual form is not so good if you go beyond excepted way of driving.

    There is even rubber buildup on driving line and all sorts of stuff, but everything seems to be working well only if you drive like intended to or that is at least my experience.

    Even BeamNG tires might be simpler, I like them lot more, because they indeed work quite well no matter how crazy way I want to drive.
     
    • Agree Agree x 3
  19. Flex

    Flex
    Expand Collapse

    Joined:
    Aug 6, 2012
    Messages:
    22
    Ok, nice detailed information here. That clears up a lot in my mind about these tire model differences. So, BeamNG's model indeed has some potential. As a physicist, i find this detailed approach really interesting.

    Nice video. So Rfactor 2's tire model is much more precise than I thought. Still, we don't know if the grip model is directly linked to tire deformation or not, or if it is still partly empirical. If it is simulated by deformation, it is probably a lot of strain on the CPU. Either way, I should check Rfactor 2 more closely. It seems to be the most realistic sim out there.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice