Drive to the limit, look at telemetry data of Alien Sim Racer and real Pro driver

Discussion in 'General Discussion' started by Joe, Dec 17, 2015.

  1. Old Hat

    Old Hat Registered

    Joined:
    Feb 22, 2011
    Messages:
    163
    Likes Received:
    8
    I agree.

    But I'm not convinced it's an inertia problem simply because those calculations are a basic aspect of sim vehicle dynamics that's supposed to be well understood and agreed by now. It does look like lack of inertia alright. And if a car is rotating so easily it will snap around with a small lockup in straight-line braking, ending up on the opposite side of the track, e.g. F3's, then you'd have have thought people would realise cars don't do that so easily or they'd be death traps for everyone. Forget data, it doesn't even look right - not for a full size vehicle. They're like little toys snapping this way and that. So it's a case of cornering completely with the pedals and steering is just just for straightening it out. LOL

    One reason some may not see it is that they're using brake and throttle together which has a tendency to "dampen" out these effects. It has always had an exaggerated effect in sims. "Throttle-brakers" are, in effect, driving a different sim car to those who don't do that. Another reason might be that they're such incorrigible fanboys that they think the physics most real racing teams import into rfPro (a different company) validates what they're driving in a game.

    As for data, in iRacing, for years people including some devs, ridiculed and demanded data for iRacing's supposed daft behaviour. But low and behold, despite its absence, most of that "ice racing" behaviour has gone now. In fact it's gone the other way e.g. you can take your feet off the pedals in the Skip Barber with the max oversteer setup and it stays planted. Nothing like the rF2 regional - which I'd rather drive all things considered btw.

    AC was supposed to be an antidote to this skittishness but it's not quite reactive enough IMO. But the best sim racer of all time, plus some of his team members think AC has great feel. So if you disagree, fine, but I wish people wouldn't imply it's because someone can only handle simcade physics. That makes no sense. I used to h&t an iRacing car on the old tire model that was widely considered undriveable that way, and did fine thanks. Doesn't mean you don't want to feel more "inertia" (and thus, immersion).

    All sims have strengths and weaknesses. Sometimes, I feel like a curry, other times, fish'n'chips.
     
  2. unknwn

    unknwn Registered

    Joined:
    Aug 11, 2015
    Messages:
    132
    Likes Received:
    10
    The "steering amount" discussion example below (also see the screenshot of RL vs rF2, although this isn't scientific as no hard data is provided):
    http://isiforums.net/f/showthread.p...-Now-Available?p=392514&viewfull=1#post392514

    I don't think the main point of Spinellis' theory is amount of steering lock as in general, but how sensitive the cars are. In rF2 cars drive like you are always close to limit, small inputs changes the balance of the car easily, small throttle adjustments allows to simply max out slip angle of rear tires and rotate without much of effort despite driving way below car limits.
     
  3. hexagramme

    hexagramme Registered

    Joined:
    May 25, 2013
    Messages:
    4,245
    Likes Received:
    194
    And instead of bringing anything tangible to the table, you start with name calling and ad hominem.
    Lovely. And you expect people to take these notions of yours seriously?

    You won't find one single quote from me, anywhere, that says everything in rF2 is perfect. Period. I've never said that, but if I had you'd be more than welcome to call me a "fanboy".

    But you'd like to create this image of me as being a mindless fanboy no matter what, because that way it's easier for you to dismiss anything I have to say.
    Sadly, that is a warped and wrong image. I thought you were bigger than that. Silly stuff indeed.

    And prancing around claiming that there is a very serious flaw in rF2's physics engine, a flaw that affects every single corner in every single car... What good does that do?
    Does it turn more people on to rF2? Not likely, huh? You claim you love the sim; why on earth do you then insist to almost only talk about its flaws (existent or non existent)?

    You can't back up that claim with anything but subjective descriptions, walls of text trying to sway people into believing that the cars in rF2 drive nothing like real life, at all.
    However you've yet to produce one single piece of irrefutable evidence to back up your claims. The amount of evidence that speaks against your notions is piling up however.

    If the car physics feel that awful to you... then why do you love the sim so much, as you say?
    I find it very odd, I must say.

    Thanks for the speculation BTW, yes I am "pretty fast" indeed, and contrary to what you think, I actually do perceive what's going on in the car while I race.
    And my driving style works across different sims, so no, it's not a matter of me just "mastering the quirks" of one specific physics engine.
    The fact that I do perceive what's going on in the cars while racing is the very reason I react to your posts the way I do, because I simply do not agree/haven't felt what you describe.
    On top of that I've not seen any data evidence that backs up your claims. So I'd say my reaction is appropriate.

    Did you notice how I refrained from calling you names just now?
    It's called adult communication. You should take note.
     
    Last edited by a moderator: Dec 19, 2015
  4. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    I've experienced "Spinelli throttle rotation" condition. For example, for me, the Mercedes in the AM GT3 mod with default settings has that condition. I have experienced this to some extent in real life, too, when at the limit in a Miata (MX-5), use a little throttle lift-off to cause amazing turn-in without turning the wheel.

    I don't believe it is a game engine fault. Especially since the amount of "throttle rotation" depends on which car is chosen and the setup configured. Rather I suspect it is an incorrect range of values for the dampers, such that the shocks are acting more like springs than dampers coupled with a particular driving style. Dampers are the "dirty science" of racing that doesn't get talked about much in sims despite having a huge impact on vehicle dynamics. I suspect this is partially why some people derail the discussion by looking at inertia, because inertia is what they understand and not looking at the parts of vehicle dynamics that they don't understand.
     
    Last edited by a moderator: Dec 20, 2015
  5. matf1

    matf1 Registered

    Joined:
    Dec 27, 2011
    Messages:
    671
    Likes Received:
    2
    A combination of soft or unbalanced suspension(especially bumps), tight diff clutches and over driving can produce this effect but I know Spinelli is adept at understanding the car setup options.
    In fact, I've taken a new approach to my race setups which is paying huge dividends and let's me feel at one with this sim. In no circumstances can I jump on the brakes or throttle, but I'm faster, more consistent and wear is much more even and consistent than at any time previously.

    Don't be soft, harden up a bit. You lose bite, but this is a good thing. Sliding is just magnificent when the balance is right.
     
  6. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    Per Paul Loatman's claim that rF2 STS is too high, he proposed to reduce down to 0.27 for more "realistic". Of course, this is purely subjective. I am puzzled by this since I am lack of real live experience on racing car. Now, by reading this thesis, I found the main conclusion on the FFB force studied on rF1 supports his finding.
    From table 3.1 to 3.6 (or from fig 3.7 to 3.14), all the results showed rF1 FFB is Over-Strength. That says given a resultant of same lateral-g value, the rF1 steering wheel FFB strength is far more than real car FFB strength (Steering torque gradient of rF1 is much higher than that of real car)- -- some corners's results with over 100% different from real car, corner 5 showed over 700% off in comparison with real car). This is consistent for all the corners on the track they found. If this finding holds for rF2 too, then Paul's claim is true. At least this leads to a meaningful objective argument.

    Note: the data they shown are with huge variance, so confident level may not be that high, unless they provide huge sets of sample data. The detailed studied on turn 2 and turn 3 later in the thesis (pg 72 - 79) showed higher confident level with far more data.
     
    Last edited by a moderator: Dec 20, 2015
  7. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,346
    Likes Received:
    6,572
    STS doesn't affect the maximum force, it just changes the 'shape' on the way there. So a low STS might give you a better (lower) gradient at low forces, but a higher gradient at high forces. In that case the mid forces would be lower than they were originally, but as you near maximum they will scoot right up there again. (high STS would do the opposite)

    Also, the steering arm force they are logging via a plugin in rF1 is the internal rF calculation; it doesn't reflect what would be felt by the driver via the realfeel and leoFFB plugins (note the gradients generally point towards the origin, something leoFFB 'fixes'). It probably would, however, somewhat reflect what rF2 outputs (as its FFB reflects the calculated steering arm force) albeit with an entirely different tyre model and lower physics rate.
     
  8. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    Reducing STS from 1 down to 0.27 would not just reduce torque gradient at low portion but also at mid/high portions as well. The large portion of curve would be pulled down, although max point is fixed.
    Their studies were not done (or target) at max torque. Except for turn 8 data shown the rF1 torque up to 35 Nm, at which might be implied to close to a "max" torque. If you remove the data near 35 Nm in turn 8 chart (fig 3.8), you will see it would not affect outcomes and their conclusion.
     
  9. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    A lower than 1 STS will causes the initial part of the final ffb output vs raw ffb gradient to be lower than for the default STS of 1. Then the gradient increases, passing through 1 at some point and continues up to some maximum gradient greater than 1 at max ffb. The lower the STS, the greater the min and max gradient difference in the curve. Same but vice-versa affect for higher STS.

    Example graph of how STS transforms raw ffb into a final ffb output:

    [​IMG]

    And how that would affect a section of ffb output from the sim during gameplay:

    [​IMG]
     
  10. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    your "gradient" in your chart is NOT same as gradient in their chart. theirs defined is the slope of a regression line that fits to scatted data torque vs lateral-g value.

    anyhow, the best way we can do is just to run rF2 on same corner with two different settings: STS=1 and STS=0.27
    then to plot the data (FFB vs lateral-g) to see how much would be diff...
     
    Last edited by a moderator: Dec 20, 2015
  11. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    I think you're onto something.

    Torques measured in sims are the physics calculated torques acting on the front tyres about their steering axis (which translates to some proportional torque acting on the steering system as a whole) from road-tyre interactions. But those same torques acting on the front tyres about their steering axis are not captured/recorded by a torque sensor in a real-car. The torque sensor in a real car only records the torques placed on the steering system by the driver. If there is a torque from tyre-road interaction acting on the steering system but no driver input torque on the steering wheel, then there is no net torque recorded by the torque sensor.

    So it maybe that the "thesis" is comparing apples to oranges.
     
  12. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    We're talking about STS's affect on the final ffb output right? The curves i've drawn to illustrate STS's affects were arbitrary (i.e. not an accurate representation of how STS affects the ffb curves).

    Can you send a link to the "thesis" and the page you are referring to please.

    (For clarity, i haven't read it.)
     
    Last edited by a moderator: Dec 20, 2015
  13. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
  14. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
  15. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    starting pg 23
     
  16. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    Ok Joe, i read up to the part talking about "steering torque gradient" and i understand what you mean now that it's fixed/constant/linear for all lateral accelerations (per a specific cornering scenario) in the "thesis". However, this does not appear to be what you were saying/talking about when you said…

    …which is what i was responding to.

    "Steering torque gradients" in the "thesis" maybe linear but changing the STS (away from the default "1" value) applies a non-linearising transformation function to rf2's otherwise linear steering torque/ffb gradients per lateral accelerations (per a given cornering scenario).
     
    Last edited by a moderator: Dec 20, 2015
  17. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    linear or not, there is no way we can tell from their real car data. It could be indeed non-linear.
    What I try to say is that Paul's claim is NOT try to fix something, but his setting makes more "realistic" as He claimed.
    The thesis found rF1 torque sensitivity is way too high in comparing with real car, this may hold true for rF2. This could be a reason why he found lower the STS helps.
    As first step I think may help on this to run rF2 on same corner with two different settings: STS=1 and STS=0.27
    then to plot the data (FFB vs lateral-g) to see how much would be diff... and if the diff is similar found between rF1 and real car.

    I will be out of town for a week vacation. If some one can do that would be great. Otherwise, I will do once I come back.
     
  18. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    Joe, i was not making any claim that the real car data is linear or non-linear. I was simply stating that STS (which stands for "Steering Torque Sensitivity") in rf2 determines whether the final ffb output is linear or non-linear with respect to the raw ffb which is taken from the virtual steering wheel torques in game.

    Yes and i realise that now having read the "thesis". But even when i hadn't read the paper, i was not talking about this. Just addressing the point about STS function in rf2 (which i presume is the same as in rf1?)

    Sure but it's a dirty way of trying to address the problem of dissimilar torque gradients (if even there). If it's there, as the thesis shows, it needs to be addressed within the car physics model really since that is where the problem stems (again, if the problem is really present for some specific car for a specific sim in question).
     
    Last edited by a moderator: Dec 20, 2015
  19. unknwn

    unknwn Registered

    Joined:
    Aug 11, 2015
    Messages:
    132
    Likes Received:
    10
    Last edited by a moderator: Dec 20, 2015
  20. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Thank you, thank you, thank you, thank you. Everything you said is exactly what I've experienced as-well especially what you described in your first paragraph (but really everything including what you said about iRacing, and how AC has gone a little overboard with their attempts of correcting all this with not reactive enough behavior).



    Exactly. Thank you, too, so much! Every word in your post.


    The both of you seem to know exactly what I'm talking about regarding specifically the steering and getting cars to go around corners thing.



    To Hex:
    The reason why I continue with ISI physics is because they never feel "videogamey" even when things happen that are just wrong (wrong/inaccurate/"digital" yes, "videogamey" no). I am most passionate about the ISI physics engine as I love ISI as a company who don't do all the marketing crap, and all that stuff; they seem like the type of company I wish all companies were: simply making as good a product as they can while staying humble rather than trying to "take over the world", make billions of dollars, destroy every other competitor, etc. Driving-wise, I just see so much potential in the raw physics engine of the ISI motor (even pre-RF2) and see a lot of great stuff. I just feel there are some long-standing (15, or so, years) of specific vehicle behavior things that need to seriously be looked at. If it wasn't for those, I honestly don't think I'd ever feel the need or urge to play under any other sim-engine. That is why I am so vocal - I think the ISI motor has the least amount of things to fix/improve in order to finally be "THE" sim we have all wanted all of our lives (although those things do seem to be pretty deep considering they have been present in every iteration of the engine since I first started with F1 2002 (and mods) back in 2003 or so).
     
    Last edited by a moderator: Dec 20, 2015

Share This Page