Automobilista - How the FFB works and what is new about it

Discussion in 'Automobilista - General Discussion' started by Renato Simioni, Mar 9, 2016.

  1. macmaniacoes

    macmaniacoes Member AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    41
    Likes Received:
    20
    But if you have a spare pad or button box you can use XPadder or Antimicro to pass button presses from that pad as key presses. For example, I use Antimicro with GT Legends, which doesn't recognise reverse, fifth and sixth in my H shifter. So I assign three random keys to this positions, and in-game I select those keys as R, 5 and 6 using the keyboard. When driving, I use the shifter, but Antimicro send to the game the correspondent keys, so the game change gears as intended.
    In your case, you need to assign macros to button presses, let's say A button to increase FFB response. That's RCtrl+6, and so on so forth
     
    • Like Like x 2
    • Informative Informative x 1
    • Useful Useful x 1
  2. ghoults

    ghoults Member AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    76
    Likes Received:
    34
    First of all the first post refers to some graph image but there is no image link anywhere in the post.

    I'm kind of surprised you went with this "FFB Low Force Boost" idea. It doesn't sound very scientific compared to having something like the ffb lut tables + minforce in ac for example. Doing it the ac way you could measure your wheel ffb and then apply the corrent minforce settings and ffb tables to get the wheel ffb to match what the game should output.

    Now with this FFB Low Force Boost it sounds like it just a feel thing. Adjust it higher or lower until it feels correct. It is not only odd way to adjust ffb but it is also completely different to basically all other sims out there which use minforce as a setting. I'm kind of wondering what are the benefits of your system vs a minforce setting that is not only well understood but works well and can be measured so you can adjust it to correct value using real life measurements instead of just feel?

    I really think the best way to make ffb adjustable is to have these things:
    1. minforce. I don't personally know any downsides to having this setting. Maybe you did not add this because it could not be added to your ffb code?
    2. some kind of linearity adjustment that is not connected to minforce
    3. some kind of underpass filter so players can tune out high frequency noise from the ffb signal

    The reason I'm wondering these things is that for me for whatever reason the ffb has been notchy and poor ever since the ffb changes were added (back in the previous title). I have not been able to get smooth ffb and I've tried all the different refresh rates and ffb and low force boost settings. It is pretty bad but not undrivable. But for me it puts automobilista clearly below rf1, rf2, ac and even iracing.

    My wheel is fine. I can get nice and smooth ffb in rf2, rf, ac and iracing as well although the last is a long time ago. And in stock car extreme as well before the ffb updates.

    With wheelcheck I get this kind of graph for my wheel:
    http://i68.tinypic.com/2ikwjfr.jpg
    As you can see it is linear and has relatively small deadzone, ~5%. That wheel is fanatec 911 turbo s.

    So if I am reading this thread correctly I'd need higher than 0 FFB Low Force Boost but doing that will make my wheel unlinear. What kind of settings other people are using who have something like fanatec gt2, gt3, gt3rs or 911? Do you guys have smooth ffb or is it notchy for you too? What fanatec firmware versions are you guys running?

    edit: the new "FFB Low Force Boost" is not mentioned in the game manual.
     
    Last edited: Aug 27, 2016
    • Dislike Dislike x 1
  3. Tony Binelli

    Tony Binelli Active Member AMS2 Club Member

    Joined:
    Apr 26, 2016
    Messages:
    422
    Likes Received:
    127
    Fanatic 911 turbo is a piece of crap in my opinion, my V2 feels freaking amazing in AMS, way better than in assetto. I owned the fanatec GT3RS and man it doesn't begin to compare with the better wheels. Just my opinion of course
     
    • Dislike Dislike x 1
  4. ghoults

    ghoults Member AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    76
    Likes Received:
    34
    It is kinda okayish wheel. I agree it is not the best and probably way below something like cswv2 or t300. However the notchy issue is ams specific and hopefully something I can solve.

    I'm not going to buy new wheel to fix ams ffb :).
     
  5. Tony Binelli

    Tony Binelli Active Member AMS2 Club Member

    Joined:
    Apr 26, 2016
    Messages:
    422
    Likes Received:
    127
    Fair enough, I went from a gt3 to a t500 and then a v1 and then v2. All 3 are worlds apart from the gt3rs. You wouldn't be buying a new wheel to fix notchiness you would be buying one to really improve your sim racing experience, I have heard very good things about the t300 wheels.
     
  6. SirVanhan

    SirVanhan New Member

    Joined:
    Mar 10, 2016
    Messages:
    13
    Likes Received:
    0
    I have to be honest: as I've never driven a real racing car, I have no idea how the FFB should behave, what should I feel...
    What i know though is that every simulator feels completely different to the others.
    I have a G27. Setting "FFB Low Force Boost" to 80% as the opening post suggests, the feeling is veeeeeeeeery different to AMS beta. I used to like it, because I could feel the car more. But was it more realistic? The point is that I don't even know what I have to achieve. I'm completely lost now.
     
  7. Domagoj Lovric

    Domagoj Lovric Moderator Staff Member

    Joined:
    Feb 15, 2016
    Messages:
    1,092
    Likes Received:
    1,225
    Can't answer all, but some answers and questions :)
    Lut tables / min force - these are two separate things right? (just checking because from your post one could figure it's same thing)

    As for minforce setting, we think our way is better because it increases lower forces, yet still retains distinction between them, not that under some level all are flattened and brought up in level.
    Yes, linearity is altered. But there is a good reason behind that. If we'd keep all cars ffb fully linear, and use minforce setting, you'd come back and complain that open wheelers ffb is weak in all but really high speed corners.
    This feature is there primarily to compensate for ffb weakness / low dynamic range / small steps of low-fidelity steering wheels.

    High frequency noise - you can deal with this adjusting damper and smoothing in realfeel.ini per car. Any such filtering brings in lag, and probably you probably lose proper curbs feedback.
     
    • Like Like x 1
  8. Dirk

    Dirk Member AMS2 Club Member

    Joined:
    Apr 19, 2016
    Messages:
    38
    Likes Received:
    9
    Just make sure you're not clipping and the rotation/steering lock is correct, try to understand how the different settings affect the output so you you're not just randomly changing things, but don't get overly concerned with searching for some objectively "right" setting.

    FFB is a pretty artifical construct, it's not going to feel like a real car anyway, so in the end I'd say go for what either feels the most enjoyable to you or what gives you the best performance, and these two are not necessarily the same setting.

    And while linearity in the communication between software output and hardware output is a good goal (I think that's what LUT curves are for), linearity between the what the physics calculate as force and what the software does put out is a different matter imo. The example of the open wheelers in low downforce corners being the obvious example of a non-linear curve being preferable.
     
    • Like Like x 2
    • Agree Agree x 1
  9. Domagoj Lovric

    Domagoj Lovric Moderator Staff Member

    Joined:
    Feb 15, 2016
    Messages:
    1,092
    Likes Received:
    1,225
    Well it will be as good as physics engine is; and even if you had totally accurate engine (which mostly practically speaking probably is way unobtanium at the moment) output controller should ideally be able to produce virtually any force, have zero inertia (so all inertia is calculated in physics), and have huge acceleration. That can't be acheived. All those low and minforce stuff are there to try compensating for shortcomings of wheels most users use (excluding few expenive products, like Bodnar wheel).

    One thing we do, is that we actually spit out to wheel whatever physics calculates, not some made up ffb based on this and that - there is default isi ffb for that. Our's isn't called "pure" just like that, it really is pure force taken from virtual car steering rack.

    Openwheelers having low force when at low speed is actually due to low aeroload. Force isn't actually low, but if we'd increase it so low speed force is high enough all you'd get is clipping of high forces, except for few very powerful controllers, like Bodnar one. This is a case where low force boost is needed. When high forces go through it they remain the same, while low ones the higher the setting, the more and more (the lower you go) are increased.
     
    • Like Like x 1
  10. ghoults

    ghoults Member AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    76
    Likes Received:
    34
    Let me preface this post by saying that I don't want you to make your ffb like assetto corsa. I'm just using some common techniques used by ac as examples to show you practically everybody does this differently :).

    Lut tables basically means you apply a correction to the ffb over its full range for every bit of resolution so that every step of the ffb is tuned to correct level. This is to prevent situations where the sim asks 50% force then one linear wheel will give that 50% while some unlinear wheel may give 30% or 70%. If 50% means 50% you can use this information to have more consistent ffb on all wheels basically.

    A lut table is a notepad file in assetto corsa that for example can look like this:
    0|0
    25|30
    50|60
    75|85
    100|100
    First value is your sim ffb output for example. 2nd one is the corrected value which guarantees linearity so your in-game ffb matches the wheel output.

    I disagree. Your low ffb boost seem to be this mix of minforce setting AND a linearity setting. The graph says it is just a linearity correlation value but in the niels' video it was mentioned this value can be also used to counter the effect of the ffb deadzone in old logitech wheels. So despite the graph it seems to be it does both. It is possible I have misunderstood.

    Anywas, in my own case I'd only need a minforce setting because my wheel outputs linear ffb. For some other wheel you'd want both and for some other wheel neither. Having two separate values allows the ffb to be adjusted better for all cases whereas your system only works well for cases that match your preconfigured values. This is because the way I understand in your system the values (minforce and overall linearity multipliers for the ffb plot from 0 to 100%) are linked and controlled from one value.

    I'm not actually asking the ffb output to be linear. I want realistic ffb. Not linear, necessarily. But to achieve realistic ffb you want to know the correlation of the ffb values your sim wants the wheel outputs. You want 1:1 ratio so if in the end when driving on the track the wheel can match your sim outputs as well as possible. If on straights the sim wants 20-50% then the wheel will give that 20-50% and not something like 30-40 or 5-70.

    Keeping it linear you then know what kind of ffb the player's ffb wheel will output for any number off ffb you are asking. This doesn't lead to single seaters becoming more on off in ffb behaviour (no ffb on straights, lots on high speed corners). In fact this kind of system allows you to avoid such situations better because you know for a fact what is the relationship between the ffb your sim asks from the wheel and what the wheel will output for that number.

    And similarly you want a minforce because most wheels don't output any forces below 10% (there are lots of wheelcheck graphs to see in internet). They need larger forces to "wake up". In that graph I posted earlier from my own wheel that minimum for my wheel is something like 5%. I've seen that number go from 3% to 30%! In your graph all the wheels start producing ffb at 0.000001% basically. In reality the ffb wheel only starts to move when your sim asks something like 5% or more.

    So driving in a corner in that single seater your sim ffb may ask a value of somewhere between 5-10%. For a lot of wheels this does nothing. With minforce setting that minforce is set higher so whenever you ask a small ffb amount the wheel can produce that.

    Damper also resists rotation which is not what I want. High frequency noise can be anything from kerbs or bumps in the road to flatspots in the tires. The two reasons why I personally want to avoid this high frequency ffb is 1. simply because this vibration is mechanically bad for your wheel. Over time it will increase the slack in the gears of the wheel. 2. And because most of the wheels are not very good at producing this kind of fast movement this vibration is then turned into oscillations on straightaways. For ff wheels this kind of really high frequency ffb output does work because of the precision, zero slack mechanical aspect (no gears, belts etc.) and the speed those powerful servomotors can output. But for slower low end wheels those signals will just turn into oscillations which is bad for ffb.

    I hope you take my criticism as just that. Criticism. I have my best effort to explain my findings in a way that is not based on just opinion but measurable values. I don't want to or intend to call you out or attack you because of the way you are doing ffb. I'm just trying to provide constructive feedback and try to see things from your perspective. The reason I write this disclaimer is because there are lots of fanboys out there who will attack anyone who dares to criticize their god sim. For those who can read I hope you understand that simply bashing or attacking ams is not my goal here.
     
    • Like Like x 1
    • Agree Agree x 1
    • Winner Winner x 1
  11. Heitor Facuri Cicoti

    Heitor Facuri Cicoti Good Member AMS2 Club Member

    Joined:
    Mar 8, 2016
    Messages:
    858
    Likes Received:
    838
    I understand the concern and agree with the logic you exposed, but those tools (minimum force or LUT) rely entirely on data provided by WheelCheck. And, from what I know of WheelCheck, it isn't the most reliable way to measure torque output on steering wheels. It is done indirectly by reading displacement after a short period of force applied, and this could lead to wrong conclusions.

    So all the effort to make something scientific and objective can go away while using misleading data, thus becoming subjective. And subjective by subjective, it's better to invest on the simpler one.

    But let's see what the Devs say about it.
     
    • Agree Agree x 1
  12. Renato Simioni

    Renato Simioni Administrator Staff Member

    Joined:
    Feb 14, 2016
    Messages:
    4,662
    Likes Received:
    42,109
    Generally our approach is keep FFB options to a bare minimum as once you overwhelm the user with options it will often lead them into entering this rabbit hole pursuing some magical setting that will turn his G27 into a Bodnar, fudging what doesn´t necessarily needs to be fudged, until he ends up with a configuration that´s potentially worse than the one he started with - and then comes the influx of confused posts. Power users may not appreciate the relative inflexibility and compromises involved with that approach, but I think the overall "feedback" if you excuse the pun regarding AMS FFB backs up that approach. Not everything should be made into a project.

    I´ve amended the opening post to include the missing graph illustrating Low FFB Slider efffect - it´s also attatched here.

    With regards to the "merits" of the slider - the code to change the FFB curve was already there via "FFB steer force exponent", we just made it customizable with the slider, and in doing so addressed a common request to improve the low force FFB for lower range wheels, which in turn also allowed us to globably reduced clipping. It is no less "scientific" than basic FFB strength output - it is subjective to taste so it´s inevitably a "feel thing", there´s no absolute true value.

    Ghoults, tweaking Low Force Boost is not a requirement - you can keep it at 40% which was the old default, or (since you mention you want to ease high frequency noise) you may just about iron them out by going lower. As Dom suggests, you can also try upping Realfeel Smoothing level to filter them out more. There´s technically no reason why AMS FFB should be less smooth / notchier than SCE, quite the contrary in fact.
     

    Attached Files:

    • Like Like x 1
    • Agree Agree x 1
  13. Dean Ogurek

    Dean Ogurek "Love the Simulation You're Dreaming In." AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    1,399
    Likes Received:
    644
    To some extent, FFB wheels compensate for what we lack in the virtual Car compared to Real Life; Seat-of-the-pants feeling that include physics feedback that most of us don't have sitting in our seat at home.

    A good FFB steering system can communicate many aspects of car behavior and with practice, we can differentiate between front-tire grip-loss, under-steer / over-steer, Self-aligning torque, wheel lockup, bump-steer, torque-steer, flat-spots, and more.
     
  14. Dirk

    Dirk Member AMS2 Club Member

    Joined:
    Apr 19, 2016
    Messages:
    38
    Likes Received:
    9
    I think we agree: You get a force from the physics engine, then you either manipulate it or not (via Low Force Boost or MinForce or additonal Damper, Smoothing and so on), then you put that out to the wheel. When putting it out to the wheel you want the wheel to react linear so you know it does what you want it to. At that point having LUT would be good, and if you find the time to put it in that'd be cool.

    However, I think if you had a linear output, and a "perfect" wheel, and did none of the manipulation internally before - no min force, no damper, no smoothing etc. - it still wouldn't feel like a real car, because the steering wheel in a real car is way more inert than what FFB does, at least with the kind of modern cars I drive (street cars with power steering).

    Really all I tried to get across to the guy is not to stress out too much over what is "right" and enjoy the racing.
     
    • Agree Agree x 2
  15. ghoults

    ghoults Member AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    76
    Likes Received:
    34
    I do kinda disagree here. I say kinda because I'm not against having the option to turn on some effects that do try to convey something like locked brakes or grip loss on rear tires. But personally I want to turn all of those effects off because personally I find them distracting

    Yeah the wheelcheck is not perfect but at the same time it is the best tool we have to measure the ffb. As long as people understand it is not perfect but just a tool. In some cases it is very accurate (sensor on the same axle as the motor) while in other cases it can exagerate an issue or even create an issue where the wheel manufacturer has built in some kind of mechanism to prevent that issue. But due to the way wheelcheck works it doesn't "see" that mechanism it could skew the results.

    Still it is definitely better than nothing. Because apart from wheelcheck there really isn't other way of measuring ffb other than feel.

    I'm kind of surprised to read this. Because in ams even to set up antialiasing you need to use something like nvidia or ati drivers to get it to work. In addition to that you may need/want to install 3rd party software like nvidia inspector. Even with your manual that can be very hard for novice users because not everyone has english language nvidia or ati drivers. Turning supersampling to this or that may be difficult because their driver tool has completely different terms for that.

    From that pov I don't really see this fear of confusion. If using 3rd party tools to adjust a basic thing like aa is not confusing then honestly I don't see how offering industry standard like ffb minforce setting is more confusing. I fail to see how that kind of setting could be too complex for ams when it is not too complex for iracing or ac. Both of which are targeting far wider audience with overall lesser skillset and understanding.

    What I do not know of course is how difficult it might be for you to offer these features. If it is too much work then of course it might not be wise to start such project.
     
  16. Dirk

    Dirk Member AMS2 Club Member

    Joined:
    Apr 19, 2016
    Messages:
    38
    Likes Received:
    9
    Isn't minforce a flat multiplier, i.e. 5% would be F * 1.05? That shouldn't be too much trouble to implement?

    Personally, I'm still more interested in getting:
    1) A force feedback meter, so I can really dial it in just below clipping.
    2) The option to set LowForceBoost per car and not global.

    You should also probably add the RealFeel keyboard shortcuts to the OP, [7] and [9] are for changing the MaxForceAtSteeringRack without having to leave the game and edit the INI:
    [​IMG]
     
    • Like Like x 1
    • Agree Agree x 1
  17. Dean Ogurek

    Dean Ogurek "Love the Simulation You're Dreaming In." AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    1,399
    Likes Received:
    644
    I think it depends on the wheel being used but, I generally rely on minimal FFB effects and I still get everything mentioned out of the FFB. To me though, instead of being distracting, I find them more immersive.:)
     
  18. Spin

    Spin Active Member

    Joined:
    Apr 15, 2016
    Messages:
    297
    Likes Received:
    33
    It could at least get a little better if we transitioned to position-based FFB instead of the current torque-based model all games/wheels use. FFB guru Leo Bodnar wrote a good article about this (and I've spoken to a few other FFB gurus about it).

    Unfortunately, other than graphics, videogame tech moves at a snail's pace.
     
  19. Dean Ogurek

    Dean Ogurek "Love the Simulation You're Dreaming In." AMS2 Club Member

    Joined:
    Mar 9, 2016
    Messages:
    1,399
    Likes Received:
    644
    There's little doubt that the long-standing FFB model needs to be updated and never could have anticipated the direction hardware has gone today.

    Not only has the hardware evolved to a high degree but, also the software; who would be crazy enough to tackle such a monumental task and what would be the reward? It would likely require some major game-side coding to be done as well.
     
  20. Rodders

    Rodders Active Member AMS2 Club Member

    Joined:
    Jul 25, 2016
    Messages:
    244
    Likes Received:
    54
    There is a controller.ini file in your profile\documents\Automobilista\userdata\<Profile-Name> folder and the game installation folder.

    In addition if you save a custom preset it's in your profile directory controller folder and looks to have the same in it as the controller.ini

    What ones takes precedent? ie what one should I customise if I want to manually tweak anything?

    Thanks.
     

Share This Page