The FFB is merely acceptable for a racing game. It has always been like that in my subjective opinion. All the way through the builds. Lets hope they can unravel this horrible mess. As for the myriad of over complicated FFB adjustments, well they are just not wanted or needed. Are they present on the console versions? I can hardly see console users messing for hour after hour dialling in some sort of usable FFB.
I really do hope they can sort this out to an acceptable degree with presets.
I understand and respect why you would opine that the 'over complicated FFB adjustments' are not wanted.
But I would also like to take this opportunity to illustrate why we felt the FFB adjustments were indeed needed, and how we approached the solution to the FFB wheel problem in general.
FFB Presets vs FFB Mixer
We had a lot of discussion back and forth with AJ about which options to expose and how (in an ideal world) we'd have both the advanced options and then later on -- once we'd explored more of the FFB problem space -- we'd introduce something that looks more like a traditional amplifier with a few knobs (like bass, mid, treble, loudness) for people to twiddle. We had hoped to include both in the released product, but compromises needed to be made and here we are.
What we have now is the FFB studio mixer table where you can 'engineer' the FFB to your tastes. What we hope to introduce with presets is something approaching the 'pop', 'rock', 'jazz', 'hip-hop' presets that some software music players offer. You could even argue that this is also what Jack Spade (and earlier in development, yours truly) has been doing all along with his FFB presets.
FFB Response and Linearity
Now for a bit of background: Different FFB wheel models don't respond the same to similar inputs. And in fact, there is also variability within each type of wheel. So even if David Ignjatovic and I were to each use our own G27 wheels, the two of us using the exact same pCARS FFB settings might not yield the exact same response from our wheels all else being equal.
For instance, it took a while to discover that on e.g. G27 wheels, if you set the FFB motor strength to 100%, the weakest signals from the physics subsystem don't actually get translated into FFB motor torque, which means that you can't feel them (as an aside, this is why some people recommend setting the Overall Effects Strength in Logitech Profiler to above 100% -- this minimizes some issues, but in turn creates other issues).
This was a particularly thorny problem because the weakest forces are sent to the wheel around top dead center, making the wheel have a much larger deadzone than the physics data would suggest. For e.g. CSW v2 wheels, this problem doesn't exist as far as I'm aware because the hardware is simply better.
G27 wheels also do not have a linear response curve when plotting input signal strength on the X-axis and the resulting FFB motor response on the Y-axis (0-16% input = no response, 16-40% input = 0-80 degree response, 40-100% input = 80-150ish degree response) similar to what this modified graph (sourced from an
isiforum wheelcheck.exe graph) shows:
(the above image illustrates my understanding of the stated parameters. It might not be fully accurate, but it ought to give a better intuitive understanding of what the parameters do)
And again, other types of wheels are likely to behave differently.
So, in order to encapsulate this natural variance in a single, coherent FFB mixing framework driven purely by the modeled suspension geometry (with a few 'physics-derived' effects optionally mixed in, such as Seat of Pants and Gut which use loads and G-forces measured at different positions in the car and not as part of the steering geometry), AJ came up with the current v3 and v4 topology of the FFB system.
The two topologies are functionally the same, where the v4 topology merely exist to allow us to tell the FFB that it should take its configuration values from XML configuration files and ignore the corresponding values in the UI (v3 = UI, v4 = XML overrides UI).
For the G27 in particular, two functions are important:
The
Scoop function compensates for the non-linearity of the wheel, which means it counteracts the fact that on the graph, the slope of the response curve is steeper at first and then at some point changes to a flatter slope for the rest of the output range. The point where it changes is called the scoop knee. So the physics-driven FFB signal is sent through the scoop function with the goal to take the physics-driven steering column torque as input and modifying it so that the end result is that the torque response from the wheel is linear.
Similarly, the
TightenCenterRange function (a.k.a.
Deadzone Removal Range in the UI) was developed as an alternative to the SoftClip function such that forces below the threshold in G27 wheels can be boosted exactly to the point where you begin to feel the FFB motors working, removing the limpness around TDC. This constitutes a form of signal compression. The exact value needed between wheels typically varies (as does user preference!).
Force Components and Master Scale
But to make matters worse, the cars in the game do not weigh the same, they do not have the same downforce, their tyre grip varies significantly and nor do they have the same suspension geometry. This means that the forces applied to the steering column can vary both between cars, and particularly, the column torque on a high downforce car can vary significantly depending on its speed (hello FA!).
This is why there is both a global Tyre Force scale plus a Master Scale for each car, and why you can tweak Fx, Fy, Fz and Mz independently, as at e.g. high speeds in the FA, Fy (lateral force) translates directly into a powerful centering spring force that is capable of completely drowning out all other signals. So by turning down only Fy on the FA, you create more breathing room for the Fx (longitudinal force -- 'braking feel'), Fz (vertical force -- 'road feel') and Mz (self aligning moment -- initial resistance to turning and very useful for perceiving the amount of countersteer needed) components, while you still retain a sense of increased lateral cornering loads.
The Power Steering problem
So, let's say you manage to configure your FFB to have completely linear torque response and your wheel cannot go beyond 10 Nm of column torque but your car produces up to 40Nm column torque. In that case you have to carefully think about which part of the FFB response is important to you because you will need to 'squeeze' the physics signal into the available FFB motor range somehow -- all without losing the parts of the signal you deem essential (think of Kimi's woes with the 'feel' of his power steering systems for the past few seasons).
The
SoftClip functions help you achieve the above, in that the useful signals can be allowed to fill, say, 90% of the range of your wheel, while spikes (high-frequency, high-amplitude) signals can be compressed into the last 10% of your wheel's torque range. That way you can perceive subtle balance and grip changes, but your wheel won't knock and rattle itself off the desk when you hit the kerbs hard or are in a collision.
In some ways, this is the same problem that power steering designers solve, except that with power steering the limiting factor is the torque the human behind the wheel is able to handle, and then you can add on top all the other mechanical engineering design challenges such as the hydraulic power steering pump capacity and how to ensure that the power steering gives sufficient assistance at low crank speeds while not overboosting at high crank speeds, speed sensitive boosting, variable ratio steering rack etc. etc.
Freedom of Expression
And that is why the FFB system seems so complex: It is a conceptually elegant system designed to allow you to engineer exactly the type of FFB feel that you personally find useful, while allowing AJ to dial up the elements that help him develop tyre physics -- without both of you having to compromise because one 'preset' or 'personal preference' was deemed more important than another.
Alas, freedom of (FFB preference) expression comes with its own set of challenges it would seem.
