If i remember, the purpose of position feedback was not to provide "canned effects" as some people have said, but so that forcefeedback could be properly simulated. At the moment, a force is applied in response to your input. The problem with this is that the force can "over-correct", due to it being too strong. This is why you can get those increasing oscillations when driving down a straight without damping, if you leave the wheel central and let go of it. With a position-feedback device, you wouldn't get these over-corrections via forces, because the steering wheel is being told where to go by the feedback. Basically, in terms of simulating the steering column, it is like comparing pushing a marble with a specific force and hoping that it stops in the desired place, compared to actually specifying the desired place and forcing it to that position. If you think about it, in real life you are inputting force to a steering wheel, not steering to an arbitrary position. When getting feedback in a real car, you are getting the feedback as a result of the steering column (and car wheels) having moved to a specific position. The required force to move from that position is obviously important, but equally so is the position itself. With regards to seat of the pants feedback, i have seen very little talk of this device: http://www.ivibe.com/ There is all the talk about sim-vibe but i think this one is better since it is far cheaper and is based on actual telemetry, with multiple (individual) vibration motors within the seat. Sim-vibe is kind of a far more expensive diy version of this. And then there are the ones that work purely on sound data (such as the butt kicker solution) which seem rather pointless!
Wrong! Force feedback is all about reaction. Such a thing as applying a force, or better said torque in this case, is just your misunderstanding of fundamental dynamics. The force feedback appearing in the wheel is completely determined by the position vs time input. The ingame ffb should just do the same. I can discuss this forever since it is really simple if you know what you are talking about. I am mechanical engineer (first in my promotion and I have always been the best in my class in high school and university in maths and physics. I barely had to study for exams since I understood what I was dealing with. If you want, you can consult it to a professional in the matter. I would like to read ISI's opinion on this. The reason why Leo Boznan has not contacted anybody is because I think he is not really convinced of what he is saying. Enviado desde mi GT-I9505 usando Tapatalk 2
One more observation. I dont know how exactly ffb devices actually work but I would like to know how accurately do they provide the expected force feedback which is the output of the game and what they should reproduce. IMO this is the most difficult part of the forcefeedback issue. The reason is pretty much related to my previous post. Providing an accurate instant force or torque to the wheel shaft is very complicated. Contact forces as the generating force source should be avoided because they are difficult to control. Electromagnetic forces coming from a controllable direct voltage should be used as in a DC motor occur. However controlling voltage could be complicated at such high framerates to provide a narural feeling. Although things occur fast in electronics, the presence of condensators and inductances play a role in the time it takes an electrical system to get to equilibrium. If transitory times are in the same order of magnitude of the time steps between one ffb output and the next one it could be certainy difficult to control a torque delivering device. With repect to input data to the sim, ffb devices should provide position, velocity and acceleration of the wheel. In order to do this the encoder resolution and working frequency should be better than the one the sim uses. Better means that if rf2 calculates ffb at 60 Hz with 1/100 of degree in resolution I would recommend at least about 200Hz and 1/400 of degree in the ffb device. Otherwise velocity and acceleration calculations will depend on several sim time steps which may result in inestability or resonance effects. Just my two cents. Enviado desde mi GT-I9505 usando Tapatalk 2
I read that article a while ago, and have not read all the responses in this thread, so let me just give you might quick take: It's a clever idea by Leo. And it would actually make the physical simulation much more clean and straightforward. F=ma is easy, that's what we do with everything else. Kinematically controlling the steering, on the other hand, is a bit more difficult. In simple terms, we have to reverse engineer it. However, I don't think it solves the fundamental issue which is latency. It might *change* how it manifests, but I'm not sure if it would be better or worse. Instead of steering the wheel and waiting for a force to come back, you would apply a force and wait for it to steer. In either case, there has to be sensing completed, communication to the computer, a physics calculation made, communication back, then a response from the motor. Still roughly the same amount of latency, as far as I can tell. It would make for an interesting experiment, but I'm not personally convinced it would be a groundbreaking change. It would take some work to undo our current system since it is baked in, and to be honest we have some higher priority work to do at the moment.
I disagree. Cars from the early 1990s had great FFB. In fact, the Megane in rF2 feels almost exactly like my 1991 Sentra SE-R. It's a shame that most modern cars have such dead and lifeless steering.
There is no need to act all superior in your attitude and make out that no one else understands. Of course force being applied to the wheel is determined by time, and position of your wheels, but also the forces acting on those wheels. If you just say it is "position and time" then with the wheel turned 45 degrees to the right, would you expect the same forces over 3 seconds, travelling 150 mph through a corner, compared to being completely stationary? Of course you would not! Force feedback and "position feedback" are both 2 methods of achieving the same thing. The system itself is a loop-back system in ideal circumstances it would make no difference which used. Due to circumstances in the real world not being ideal, what you will find is that the trade-offs are different for each system, and from what i can imagine, a position feedback system has less severe negatives to it. What Leo Bodnar is suggesting is another way of looking at force feedback in order to avoid issues where the forces overshoot and you get oscillating effects. It also would allow for easier simulations of things like power steering etc. In real life, when you turn the wheel you are applying a force to that wheel, and that force will move it into a certain position. It i s the same with the car tyres themselves - they have forces applied to them that will move the steering column (and your wheel, unless you resist) to a certain position. The relationship between steering position and forces on the steering column (wheel) is fixed. What is happening here is simply the reversing of which one is the input and which one is the output. The system loops back on itself so in an ideal world with no latency, it doesn't matter. However it does matter as discussed: Exactly, it would solve the oscillations in the centre of the wheel and other problems, due to forces "overshooting", it's just the negative impacts are that you would maybe have delay on your steering forces being input. I think it would probably be preferable this way because you will never get a case of the wheel's FFB "fighting" with itself (due to latency), and a tiny delay in the wheel responding to your input should be so small that it is rather imperceivable. Certainly it is worth trying
Thanks for the response Terence. It is a fantastic reflection on a company when a developer takes the time to engage with the end user =) Understandably priorities come first in the development process, but i can't help but hope someday you guys find time for some experiments perhaps with Leo on board as well. Thanks again =)
Now that I am in the computer and can write easier I will explain in more detail so that it is better understood: When I say FFB is only conditioned only by the position vs time profile of the input, it means regarding the input itself, which is what we are talking about. Everytime you apply to a determined system the same input profile in terms of position vs time of the wheel, the reaction of that system will always be the same. There is a unique force that provides this input profile and it will need to be provided by the driver to behave in that manner. It is pretty obvious that FFB depends as well on the reaction of the road to the wheels which is simulated and considerd by the sim FFB algorithm. Besides the wheel force component is the biggest in the equation in terms of FFB. However, wheels are in the other end of the mechanism and we are talking about the effect of moving the wheel in the FFB itself and how that input profile should be "measured" and provided to the sim for FFB calculation. If you don't move the wheel you get different FFB that when you do for sure. However, the difference between them should not very big since steering mecanisms don't have big friction, inertias and are quite stiff. When you leave the wheel loose is the only time that you can talk about applying a measurable and explicit steering force: No force or zero force, as you want. Then the movement of the wheel would be completely dictated by the reacions in the wheels. Therefore we sumarize: - FFB in the steering device is dictated by wheel and steering inputs - The steering input profile can be easily obtained by providing the sim with steering position, speed and acceleration values. - Wheel inputs are calculated by the sim according to the tire forces which as simulated in the physics engine. - Therefore FFB can be perfectly and accurately calculated according to the above. - Wheel reaction plays the most important in the combination of both. - You will need to apply that force to the wheel to move that way. If you understand the above, you will easily understand that: Trying to measure an applied force to the steering device if this one is not provided by the sim as a counterreaction is just impossible. It is trying to measure FFB to calculate FFB. Another way to see it is if you suddenly disconnect the electricity in the middle of a fast turn when you were applying quite a big steering force. The wheel will rapidly turn and you will not be applying that force any more because you have no reaction in the steering device. Remember FFB is a REACTION Once this point hopefully is more clear now I would like to know if somebody knows if the FFB output from rf2 (and other sims) would be the same independently from the wheel used. It shouldn't since its steering device has its own inertias, frictions. ISI would need to actually simulate each steering device so that they can compensate for the actual forces coming from the wheel internal inertias, stiffness and damping. Maybe this is the aspect that is not fully well simulated and is providing the difference between real and simulated FFB Leo claims (I don't either agree or disagree in this one since I have very few experience driving without powersteering. Apart from karting and one time that I lost power steering (due to the parking distance sensor was broken I had to drive home without it) I have no other experience so it would be just stupid to give an opinion. I just do it when I think I have something interesting to say. To prove that I am qualified to do so I gave certain references. It might sound superior but I don't care about it. I had my purpose to do so. The reason why I say this is because if there is a difference between real and virtual FFB must be because one of the following: - FFB device cannot provide the SIM output force due to a lack of precision, available maximum force or at insufficient speed. It would be nice to be able to measure real FFB in the device to make this comparison. Not for the input, just for calibrating and testing the device itself. - The error in the physics engine providing inaccurate values for tire forces. - FFB device does not provide inputs at the required rate to accurately calculate position input profile in terms os position speed and acceleration. I don't know and I think it is important if the wheel is just providing position and the sim estimates acceleration or the FFB is doing that. I think it would be better that the device did it since the calculation is very simple and the internal firmware should easily handle it ad avoid stressing the computer and the SIM unnecessarilly. - FFB device is not considered within the sim for FFB calculation and compensation. This last one which is what I explained in the paragraph above is for me the most complicated part for ISI since they would need to test and parametrice all FFB devices. The compensation of forces should not give an important difference in terms of overall FFB magnitude I would say. However is a guessing and I would have to make some calculations to evaluate it more thoroughly. For being difficult an costly and probably not very important is why I think this concept might have been simplified or at least standardized among all wheels. Position feedback is just impossible. This is the craziest thing I have ever heard of due to the explained above. The output of the wheel cannot ever be to move the wheel to a certain position. We are driving and putting the wheel to the position we want. The SIM should not calculate what is real as the position of the wheel which is controlled by the driver only the virtual part. Imagine in a laterally cambered straight where you needed to provide a constant force to maintain the steering perfectly fixed. How would you calculate the position for that? The wheel is fixed. only thing to be calculated is FFB. It is not that hard to understand I think.
I think we have a language barrier going on in this thread. Some people are arguing the same points in 20 different ways. This discussion would probably be more productive in person on a whiteboard. Also, in person there probably wouldn't be personal attacks because the fist could be utilized.
I've only browsed this topic, and haven't gone too in-depth with your posts or 2tyred's, but is it possible you're misreading what is meant by position feedback? (I'm not totally sure myself) Take your example: On a 'cambered' straight the wheel pushes to the side as you're driving straight. Why is it pushing? Because the camber wants you to drive downhill - if you let the wheel go it would turn down the slope. This force could either be the game saying "you're holding the wheel straight, but your tyres want to go this way, so here's a force", or, the game could say "your tyres want to point 4° to the right at the moment, which is 60° wheel rotation" and your FFB wheel says "I'm 60° left of where there'd be zero force, so I'll apply force X". Either way you end up with a force on the wheel, for the same reason, but in a different way. Is this somewhere close to the mark? (obviously this second method wouldn't work without hardware designed for it [without introducing more latency as the wheel driver intervenes] but isn't that was Leo was saying?)
Interesting thread and thanks for the replies. I suppose there is always room for improvement but I'm quite happy with the current state of ISI's ffb.
Physically when you turn the wheel in a car you move the wheels on a position basis and any resistance to this turning is relayed back to you as resistance. This is exactly the way FFB works for us in rF2 and in any other sim. Leo sounds a bit confused here, since really he is missing the fundamentals of Newtons 3rd law. If you want to actually turn the wheel, you must apply a force to overpower any static forces and this is how FFB works. What he postulates is a FFB system that acts exactly in the same way as it does now, but in a completely backwards operation. The only problem we have in getting proper FFB is related to physics in the sim, update rate/latency of the connection and also that most of us are using relatively cheap FFB wheels.
I can assure you, 75%, probably more, of feeling the grip of a car DOESN'T come from the wheel. When the car is talking to you most of it is not done through the wheel, and a road car is even worse than a racecar.
Are you sure? I might be wrong but when I am in the passenger seat I dont have the same feeling if the car is close to the limit of grip as much as when I drive. The only difference between both situations is the steering wheel. I am not discussing your point but introducing into the FFB artificial sensations is wrong IMO. Since you seem to know a lot about where sensations are coming from. Can you explain what you are basing on? Can you provide some bibliography or related links to support your statements? Are you amateur or pro racer? Do you drive usually enough without power steering? Enviado desde mi GT-I9505 usando Tapatalk 2
As someone who can suffer motion sickness within 5 mins as a passenger (depending on the situation) but can drive for hours on end with absolutely no sign of problems - I can assure you your brain learns a lot about what to expect from your actions as a driver. You lose all of that as a passenger. So a car's movement in relation to what you'd expect from the inputs, is much much harder to ascertain if you're not the one controlling the inputs. Basically, especially after you ask for references and racing experience from someone else, talking about your experiences as a passenger in a thread about FFB is completely nonsensical.
Excuse me but I cannot fully agree, Lazza. I didnt talk about dizzinees but about the grip/sensation which is related to the thread. It has been stated that more than 75% of the sensation is coming from other than FFB. Those other inputs would be common for a passenger or a driver. Now you say that the driver get a lot of info from the response to the driver actions. I guess that those or big part of them at least should be in a sim as well. Since you are relating that these things are being felt by the driver feels and not by the passenger we cannot be talking about actual g forces which are the only not possible to be simulated properly unless experiencing the same real accelerations which would mean that you would be describing the same movement pattern as in a real car. When you drive to the limit, a variation of 25% in the grip feeling is just huge IMO which could cause the mentioned difference between driver and passenger. Visuals and g forces are the same for both. It has been suggested that FFB effects should be canned to compensate the lack of acceleration in your body and I dont see it that clear. In my opinion the human brain is too complicated for us to understand. The theory I support about this is that the brain learns to account for small details to get perception of things. If you lack g forces you will enhance other senses the same way a blind person does with smelling and hearing. In this way FFB plays a determinant role and my opinion is that it should be as relaistics aspossible. Trying to switch FFB and be fast in the track would be impossible for me. Visuals are important but FFB as important for a good racing simulator. Now if you want we can open a debate of why people are usually faster with a good wheel than with a bad one. Is it because T500 or fanatec are doing a better job canning the effects? In my opinion is because they offer a more realistic FFB than the cheap one. BTW I still havent received any reponse about the questions I ask so I understand nobody has any idea. Enviado desde mi GT-I9505 usando Tapatalk 2
http://it.wikipedia.org/wiki/Niki_Lauda Niki Lauda was nicknamed "The computer", because of his incredible ability to identify, like a computer, all the flaws, even the youngest, who was driving the car and the meticulousness with which he put up his own mechanical means (after the accident of 1976 that left him disfigured, said that he preferred his backside to a pretty face, just because he was convinced, rightly, that a car you drive especially "with the butt" ).
I'm afraid you've rather missed my point. Your brain picks up a LOT of movements (acceleration) in a car, and if you're the one holding the wheel and pressing the pedals you have a good idea what is expected and what isn't. A passenger won't know that a particular acceleration is a sign of, say, a loss of grip until it goes far enough to be obvious, or there are other signs (sound, visual). That will happen a long time after the driver knows about it, because the instant he feels something he's not expecting (based on his inputs) he knows something is up. Obviously I'm talking in terms of the brain here, not conscious thought that would be much slower. You mention FFB as something you need to be fast, yet a lot of people have gone very fast without FFB (remember GPL had no FFB at the start, and there are still people who shun FFB wheels and go very fast). I'm the same as you because I learned to drive a sim with FFB, but losing sound has the same effect as well. So, let's say FFB helps, let's say sound helps, and along those same lines let's say the forces you feel in a real car also help. The question is where do those things help - and in the moment of starting to lose the rear end, the force through the steering wheel will drop slightly. Your front tyres haven't really changed what they're doing, you're not suddenly losing front-end grip, so there's no 25% drop in steering force. What's happening is the car body rotates maybe a degree or two more than expected. How much do you expect to feel that through the wheel? You will feel it through the seat much earlier. And in many sims - rF2 included - you might start to hear it from your tyres, probably earlier than in real life. If you make the FFB completely realistic, and the tyre sound completely realistic, you're going to make things a lot harder than they would be in real life. So in the absence of seat/rig movement it makes sense to 'enhance' both.
