Well hardware plays a role in how good you can handle a car regardless of how bad/good/realistic/unrealistic the physics may be. If it wasn't a black art then iRacing, gMotor, Live For Speed, Netkar Pro, Driver's Republic, Assetto Corsa, World Series Racing, etc. would all be very similar to drive and they're not.
You don't seem to understand why sims are different so you simply call it black art. It is clear that you ignore the reason for it so you simply give the best reason you can think about. It has been quite common throughout history that human beings need to provide an explanation to whatever they cannot understand. You know God created the universe in 7 days.... As I said before, different sims take different approaches to simplify a more complete and accurate model which otherwise wouldn't be playable unless you had the same workstations being professional sims. Regarding the car itself, a nearly perfect simulation can be achieved in reality. Sadly it won't run in realtime. All sims use instead very simplified models with masses, springs and damping elements. Remember for example when flex chassis was introduced. Was it black art before? For sure not. Another very clear example of this is the tire model approach by ISI. The tire model used would not run in real time so they used a lookup table strategy. The data is actually obtained by more advanced simulations that take a lot of time to run. Other sims don't use that strategy but for sure it is no black art. The data available to each company is also a key for this. Formula 1 is the category where more data is available. Only software companies that have been working for F1 teams will have access to that data. In lower categories not so much data is available. So please do not repeat again that the reason for sims being different is because it is black art unless you like appearing as a big mouth. Enviado desde mi GT-I9505 mediante Tapatalk
OMG, what are you think you are? You know the the approaches rF2 team used and their models and methods how and where to transitioning to sliding state and on sliding state? You actually know the approxmations they did on they tyre solutions by reading those 25 page of ttool paper? You are joking right? The term of "black art" he used is totally proper. If you do not like, just be it. I had enough all those BS here!
As I said in the other thread, the ttool doc also links to (just one) document with papers dealing with various aspects of tyre behaviour (as tested, physically, with real tyres) and some observations on tyre models used over time. You'd find a lot of your answers there, and I'd be pretty confident in suggesting it's not ISI's only reference. Again, a fully sliding state isn't the most complex situation they're trying to model here.
Your assumption is that the approaches of ISI for sliding state is totally transparent to you and well understood and can be verified. He said in his video that it is NOT. They must be cooked by each developer teams. So he call it "black art".
My understanding of it has nothing to do with ISI's understanding of it... they've been doing sims for nearly 25 years.
Yes, what they are doing at a low level is a mystery to me - it's part of what makes their product theirs. They don't tell anyone how they do things, or other people will become direct competition. It wouldn't make any sense. The problem with talking about black art, black magic, is it starts to suggest no one knows anything about what should happen (with sliding state, for example) and therefore without them telling us what they're doing we can't trust it. That's how it sounds. Too often people who know nothing or learn only the basics then question what ISI is doing, or want to know more about it so they can somehow validate it - again, ISI have been doing this for years, not just for rFactor/2 and as an engine for other games, but in developing simulation engines for companies/teams whose success and ultimate livelihood comes from ISI knowing what they're doing. If you do that badly just once or twice it's unlikely you'll have a viable business in subsequent years. So I don't see much point in a lot of stuff people post about specific issues, because it won't lead anywhere. And in the longer term it becomes obvious ISI will do what they do regardless, sometimes with negative feedback but mostly with positive. They seem to be going ok with it so far so fair enough.
Geez, another stop feeding the troll post... I can work out when what I'm saying is being completely ignored, thanks
Science that is based on unproven theory has more unanswered questions than the science based on known facts, hence the sliding tyre is more of a mystery than the one that is within the parameters of mathematical certainties. I 100% agree with your statement that the differences in sims are not only explained by this "black art", which means in this case the data that is either inconclusive or totally absent and has to be interpreted or invented, and then tested within the simulator while making adjustments to get the desired results. Kinda like a crossword puzzle, if you have a word wrong, the whole puzzle is affected by it. Obviously you cannot use this kind of simulated data for any real world application that requires the up most accuracy. These driving simulators we are all about, has to interpret a lot of data and end result is only as good as the interpretation. The result is then pulled through all kinds of gatgets one uses to experience these games, then we enter to our personal preferences and it's anybodys guess whatta hell is that experience going to be like for any one individual. Without black on white, all one can say is "I feel like", and in scientific terms that is useless waste of time when debating facts. "I feel like" this whole topic has turned into a mere interpretation contest of the frase "black art", rather than talking about the OT, and that is just like a question Batman here has to ask:
Golanv, I basically agree in most what you said. However, I disagree in one of the aspects which might be due to misunderstanding. You talk about proven theory vs unproven theory. I am not sure if I am understanding those terms correctly. Let me explain. If you measure a material Young's modulus over a range of temperatures you will typically a variable figure, usually higher at lower temperatures. If those measurements are consistent throughout different tests, you can actually affirm that it is the way it behaves regardless you cannot fully explain the reason for it or predict it's value out of the range of testing. Using those values for simulation is not black art. It is simply using empirical data. For sure, kinematics is a much precise part of physics. The basic F=m•a applies 100% accurate. Imagine if it is precise that recently a spaceship was delivered to encounter and land on a comet. Rosetta mission. If you take a look at the fly byes it needed to do it looks like a miracle that it actually reached its destiny. The same applies for friction which I guess it is the main point in this thread. For sure it is a complex phenomenom due to its non linearity and the large amount of variables involved. Most of the data to be used are empirical but have been deeply analysed. It should be noted that friction and wear have a dedicated branch of physics: tribology. Friction and wear affect all mechanical devices and needs to be controlled to guarantee a correct functioning over time. In some cases friction is not an enemy but a friend: clutches, brakes, tires... In those cases wear is topically the enemy. Pin on disk/wheel tests are extensively used over different ranges of temperature to account for both fraction and wear. For sure all the data obtained are empirical and to be used with typical coulomb 's formula Fr=mu•N where mu is the friction coefficient and N the normal force applied. Wear is also measured and some wear vs sliding distance coefficients are obtained. Normalized values would take into account the effect of normal force providing a coefficient of wear vs amount of work done by friction: Heat. So if sufficient friction tests are available a car sliding should be easier to analise than a car rolling. the reason for this is that the relative sliding is much more constant for each wheel across the contact patch. This situation is very similar to those pin on pisc tests that are performed. So that's why I say that sliding state is easier than rolling. The main difficulty here is the effect of rising temperature which certainly affects friction. In order to account for this, pin on disk tests are also helpful but need to be carried out with a different approach: short tests to analyse the first seconds of the test where temperature rapidly ruse before reaching a steady state. Really I cannot see where is the problem here in using a normal scientific method to simulate tire behaviour. In my job I have to simulate molten glass and how it deforms in the process of bottle forming. As you might know, viscosity and thermal properties of glass are very variable with temperature. Besides they are incredibly hard to measure precisely due to the high temperatures involved (from 500C to 1200C). That is black art nowadays due to the lack of data. But friction??? Come on!! Enviado desde mi GT-I9505 mediante Tapatalk
I was referring to theory as already been accepted, like the term is used in science, as valid hypothesis that is supported by evidence. The amount of evidence determines how good of a theory it actually is, and that is what the ants of science are doing, trying to find more and more evidence to support their models that explains things. I do not see a problem with using normal scientific methods and the data available to simulate tyre behavior either, that's what the car and tyre industry have been doing for decades, and with great success I may add. The margin of error has become down to the point that you may say the simulations are actually accurate enough for pretty much any use where regular vehicular activities are concerned. What comes to the simulators we play with, it's all about how much data is available, and more importantly as you already mentioned, how much of it can be used before realtime simulation is not an option anymore, which is where the tyre model and the physics engine needs to start to interpret data and use look up tables for example. There is a lot of data about friction concerning rubber, what makes tyres a difficult topic though, is the nature of the materials and circumstances involved. Tons of variables that are very hard to measure accurately, not to mention repeat the experiment with consistent results. As you stated, it is hardly a black art, merely a difficult to get it right with these methods our racing sims have available. I found this kinda cool study on frictional properties of rubber from 1942. http://nvlpubs.nist.gov/nistpubs/jres/28/jresv28n4p439_A1b.pdf
No one disagree with that. But if I understand you correctly you might imply that at a high level is transparent to you. On this subject—Sliding state, as he said in the video that when and how to start to sliding and/or to spin-off is not known and no theory published about that and cannot be validated by an object way. Even a high level approach to deal with that may well be as “black art”, not revealed to you. Sure, we all have some faith on ISI. Otherwise, we would not be here. I stated my reason very clearly when I brought this issue. I did not feel realism of rF2 sliding/oversteering. The rF2 cars were way too easy to enter spin-off. A year ago if one had claimed this is realistic, I would be sure LOL. I love the rF2 grip state I feel very realistic. Although rF2 improved a lot recently but I have no time to drive too much in order tell. I did not feel realistic AC cars sliding either. Their cars grip ground too much and hardly getting into sliding, and far less feel on tire dynamics. Once I had, on rF2, my FFB and motion rig shaking experience caused by “tire flatspot” (as most of you suggested), I have a doubt on realism of their tire damage model. Naturally, it is in sliding state. This is why I brought the term of “black art”. If you do not buy, it is find. This is my concern. Sure you love their foods, taste that good. I do too. They offer a great menu tell you what they are. The cooker chats with you every day. For a particular dish, you just could not find anywhere else. You know what, you will never find out how he cook, unless he tells you, no matter how long you have been here. Because it has no theory base and no data can be validated against so he cooked in his own way-----“black art”
I don't think people here understand the difference between hard data and making a physics engine which controls all those thousands/millions of variables. iRacing, ISI, Live For Speed, AC, World Series Racing, Netkar Pro, etc. etc. all most likely have tons of hard data including the easiest most basic stuff like the gravity of earth makes objects accelerate towards the center at 9.8 m/s^2 or whatever, YET, even the simplest stuff like how cars float/fly in the air (upon getting airtime) is completely different in every sim, let alone actual vehicle dynamics, lol. "Ok, I just got some hard data from tyre manufacturer 'x', plug in those numbers and, voila, my physics engine is all of a sudden the most magical, god-like physics engine of all time with the most perfect vehicle dynamics behavior"....Right.... lol. Gmotor, after 15 years and many physics updates and multiples tyre models, still doesn't simulate oversteer anything like how it happens in reality. Netkar Pro, Live For Speed, and Drivers Republic are the only games that do this extremely close to reality. Gmotor, after 15 years and many physics updates and multiples tyre models, still behaves like the car rotates around the body rather than around the wheels/tyres that are sliding if that makes sense. Gmotor, after 15 years and many physics updates and multiples tyre models, still has this weird behavior where cars want to keep turning-in and keep turning-in even though you aren't applying much lock. I went around all of Bathurst with the latest-and-greatest Corvette without turning my wheel more than 60 or so degrees with the stock steering lock. Think about that for a sec. That is ridiculous. I did the same thing with a Mod Porsche at Adelaide. In-fact, I didn't even have to use more than 45 degrees of lock with the Porsche. That is completely retarded. This front-end wanting to keep turning-in for you allows this to happen. Gmotor, after 15 years and many physics updates and multiples tyre models, still behaves as if - instead of the car wanting to go straight but only turning because of the friction of the tyres making it turn- with gMotor it's as if a hand comes down and grabs the nose of the car and turns the nose according to your steering inputs rather than the car wanting to go straight and turning due to the tyres' friction making it turn. Gmotor, after 15 years and many physics updates and multiples tyre models, still has this effect where you can't properly control and dictate oversteer moments, but instead, it's just a quick lift-off-throttle-and-correct-and-snap-back-to-centre reaction. Gmotor, after 15 years and many physics updates and multiples tyre models, still has issues with low speed grip/slip behavior (anything under 80 km/h or so). I could keep going, the point is, stop being a fanboy and thinking that RF2 is some magical, superior-in-every-physics-way sim just because they have a complex tyre model. Complexity doesn't mean anything if the end-result is crap (I'm not saying RF2 is crap, it's just an example, in-fact RF2 is my favorite sim). --- Pay attention to what Jameswesty says, as-well as hiohaa, and BSNismo (FONismo). I know it's a lot to get through. Ignore most of my posts, I was a little more blinded back then as I was a fanboy and not absolutely fully unbiased (even though I probably said I was). I can't believe I did so much bashing of Netkar Pro back then, geez, the fanboyism absolutely corrupting my mind and sense of un-biased, critical thinking. http://isiforums.net/f/showthread.php/6358-NKP-and-RF2-why-is-NKP-superior-in-terms-of-raw-car-handling http://isiforums.net/f/showthread.php/7152-What-I-believe-to-be-wrong-with-the-RF2-tire-model-grip-levels-lack-of-control http://isiforums.net/f/showthread.php/6543-Is-ISI-still-working-on-the-tire-model http://isiforums.net/f/showthread.php/19348-BSIMRACING-article
No one's saying that Spinelli. Of course approximating tyre behaviour isn't easy - if it was everyone would be doing it, and all sims would feel the same. But are they generally close enough to be close enough, without quite feeling perfect? rF2 seems pretty close... but then along comes Joe questioning various aspects of the physics, questioning flatspotting, wanting it turned off because he thinks it's some artificial extra effect on top of the tyre wear that's part of the model. The forces acting on a tyre in a sliding state, the shape of and pressure distribution across the contact patch, the temperature (change) across the contact patch via multiple repeated tests - there is data for all of that, from years of research into tyre performance and design, and when you build a model to simulate that of course you validate your model its output to real data. How close is close enough is up to the developer, but coming along and saying it must be wrong because you think it doesn't seem right, and flatspotting must be separate damage because it feels like the tyre is damaged (sorry to harp on about flatspotting, but Joe did bring it up), really makes it sound like ISI threw together a model in a week and are hoping no one notices. Joe doesn't help himself by questioning the entirely wrong data (see the current thread), and making assertions without logical reason (or saying that my quite logical assumption is my belief, insinuating that his illogical assumption is just as likely). By the way, when a moving vehicle gets airborne gravity is the easy bit - it's the aero forces generated by the various parts of it that make the difference, especially at extreme angles that probably aren't really measured for normal driving, plus the weight distribution. Just as an aside. (and that's a case in point - getting everything perfect is impossible, but getting a lot of stuff very close is achievable) *I obviously replied to your pre-edit post. I think I'm done here.
FWIW, Dave Kaemmer addressed a lot of these points in old videos. He says that most tyre testing and research has been for road cars and there's "big dearth" of data and understanding of sliding in race conditions. He also describes the coefficient of friction as being a "useless concept" for tyre rubber. Not sure if it's this video, but he has said that aero is complicated but you can make measurements and "parameter-ise" (is that a word?) it well enough to get reasonably consistent results. But tyres are more complicated and one of his team who was a race team engineer said that everything was understood except tyres and thought what he was doing was really cool. https://vimeo.com/15935676 I'd have thought that with so many variables (track surface, camber, tyre carcass distortion, degradation, hysteresis, wear,temp, pressure, non-linear transient suspension forces, aero etc. etc.) you can't actually achieve exactly the same circumstances twice to make accurate measurements that are generally relevant. Chaos theory comes to mind. He thought back then that he'd got the right numbers coming out of his model but there have been big changes in his tyres since then. Over the years I've become more worried by the people who think they understand this subject than those who know they don't.
I agree Old Hat. I'd like to just be able to better feel the rear of the car via the FFB. Some other sims have a more natural/better feeling of the weight and weight shift, even if the physics are less sophisticated. At this point I do not know enough nor do I have enough data (even if I did know) to distinguish between a true problem of lack of rear grip and a problem of the right signals not being transmitted to me so I can manage limited rear grip. It seems like the problem is too little grip at the rear, but I'd like to be able to better manage it via better FFB before coming to a conclusion. And the ability to have what the rear tires are doing transmitted through the FFB to a greater degree than now would be a good thing to have, especially if it was tweakable.
I don't agree, Marc, about it being about not feeling the rear through FFB. I can feel the rear just fine. I can even save the majority of oversteer with no FFB at all, you just gotta get used to it. The problem is the vehicle behaviour itself regardless of user inputs and user feelings (FFB) and the weird way oversteer occurs in ISI engines which is not akin to reality and many people have noticed this over the past 15 years of the ISI engine. Something is off regarding forward and sideways vehicle inertia and momentum as-well as how tyres slip and re-bite throughout the sliding-and-correcting process as-well as possibly body/corner weight movements and inertia transfers as-well as super odd pivoting based physics/forces where the vehicle rotates around some pivot point rather than around the tyres/wheels that are slipping/sliding. Although it's worse in RF1 (SCE) and prior engines, it's still very much there in RF2 despite all sorts of advancements in the core physics engine, a completely new (and much better) tyre model, etc.
