Automobilista 2 V1.2.0.0 & Racin´ USA Pt1 RELEASED - Now Updated to V1.2.1.4

I thought it was sim(ulation) but probably I got it all wrong.
By the way: never heard of mathematical equations describing physics? There seem to be a few thousands guys every year spending their time studying them and getting a master degree on those and several companies making a living out of selling software simulating mechanics before the prototypes are even created. All of them wasting their time I must assume?

 

Until the internet it was the guys that knew the formulas to design things properly that made a good living at it. Now i can find a formula to build pretty much anything i want on a car correctly with a push of a button.

 

Bruno is right in that "a model of an object is not the object". Even celebrated models in physics like Newton's laws of mechanics fail at very small scales and very large scales, hence the need for quantum mechanics and general relativity. Every model has a limited scope of validity, which should be kept in mind when striving for maximum "realism". That is true for a scale model of a building and true for a complex system of nonlinear partial-differential equations like a physical tyre model.

To make this philosophical argument relevant to sim racing, some common fallacies can be linked to this erroneous thinking:

1. Comparing sims to others sims instead of reality ("AMS2 doesn't do what ACC does, therefore it's unrealistic").
2. Expecting to see things that occur in simulators also in real life ("race cars must spin like tops if you give them a bit too much throttle").
3. Expecting that using real-world values of setup parameters etc. will make the simulated cars drive better/be more realistic.

 

I think something worth discussing is that a more specified model does not equate to a more accurate model. Whilst I come from a statistics background and not engineering/programming background, I still believe that this is relevant.

It’s possible that when creating a model (any model, not just a tyre model), that the model could be over-specified. It is possible to run into the trap of modelling error terms instead of the data-generating process. This results in a model which generates highly erroneous results, as opposed to a simpler model.

English: More simulated parameters in the sim does not necessarily translate to a more accurate sim.

 

Thanks for that explanation and video links, will check them out later. Diffs are the major setting that confuses me so hopefully this will give me some clarity, cheers.

 

Well F1 teams do a lot of CFD for aerodynamics now because of restrictions to wind tunnel testing. So that is a form of simulating air flows. If it is good enough for them.......

 

I understand the sentiment that a model of something isn't real, and so won't behave exactly as the real object would, but it should be very close in behaviour for the model to be considered a success. When I worked at BAe, we would almost exclusively deal with simplified systems because the full systems would be too hard to solve. This would be through linearising systems, which is a fancy way of describing throwing away the higher order terms, which is fine so long as the higher order terms aren't significant. Perhaps the solution to the linearised solution would then be used as the starting condition to solve the non-linear form. Whether we like it or not, maths is the language to describe physics and it's always an approximation, though things like F=ma are almost exact for most practical cases (i.e. no quantum effects and far from the speed of light).

Now, to argue that this is a reason we should expect or be happy that it isn't like reality just shows that certain things are far from right, and when this happened at work, many engineers would scratch their heads and work hard to understand why the mathematical model they spent so much time devising had failed. I could use this attitude to argue that I should be happy that in rFactor 2 I have to disconnect the rear anti-roll bar even though this is never done in real life. Well, I'm not, and it just shows that work needs to be done.

Since I just mentioned another game, I don't see why some are so averse to comparing games that are trying their level best to do the same or similar things. Bearing in mind that we enjoy sim-racing, and that we mostly have experience only of games, then when a new game we try feels different it may change the nature of the enjoyment we feel towards it. One may be right, and it may not be the one we've grown used to and like the behaviour of, or they may both be wrong (to varying degrees). Certain games may do things better than other games, and using one to hold another to account is a way of testing the validity of the modelling in each and so will help things move forwards (as will comparing to real life where there are marked differences, but the only real life experience I have are road cars, karts and a Formula Ford, so that direction of comparison is limited). You have to be sensible with your comparisons though, as it would be silly to argue that AMS2 is wrong because you don't have to disconnect the rear ARB to be fast, but it does show that point 3 made above is wrong because they don't do this in real life and we shouldn't be happy to have to do it in a game that purports to model real life. However, when trying to win races, you have to do what actually works rather than what works in real life. I hope that we all hope that one day the simulated and real world will be in broad alignment.

Comparing FFB is difficult because we're not in a real car feeling real forces - we feel nothing but what the wheel gives us. I've heard that ACC just models real forces resolved through the rack, and some love it and others say it's lifeless and awful. rF2 adds imaginary forces to give back some of the lost feel, and some love this and (guess what?) others hate it. I don't know what AMS2 does, but I'd say it feels more rF2-like than ACC-like (I only have a G29!). This is the one area we can all argue loads about simply because we're not driving real cars and never will be while sitting in front of a desk or (if we're lucky) in a rig.

Sorry, I typed too much again...

 

True, but differentials have quite known and relatively straightforward equations and unless you are looking to describe the behavior of the atoms inside the clutch, they work reasonably well

 

Please be aware that Im not in any way trying to rate the different reactions/answers to my The map is not the Territory post.

But I think the following post is pretty interesting.

Because this shows one of the fundamental differences between any sim(plified) model of reality and reality itself.
Because in reality (like as example the differential on a RL racing car) there are no excludable "error parts" vs some urgent data-generating parts.
They are all part of the events happening when as example a differential is doing its thing on the track.
Different parts have ofcourse either minor or major influence of the outcome - but...
But until a virtual model like a tire model or physics model in a racing sim is able to model RL events all the way down to the complexity on the atomar level - then any model is only a sim(plification) needing a lot of shortcuts its RL counterpart dont need or hold..

ByTheWay: Im not joking when I say that David Tucker one of the most uber-loyal staff disciples of iRacings famous TM guru Dave Kaemmer actually told everyone who wanted to hear it - that "our Dave" is actually researching in events on the atomar level to make iRacings tire model even more closer to reality.
No joke.

 

What’s the point of this „simulation is only simplified imitation of real life“-discussion?
We can either enjoy what we have now or go racing in real life (with all its advantages and disadvantages) or wait until simulators become lifelike (and maybe die before this will happen). It’s all your choice…

 

Answer:

If you read back in this thread then its pretty easy to find a lot of car setup advices and conclusions taken naively 1:1 from RL car setups.
And Im not sure such kind of mental shortcircuits allways just give people virtual joy and happiness

 

Yes, but in all the mathematical physics simulation we know the underlying algorithm, in Ams2 we DONT know , it's a closed black box. and how to try to understand how anatomy works without autopsy, so result, try to do the best times and laps with what you have.
Are you sure that all F1 drivers fully understand how the physics of their cars work? And where is the emotion, the instinct, the inexplicable? Was Fangio a physicist?
Our result on the track is all the fault of the physics of Ams2?
Are we no longer able to adapt?
I need a donuts.

 

This is similar to music compression. It doesn't matter if you'll have a lossless FLAC file or 128KB mp3, neither will replace full blown live orchestra. Same with simracing calculations. Even fully simulated down to atomic level system won't feel like real car. Period.

 

Simulators cost millions. Emulators not so much. I was in a simulator this morning. It has a room all to itself housing banks of computer stuff to run it. The parameters that it follows are all rated to the certification of the vehicle it simulates. The funny thing is the graphics aren't even particularly good! Trust me when I say that simulating reality comes with a hefty price tag and a lot of engineering support.
Emulating reality is no less potent though. Often times there is very little difference in the end user experience and the same human behaviours can be observed in almost identical environments. What you can't do with emulators is develop systems and procedures that can then be transposed into the real vehicle with almost total reliability in terms of expected outcome. In terms of sensations though, they can be just as good.
Arguing pointlessly over a definition that isn't particularly relevant for most users I would suggest..

 

To Simulate you have to calculate,
To Emulate you have to imitate.
Big difference there is,

 

Agreed. But if the user doesn't notice the difference and its not being used to develop something for real world application, why spend millions when hundreds will do.. It's like computing the movement of leaves as the result of airflow in a tree. Possible but why bother when a simple emulator will do in creating the same effect.. in this case simulating reality is just not necessary when emulating it creates the same effect. The additional processing isn't altering the experience in any way.. It's not as though we use computational flow dynamics on our desktops..

 

When saying simulator are we talking about software or hardware? Because the simulation software that is available to professional teams is not all that different from what we have at home, and there have been several real drivers that have said so. Even rFactor Pro was pretty much just the launching pad for ISI to develop rFactor 2.

My guess is that the professional sims might be running at a higher physics rate (thanks to using more powerfull computers) than the commercial versions, and they're probably running more accurate numbers that otherwise manufacturers might not want to disclose to devs for use on commercial sims, but the basic mathematical formulas used to calculate each element of the physics of a car are formulas that are known and docummented in plenty of books about the subject, so they would not be all that different between professional and commercial sims.

For a long time now racing sims have done a remarkable job of accurately simulating the behaviour of a car in relation to it's inputs, even rFactor 1 could produce telemetry results that could be incredibly similar to real world telemetry of the same car/track combo, and ultimately that's all that is necessary, because the inputs (via pedals, wheel and shifters) are all that we have available in a sim as well.

All this to say that this conversation between emmulation and simulation is just a very weird one to me... but I guess it's to be expected now that discussions on this forum will always get weirder and weirder the closer we are to the end of the month

 

It does help to pass the time

 

Ref OP
Qs
Have you set and/or experience of changing the track states.? in this build.
green, rubbered, heavy rubber, progressive, etc
Only interested in wet with wet tyres pls.
I would like to correlate some results , In particular Porsche GTE grip levels from green track to heavy rubbered in wet conditions. Full rain and thunderstorm only.
(more specifcally heavy rubbered thunderstorm)
Feedback from other cars/classes welcome also.