A discussion about simulation accuracy

I recently drove the Huracan GT3 back to back in AC and ACC. In the two sims, the same car feels very different, primarily through differences in the tyre model, I expect. What I found interesting was that for years we felt that AC was the pinnacle of sim physics accuracy. The way the cars handled in AC was very lifelike. And yet now, ACC feels quite different. With all of Aris's explanations of the additional nuances in the ACC physics model, you would assume that ACC is more accurate. Because the cars feel so different, does that mean that AC was way off the mark? Or was it simply the closest simulation that we could get 5 years ago, with technology moving on in the interim?

In any case, it highlighted to me that we need to take any current feelings of simulation accuracy with a big pinch of salt. Even the real racers can't agree on how accurate some aspects of the sim are. The goal continues to be simply to make the physics feel "believable", rather than "accurate". So long as we can believe that yes, this what the real car probably feels like, without too many "WTF?" moments, then we sim racers should be happy.

 

The difference comes probably from the fact that ACC does not yet have a Simulation Value adjustment

Jokes aside, because fun and simracing don't go together:
The differences between AC and ACC in my opinion come on one hand from all the new features in the tyre model and physics refinements Aris talked about, and I guess combined with more capable hardware for the users this is the main thing. On the other hand, AC simulated a lot of different cars and car classes/types, many different tires and suspension models, so maybe the physics model had to be more generealized in AC than it is now in ACC, where it can be specially tuned for GT3 cars with only two tires and can therfore communicate the nuances better.

Generally speaking, I think what anybody feels in the wheel is completely subjective. That's why we not only have people who say ACCs FFB is very dampened and uncommunicative, and others say it's the best anything ever felt, me amongst them. That's also why different people prefer different sims, and there are myriads of people who'll prefer rF2, R3E, AMS, AC or pCars2 and in their eyes their respective sim of choice has the best FFB ever while using vastly different FFB settings for their wheels. Maybe because they simply got used to it over time, or because their sim's FFB offers something others don't offer, like some detail in the wheel, or better weight transfer feeling, etc.

 

IMO, minor differences in FFB, the physics, or tyre model can make a big difference.

That said, i don't feel like ACC is massively different to AC though. I won't lie, i also had some difficulties with correcting drifts, but, it seems a whole lot better since i turned down the graphics options, and got more frames, and less input lag in the process. I think especially the post-processing option has a big influence on the severity of input lag.

I think we should not discuss this based on a single car. There's so many more cars to come, and, i remember that, in AC, the McLaren, and the Ferrari GT3 could also be quite some beasts to handle...

 

It will always be just "believable" till you won't start breaking things into little details, examining them closely and then going even deeper till you start actually be precise at such degree that you will be confident about something being accurate, and yet consider a possibility of some error. The combination of simracing and believing is so medieval lol But it is totally normal to feel if something is just enough. After ~2k hours AC started to feel a bit limited, even though it is really epic, ACC is such a jump forward really happy about it, even with just GT3 cars. Most other cars probably will never be simulated so well like GT3 seems to be in ACC, but who knows... some principles could carry on to other vehicles, doors to some data might open up, new ways of simulation appear...

I also think that if a person is a real racer or experienced driver, it doesn't mean that he is able to judge reality in such way that he could properly compare it with simulation.. I think they usually just drive and that's it. However if there is a possibility to get some numbers, or get in touch with people who knows numbers, then it is a different story. Kunos certainly got some numbers now !

ACC is surprisingly a big leap in realism, I'm pretty sure about it. Who knows how many big leaps there are left, I think ACC could be one of the last big leaps when we talk about the vehicle simulation. Now add way way more powerful graphics and sound, and you get the whole picture. Later it will probably take more and more work, and progress will be less and less noticeable for an untrained feel, just like Nvidia with raytracing for example a lot of people doesn't care how light works, it just happens, what do they know about it, just the basics of basic stuff, same with physics, sound and so on, basically everything is physics.

 

That statement is a huge slap in the face to Kunos.

 

For me that's a given and clearly isn't but should be for anyone else also. The reality is that a racing simulation at the end of the day isn't a perfect replica of real life and probably never will be.

 

A few things to consider:
There will always be things, like tyres and aerodynamics, for example, that require some simplifications and are impossible to recreate like the real thing. Every developer will do these differently, and they can make the same car feel different in two simulators, even if they behave very similarly to each other and to the real car.
You said the car feels different. Well, feeling different doesn't mean it behaves all that differently. ACC has a completely different FFB to AC, and even if you take the same car with the same physics and change the Force Feedback, the car will feel a lot different, that to me says nothing.
I believe the stock setup for the Lambo on ACC is completely different from the one on AC, that alone can also change how the car feels and handles.

If you drive the same car and same track on five different simulators, they will all feel different, doesn't mean one is more or less accurate than the other.

 

It means that they are all inaccurate. Mother nature doesn't have five different physics presets

 

Yeah his sentence was poorly thought out. It is so ironical, basically it is: "hey I thought, I believed that AC was the peak, but now you make ACC and it is even better - that means that AC was not the peak, I don't trust you anymore and I have no idea whats better or worse, I am placebo guy"

It is true, when they all be totally accurate they will become totally identical, or they would if they would. It depends on how high your standards are. But not only that. Also depends on what sort of perception of realism you have, what you know, and what you just believe to.

However, nature has more than one kind of different tires. All sorts of mechanical stuff going on.. But seriously physics are so deep it is not unusual that same thing is very different in different times.

 

What a ridiculous misinterpretation.

Not at all. My point is a philosophical one. When you enjoy a sim, it's because the car behaviour feels believable. You think "Yes, this is pretty much what driving a real car feels like". On the other hand, you try some sims with poorly modelled physics, and you think "WTF, whay did the car do that?". It's just not believable, and you don't play that sim any more.

In our AC careers, I would say the vast majority of us felt that the behaviour of the cars was believable. It lived up to our expectations of how a car should behave. Now lets say ACC has stepped up the accuracy of the simulation, or in other words, AC is not as accurate as ACC in terms of the physics simulation. We still enjoy AC, right? That's because it's believable. Being believable is more important than being accurate.

I just think it's interesting, from a human nature point of view, how we still enjoy AC even though in all likelihood its physics simulation is less accurate than ACC's. That's why I said being believable is more important than being accurate, if we're going to enjoy a sim. If you know AC's physics is less accurate than ACC, do you enjoy it less?

 

The goal of a simulation, any simulation, is to replicate the real world. Accuracy is the the priority, you want your simulation to give you results that would match the real world or it's not a simulation. The problem is the real world is highly complex and has millions of variables, and all simulations take massive short cuts like not simulating the electrons in an atom, or the effect neutrinos have on the fly that just hit your windscreen.

Belief isn't a word I like to use, I don't think it's a particularly useful word, if a sim is accurate you can use it to become better at driving, it doesn't matter if it's believable or not as long as the results are positive.

 

It would be interesting to know what the approach was when they knew they would only model GT3s vs all the manufacturers and models they had to deal with in AC. Is the ACC physics engine tailor made or would it work okay with a KTM?

 

I don't agree with the op.
The cars are not proven to be different.
In AC we had a default generic setup.
In ACC we have 3 setups, which are not proven to be identical with AC's setup.
Forget about FFB.

Someone needs to take both cars with exactly the same setup (I don't know how we can do this) and compare their telemetry on the same conditions.

I bet that it will be really close.

Sorry but this thread comes with a not scientifically explained assumption.

Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk

 

Hmm..

[more believable>more realistic];
[enjoying less advanced simulation=enjoying more advanced simulation];
[0=1].

much sense. very logic

I think you can still enjoy AC, or any other simulation, there are many reasons for that. But if people will use AC for GT3 cars, while owning ACC I will say that they are stupid. Stupid. But that is not going to happen. The question is how about iRacing. I think iRacing is already behind by now. As well as rF2.

 

This thread is not about opinions or thoughts.
Is about scientifically proved fidelity of a simulation.
Please stop sharing your thoughts.
If you have anything to prove your 100% meaningless assumptions, please do share them.
This is a common mistake that people do when comparing simulations : they use assumptions, feelings, not same conditions, compare between a simulation and another and not with real life, without mathematical or scientific tools.

I would expect someone to appear and say : look I have real data and I compared the X sim and the Y sim in various aspects , at the same conditions as real car and environmental and I discovered that the X sim is Z% more accurate as a result ( added or means values ) compared to real car than the Y sim.

Wake me up when this happens.


Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk

 

Considering that nothing has been scientifically proven in this thread, please take your own advice.
------------
You guys remind me of mumble rappers.
------------
GTR2, GT Legends, and Netkar Pro are all 'obsolete' compared to ACC yet they are still fun to drive. Or maybe they aren't, if the OP is to be believed.

 

Exactly . That's what I meant. You are indeed sometimes useful...

Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk

 

I rather like to think that if devs strive for as high accuracy as possible, then the "believeability" will come as a result of that.
In any case, "believeable" is a word that comes pre-loaded with a buckload of preconceived notions.

 

I and others can scientifically prove, ACC feels like ****, sorry but it does, it just does, it does, it feels like ****!

 

Every simulator is inaccurate, but if even professionals use them to practice and help set the cars up, I think they're accurate enough for me.