AMS2 vs AMS1 Physics

I did testing with Lotus 49c. It has enough power and flimsy enough tyres, and I did set wings at 0 degree, shortest gear ratio and preload to 0, so that nothing could mask the effects of Clutch LSD settings.

And you are completely right.

At least ramp angle settings in Clutch LSD for both power and coast side settings seem to be completely dead, at least for Lotus 49c. I even forgot many a time during the testing which way around they were because I could not make any difference in handling.

Edit: I did also test again Mustang'66 in PCARS2, and the changes to handling there with adjusting clutch diff ramp angles were equally unnoticeable. No idea since when it had started acting like this, because I remember testing the clutch diff with burnouts during the development at WMD and did not notice issue then.

Edit2: Testing with Ginetta G55 on half grass, half track, shows that Clutch LSD works at least in this car very visibly.

You were right about ramp angles too. Clutch dump test with full revs, one side of the car on the grass:

Power ramp angle at 25 degrees locks more and makes the car spin to the grass side, as the road side wheel has more bite while the wheel on the grass keeps spinning on spot.

Power ramp angle at 89 makes G55 just push onward, as both wheels rotate on their own speeds producing same amount of traction. Sometimes very small twitch towards grass happens when dumping the clutch, but mostly it just goes onward.

So, overall it does not seem to be a bug in the game engine. Possibly it is a bug in the car, saved setup or setup screen handles.

 

This seems to be such a common misconception for people who have obviously never driven a real car on a track

Please for the love of [insert your preferred deity here], go and get a track day.. with any car. Heck, if you have a large empty parking lot somewhere and can afford to waste your tires on your daily driver, go have some fun (safety first!) and see how a car handles in real life while on and around the limit of tracktion. If you can find a dirt road that is closed for traffic, see if you can drive a few miles on that stretch and really push the car. It's fun and teaches you a ton about how forgiving it really is.

In my limited experience I've found that most ordinary cars are not rocket science nor are they super difficult to handle. Yes, you can do stupid things and end up spinning around but it's not as difficult as simracers keep insisting it being. For instance, I have no idea why people think that locking up your wheels completely while braking in a straight line should send you into a spin. This simply doesn't happen unless you input some silly amounts of steering lock or have a broken car.

About 8 years ago I drove my very first ice rally in a local track day setting. I had never driven on a frozen lake before.. and I was fairly competitive and never spun the car even once. I did shunt a few people on the track but we were racing cars that were going to the scrap heap anyhow after the race (ours was a banged up, almost dead, completely stripped out Peugeot 206 with some old semi-working custom mods to the engine). Speeds on the main straight went up to about 100km/h and the first corner was a shallow left turn. A few cars did go out in that corner over the course of the 4 hour race (we had driver swaps) but not many.. and there was like 40 cars on track at any given time, all different makes and brands. Just like me, about 50% of the people driving there had never driven on a frozen lake before.

My point is: Driving a car is not that difficult. How many times have you actually spun around and completely screwed up, simply by driving errors while carting? For regular cars/tourist track day boxes, the somewhat whereabouts of "the limit" is not at all walking on eggshells, like so many past sims would have you believe.

.. now when it comes to truly powerful cars like F1, LMP etc.. I have no idea.. but logic would dictate that it would be silly to make these cars death traps that insta spin from regular steering inputs. If these types of cars behaved like iRacing portrays them.. we'd see a lot more extremely weird crashes and have a lot more dead drivers.

 

You got me wrong. I'm not criticizing your skills. I think you can be more skilled than myself. The point is, it's not a good rule to measure how the game is realistic, but you are using it in complete subjectivity.

My point is: Whatever real theory or technique I use, the game responds correctly to it. I even used, for example, the "Alonso steering" theory and it worked as it should. Well... even my setup, that you used to get a P2 on LB, was totally build over real theory and it works exactly as it should. In my standards, when a simulator responds to real knowledge means that it's a good simulator. Maybe we can question how precise it is, but I really believe that there is not a single simulator on the market that get the 1:1 perfection, or even close to that.

 

This is another common fallacy among simracers. Just because all the other sims did this or did that doesn't mean they do it correctly.

iRacing is yet again an example of just how WRONG the physics have been for the Dave Kaemmer titles. Heck, real racing drivers have literally been silenced on the iRacing forums back in the day.. and now the "noise" from real drivers, criticizing the physics is at an all time high.

Yet you have the vast majority of simracers, who have zero experience with real cars, almost religiously spewing the nonsense that iRacing is the most realistic simulator ever. Yet all the real evidence points to the opposite.

Just some examples from a dude with various driving experience.

Having said that, I don't know at all how 80s or 90s F1 cars should feel.. none of us do. You can only compare with more ordinary cars and extrapolate the accuracy from there. iRacing gets even a basic Mazda MX5 physics completely wrong so yeah.. chances are that all the other stuff is also completely broken.

My point still stands: Just because other simulators have done something a certain way doesn't mean the current new games get it wrong. It might just mean that we are finally arriving at physics simulations that actually get things right.. or not. Who knows. Maybe in 10 years we look back at AMS 2 and go "yeah, those were strange days!".

 

Absolutely. AMS1 is actually very nice and easy to drive as well, so at least it can't be "that" broken.

I'm absolutely positive there are some strange oddities in the Madness physics engine but it's hard to put a finger on it.. but the overall general driving feel is in the realm of believable, and in my opinion even more so than AMS1 which feels a bit dated somehow. The older sims all have a sort of "simulator on computer" feel to them.. whereas the newest breed of tire models have some sort of interesting complexity to them.

In the realm of music/audio production, we are slowly moving towards physically modeled instruments instead of using sampled instruments (basically just multiple audio recordings of all the possible "states" of an instrument). These physically modeled instruments are much more capable than their 'static snapshot' brethren but yet in some cases fail to sound as realistic. This is however slowly but surely changing with advancements in modeling. I suspect this is the very same case in simulators. Having anything based on look-up-tables (aka sampling) can be thoroughly convincing at a large.. but it can't account for all variables in all situations in the way physically modeled physics/elements can.

For instance weather.. you can build various snapshots of weather into a weather simulator that are based on other snapshots of elements that affect that weather (humidity, perspiration, temperature, wind etc) and you got a sort of semi-predictable system that can be somewhat convincing. But if you actually model the whole system, each and every part that goes into it, you'll have a much more convincing overall experience and more fluid transitions between states.. you'll also end up with occasional hilarious bugs where the system "explodes" and goes nuts.

I'm definitely rooting for all simulators going the physical model route as it will ultimately be the correct way of doing things. Every other sector in the world is going or has gone this route (science, economics, medicine etc).

 

This is very possible. The pCars franchise was notorious for it's lack of reliability when it came to saving and actually recalling setup changes! I still to this day never overwrite a previous car setup.. as it was buggy almost all the time. The only way to somewhat guarantee that the setup is actually saved AND loaded on to track is to save a completely new file, then exit the session and load that newly saved setup! Aggravating to say the least.

 

For me saving and loading setups in PCARS2 hasn't caused me issues (that I'd have noticed) after the final two patches anymore. So it might work ok these days.

I was thinking saved and loaded setups would have been cause here too, especially because AMS2 saves the existing setup always after any changes into the second setup that has *-character applied to its name. This is very convenient and nice from setup tuning point of view, but it irks me when I remember all the issues we had with custom setups in PCARS.

So I tested resetting setups to default and even deleting my profile (sigh again), but that didn't help. So maybe the LSD setup screen of Lotus 49C is just pointing at thin air there or something. The car definitely drives very well with its default settings, but those F1 Retros with their big fat slicks and tiny fronts may be different story.

 

Haha, true that. Any physical model can be severely botched by just a few things being off. That's indeed the difference to a typical look-up-table "simpler" solution. Then again, I've heard making the tires for rFactor 2 is notoriously difficult simply because of the sheer number of parameters. I mean, rF2 is a sort of pseudo physical model as far as I can understand it.. but instead of running real time, it lets you "print out" the results of the simulation into a list.

 

Hi. It is 2024 already. I bought the game in the last Steam, apparently there was a physics update last year.
I have to agree with the OP. All the cars feels like in an arcade game. When I brake I dont feel like I am grounding the front of the car but I just make the rear tyres slippy, the car rear stars to slide and it is a very easy and controllable slide. I find myself counter steering like I am playing a rally game which I doubt is very realistic. It is fun but does not feel very realistic

Then I found some videos that seem to agree with this view of the game. For example this one, it explains the problem very clearly.
More relevant is the Nils Naujoks review of the current physics of the car. You can tell that he is deeply disapointed with how the cars handle. again with this fake slide that is easy to correct happening all the time.