Automobilista 2 V1.4.3.3 & Racin´ USA Pt3 Officially Released!

Some thoughts on the topic of performance discrepancies of the sim cars vs real counterparts: Generally speaking, most cars in the sim can achieve faster times by 1-2s in an average size track. We consider that a reasonable threshold given variables that aren´t thoroughly simulated and inherent advantages of sim racing vs real life driving - naturally when we learn of any measurable innacuracy that may be contributing to a performance discrepancy, we will address it.

There is a different approach in classes with mixed cars, where we may slightly over or undermodel a given car´s strengths and weaknesses in order to retain some balance of performance against the other cars in the class - this way models that were faster than others in real life are still faster in the sim and their relative strengths and weaknesses represented, but the gap usually reduced to tenths rather than seconds. Real modern GT series do that via balance of performance rules, and it´s just a logical gameplay measure that does not fundamentally detract from its realism - you can give a car slightly better engine, tires, aero or take a little weight off, but the car is still fundamentally the car.

Historical cars are trickier because data is both more scarse and less reliable, including performance data - motorsports as a whole just weren´t as optimized back then as they are today, and the chances teams and drivers would extract peak performance from a car over a couple of qualifying runs in any given year were substantially reduced relative to what is seen today.

Take for example this result from the 1000km of Spa in 1988 - the Sauber C9 set the pole with 2:02.2; our Sauber C9 in AMS2 in turn is capable of running in the 1:59s in the similar (if not identical) Spa 93 layout. Can we say for sure that means the sim performance is innacurate though, or even by how much? Do we believe a Jochen Mass, JL Schlesser or Johansson were really extracting the maximum lap time from that car, Ayrton Senna style? If we can´t, we don´t know for sure what peak performance of the car was and thus have no reliable frame of reference. Schumacher and Frentzen as rookies were almost immediately faster than any of the old timers once they arrived at Sauber, hinting that would be far from a safe assumption.

Then there is the element of team preparation - in that same qualifying session, the faster Porsche 962 C was a whopping 8.6s than the slowest one - naturally you have to take the faster car as the more appropriate reference of its performance, but what if there were no other Porsches to gauge from? Some cars like the Corvette GTP had so many variations over the years and so little sampling of results to pick from that any random lap time is hardly conclusive evidence of its ultimate capabilities.

Than there are the tracks - even tracks that have for the most part retained their old layouts have become faster courtesy of flattened curbs and better tarmac quality since the 00s relative to the 70s and 80s. If a modern track is all you have to gauge lap times relative to real life results from the same car and track in the 80s, it´s already far from a 1:1 comparison.

In light of these variables, the best we can do as sim physics developers is to reach a very educated guestimate that tries to account for all these unknowns vs what can be learned from the car itself, in order to deliver a reasonable representation of the car that works in the context of the game.

There is no question the Corvette GTP performance is overmodelled, as is in fact the Porsche 962C - the extent of which however when you consider the above is extremely debatable, and whatever 1-2% discrepancy there is in each component do not detract from the fact that the car as a combination of chassis, engine, suspension and aerodynamics modelling remains essentially the car as it should be.

 

The next update will add support for cars to be set with front / rear roll bar adjustments independently.