I don't disagree at all, for ex, Netkar feels very different to RACE07, but in both instances, I have excellent steering precision and FFB that I can use to help me judge the limit{my limit}, however, this isn't the case with pcars, as the FFB seems to drop in and out down the mountain at Bathurst, yet the best sims will feel fully connected thru there and any tricky corner section/s.
That fits what members say
about FFB in pCARS. The one thing that bothers me is why we're still having FFB issues, and basically of the same nature. Hmm...
As for physics/realism deficiencies, I'm sure we can find area's where they all suffer, but until the basics of the driving are fixed{steering+FFB}, no game or car can ever be judged as sim worthy as the basic underpinning of the game isn't up to standard.
Firstly, you misunderstood, but that's largely my fault.
I meant to say,
There are many indicators of wrong/under-developed physics for any sim, meaning if one looks beyond the feel and handling aspects there are many things we can focus on (sort of
"like" performance indicators in companies or employees), things which are quite telling and no developer can avoid them.
If a racing sim shows no obvious deficiencies in these aspects and telemetry confirms our understanding, then the obvious conclusion is the physics engine is doing the job properly.
You can spot them in videos, you can spot them when driving and the reaction of the car is wrong (ex.: linear motion of the CoM when loss of control occurs and you go sliding across the tarmac and grass, what you call center-pivot physics), and you can detect them through telemetry.
Again, the basics of driving for you are not quite the same as they are for others, or me. I look at many aspects, FFB being the least important (given its nature and the problems from wheel to wheel).
When it comes to weight transfer, and this is one area where genuine sims excel, one of two things seem likely, ie.....
1....weight transfer is reduced to enable easy driving in arcade/simcade games
2....weight transfer and the level of FFB detail requires enormous programming time, therefore must be reduced to allow for large carlists.
It could also be a combo of the 2, but either way, Netkar loads beautifully, and you can feel the cars weight under braking and compensate either immediately, or on the next lap if you fully overshot.
Point 1 you made: a known developer (its titles are sold on a massive scale), when asked about this, told me no compromises were made. The titles are sold for the PS3 and XBOX (and PC), and given the fact that statistically, over 90% (I was not aware the number was this high, but he assured me it is) of gamers use gamepads to "play" driving/racing games, then a layer of control is put in place to properly translate motion and feedback - this layer of control also implies a series of driver aids may be permanently on for some types of controllers.
Which is fine. Build the physics engine from the bottom up, and design code to help gamepad/console users control the cars better. This does not imply that the physics engine is botched or arcadey, though.
Point 2: Stefano Casillo told Autosimsport that the FFB for NetKar Pro was extremely simple...and short (a few lines only). So, no, you don't need hyper-complex FFB code for it to be good and have a fine level of detail. At all.
Regarding weight transfer: it is not entirely separate from the tire model code. When driving in RL, the car is either in static equilibrium or not - in which case we have a net torque and net force. Though roll centre locations are determined by suspension kinematics, for the roll axis we need to calculate and consider the car body motion in relation to the four contact patches and the existence of a force (or forces) that acts longitudinally or laterally on the CoG. Thus, in a sim, if something is not right with the forces calculated in the contact patches, weight transfer is also affected.
I honestly don't see any reason for a large car list imposing a simpler code - the physics engine, any physics engine, is fed a set of parameters and this sets the car model for any car. This is model calibration, and it is the same for cars (usually a large set of parameters to be configured) or tires, or even airfoils (wings). If you have 1 car, then all you need is a set of parameters for that car. If you have 1000 (GT5) then you feed all the numbers for all the cars and you get 1000 different cars.
The Problem is not there. The problem lies in the amount of detail you wish the physics engine to have. Some of the developers I know have made NO compromises (though they're working within some restrictions), others have. Some operate under very strict budgets and time constraints from publishers and distributors, others do not. Too often, decisions are made to simplify the simulation because these time constraints demand shorter development cycles. Sadly so.
