TheoLogic wrote:The problem is when putting a big amount of torque on a wheel it spins out of control. So at this moment, I put a downwards force on the body, making it drivable. But at lower framerates, the vehicle starts buming around...
That's a problem you will always have when evaluating a differential equation at a fix sampling rate.
The error is a function of the rate and the non linearity of the function.
In the car when you apply a big torque you increase the amount of non linearity, so evaluating the function a low sampling rate (low frame rate) will make the error too large to reflect and realistic behavior.
using tricks like extra down forces, or higher friction will only increse the non linearity making the behavior worse.
The car joint in 2.0 is the same joint that was in 1.53 with the exception of the sampling rate.
In 1.53 I added a special sub sampling just for the car, but I took that out in 2.0 because I thought that engine was fast enough that people could do it themselves.
Maybe I need to put that functionality back.
Could you try this experiment and let me know if it makes be batter, then I will know what to do?
Make the physics time step say three time faster (180 or 200 fps)
You could try a simple level so that the experimental demo is playable; if it is better than I will add the option of sub sampling rate to joints.
That is if a car joint and all of it child joint get more a higher FPS, that way the physic tiem is oly increased when it count.
Please can you run that test for me?