If Normalized Predictive Distance is increased, energy transfer/restitution via bouncing essentially stops working. For example, if you run the Restitution test in examples2D, you'll see all the balls bouncing as expected. But if you then increase Normalized Predictive Distance (for example, to 0.1), you'll see they stop bouncing almost immediately. I believe that when impacts are detected via the Normalized Predictive Distance, the steps to account for bounciness are being skipped.
To reproduce:
In the restitution2.rs test (rapier\examples2d\restitution2.rs), add the following line to the end of the init_world method:
testbed.integration_parameters_mut().normalized_prediction_distance = 0.1;
Then observe the example.
If Normalized Predictive Distance is increased, energy transfer/restitution via bouncing essentially stops working. For example, if you run the Restitution test in examples2D, you'll see all the balls bouncing as expected. But if you then increase Normalized Predictive Distance (for example, to 0.1), you'll see they stop bouncing almost immediately. I believe that when impacts are detected via the Normalized Predictive Distance, the steps to account for bounciness are being skipped.
To reproduce:
In the restitution2.rs test (rapier\examples2d\restitution2.rs), add the following line to the end of the init_world method:
testbed.integration_parameters_mut().normalized_prediction_distance = 0.1;Then observe the example.