An interesting discussion going on at the vpforums:
And what I want to do at some point is to dig deep into the physics engine and see what I find. I have some dream projects here, like simulating the elasticity of the flipper rubber. I have no idea whether any pinball sim does that and whether it even adds any realism, but it might be a fun thing to try.
------
If you can do something about the physics I would recommend to make the ball more controllable by the flippers. At the moment it's hard and almost every time a lucky shot to precisely shoot in the desired direction. This can not be compared to a real pinball machine in any way.
-------
And the ball should have more mass. As the moment it behaves almost like a rubber ball than a steel one...
-------
The tricky part is figuring out why the ball doesn't behave as it should. All the basic parts are in place, you have the ball, the flipper, the rigid body collision, this can all be simulated accurately, but maybe something is missing. So my idea was that perhaps simulating the actual way the rubber is compressed, grips the ball and then expands again to return extra momentum to the ball can add some extra realism. But it's just a crazy idea of mine. I wish I could get high-speed camera footage of a flipper hitting a ball...
-------
I believe what's missing fundamentally, besides the elements for a full 3D model concept / ball allowing for more realistic collisions resulting in ball spin, is that the properties for the components of the virtual pinball table (wood, plastic, rubber, and especially steel) are too similar / limited and a major thing such as density / mass for an object's area are not considered. I've theorized and posted about this before and wouldn't be surprised if the benefits that my BMPR routine yields could be largely accomplished by just having the ball be recognized in that it is made of steel (i.e. heavier in it's space that it occupies than a wall of the same area). Having the ability for adjusting friction and elasticity on objects isn't enough for better modelling and more natural interactions. The momentum of the ball I feel would greatly improve if we could tell VP it's made of a heavier substance. It would take more to get going but also more to stop (just like a real table) and that aspect would yield some improvement in two problems with VP - the ball typically moves to fast / gets turned around too quick, yet doesn't carry it's lateral momentum well or transfer naturally from an up-table direction to a downward (and accurately accelerating) one.
I do agree with you in the aspect of some objects and how they "give", especially the rubber objects, as it is an important element and the cushioning factor of various materials has a reasonable bearing too on ball trajectory (this would likely be needed for some spin as a ball contacts a rubber and pushes it not only in but possibly down for a bounce back with upward direction - possibly for some air ball too).