Hey thanks for trying out it!
Yes I definitely need a shadow for the marble, and a better marble texture is in order. I actually have a card that can handle shaders so I should look into it more.
The physics for speedy tiles is controlled (by Newton) using contact, so if you roll slowly over a speedy tile you will end up going faster than if you are rolling quickly, because there is less contact. I have too many speedy tiles there as a "trick" so the player needs to decide 1 1/2 tiles is enough to get to the next platform, or to limit speed of marble when rolling over them. If you can hop right to the end using too many, that's considered a risky bonus.
The catapult on level 2 thing is just bad design of the level, there are tricks to keep your marble from spinning off (spin backward, last minute, etc.) but it's not very reliable. And since I increased the timestep of the simulation I think it is even worse.
The physics are framerate independent via time-slicing. You're witnessing the inconsistencies of a complex physics solver (Newton) that I can't count on 100% to react the same every time due to the amount of contact and forces on the marble. I might make some things zone-based instead of contact based for this reason, but I haven't decided yet. [EDIT] Actually some of what you're seeing is the difference in physics behavior when I sped up the timestep of the simulation.
Next demo will have moveables, enemies, and some other widgets (black/white holes, new fans, etc.) so there will be more to play with. And I'll try to make some fun levels instead of the 3 basic "proof of concept" levels that are there. People seem to judge the game by the levels provided, where I was hoping the mere potential of it all would be enough. So I'll try to build up some more content.
[EDIT]@Link - that sounds cool.
A 3D marble platformer using Newton physics.
