Hi!
I'm 14, from Germany and you will probably call me insane:
I'm planning and programming a 2d game engine mainly based on SDL.
I wanted to implement my own pixel-perfect physics engine.
Pixel Perfect means in this case, that the physical object's shape is defined by how their sprite looks.
The advatage of this is, that you don't have to think about a shape of something, you just design it
in Photoshop or Gimp, you have an image, and the non-transparent pixels also are the physical shape of the object.
Also it is more accurate than any shape approximation like boxes or circles.
I started to program it in Ruby with a self-made Ruby-sdl extension.
I got something running what behaved physically not too bad.
The main problem was performance.
If I had more than 5 little objects at once,
The fps got below 10 ^^
Certainly Ruby as an interpreted language is not the right thing.
So I started all over in C++.
I'm very impressed by Chipmunk Physics, because of its high performance and stability.
Now I had the Idea of just adding a new type of collision shape to the engine,
a pixel-based one.
Then you could use either polygon shapes, or pixel-based shapes together,
what would, in my eyes, be a perfect balance between performance, accuracy and simplicity.
What do you think of this Idea? Would it be hard to implement?
Any suggestions?
If you don't say its a completely insane and dumb idea I will try it and see.
Thanks for any tips and sorry for my English.
Anselm