Most of the time, we like to have scaled-up values when working internally with the physics and scale them back down when displaying them onscreen. Typically we have 256 represent one pixel (though you can do it any way you want. 256 is probably the fastest number to divide by)