imaZighen {l Wrote}:This is remarkable, If it was in c++ I wouldn't use it btw, this library gave me a lot of help, thanks mate!
Lyberta {l Wrote}:Float tends to have a very small precision when you start working with larger numbers.
For the reason @imaZighen mentioned, double isn't that suitable for games (the only thing I can think where it's really useful is for time precision). As for the imprecision of large numbers, it's actually a good practice to divide the world in chunks and translate the objects back to the chunk [0.0f, 0.0f, 0.0f]. For example, if the nearest object to the origin is [600.0f, 500.0f, 300.0f], translate everything with [-600.0f, -500.0f, -300.0f]. And use scale whenever you can, instead of large numbers.
If you were using double, your physics would work correctly with lager numbers, but the rendering still would shake, as the double values would be converted to float.