Implicit conversion from float -75.4796511627907 to int loses precision