Implicit conversion from float -94.39490445859872 to int loses precision