Implicit conversion from float -6.03715170278636 to int loses precision