Implicit conversion from float -6.0328317373461005 to int loses precision