Implicit conversion from float -1.6056338028168966 to int loses precision