Implicit conversion from float -16.261682242990645 to int loses precision