Implicit conversion from float -28.277511961722468 to int loses precision