Implicit conversion from float -25.027322404371574 to int loses precision