Implicit conversion from float -44.975767366720504 to int loses precision