Implicit conversion from float -52.672998643147906 to int loses precision