Implicit conversion from float -8.344370860927157 to int loses precision