Implicit conversion from float -5.511482254697285 to int loses precision