Implicit conversion from float -76.33064516129033 to int loses precision