Implicit conversion from float -62.18023255813952 to int loses precision