Implicit conversion from float -57.846889952153106 to int loses precision