Implicit conversion from float -57.257769652650836 to int loses precision