Implicit conversion from float -57.504159733777044 to int loses precision