Implicit conversion from float -57.16216216216216 to int loses precision