Implicit conversion from float -46.10738255033556 to int loses precision