Implicit conversion from float -26.2809917355372 to int loses precision