Implicit conversion from float -22.551159618008185 to int loses precision