Implicit conversion from float -10.24509803921569 to int loses precision