Implicit conversion from float -141.5068493150685 to int loses precision