Implicit conversion from float -25.088282504012838 to int loses precision