Implicit conversion from float -25.690376569037653 to int loses precision