Implicit conversion from float -24.94845360824742 to int loses precision