Implicit conversion from float -49.507389162561566 to int loses precision