Implicit conversion from float -26.058394160583944 to int loses precision