Implicit conversion from float -0.5960264900662224 to int loses precision