Implicit conversion from float -0.06276150627616062 to int loses precision