Implicit conversion from float -0.7177033492822886 to int loses precision