Implicit conversion from float -6.051379638439585 to int loses precision