Implicit conversion from float -6.176470588235304 to int loses precision