Implicit conversion from float -7.7763659466327795 to int loses precision