Implicit conversion from float -7.6890756302521055 to int loses precision