Implicit conversion from float -94.37288135593221 to int loses precision