Implicit conversion from float -94.85237483953787 to int loses precision