Implicit conversion from float -25.605095541401283 to int loses precision