Implicit conversion from float -56.441947565543074 to int loses precision