Implicit conversion from float -109.74093264248705 to int loses precision