Implicit conversion from float -0.42402826855123976 to int loses precision