Implicit conversion from float -25.72085889570552 to int loses precision