Implicit conversion from float -25.11142061281336 to int loses precision