Implicit conversion from float -24.087947882736145 to int loses precision