Implicit conversion from float -26.003062787136287 to int loses precision