Implicit conversion from float -22.72545090180361 to int loses precision