Implicit conversion from float -24.777327935222672 to int loses precision