Implicit conversion from float -12.6813880126183 to int loses precision