Implicit conversion from float -24.43883984867591 to int loses precision