Implicit conversion from float -48.576779026217224 to int loses precision