Implicit conversion from float -35.778210116731515 to int loses precision