Implicit conversion from float -28.255250403877213 to int loses precision