Implicit conversion from float -38.244274809160316 to int loses precision