Oh, sure you do. I probably should have used binary notation instead decimal for the output values, like this:

    case (in)
      8'b00110001: out = 2'b01;
      8'b10101100: out = 2'b10;
      8'b11110101: out = 2'b11;
      default:     out = 2'dx;
      endcase

Now it's easy to see that for the three defined input vectors, out[1] is always the same as in[7], and out[0] is always the same as in[0]. Since the output is don't care for all other input, the simplest implementation is just two wires.

Am I being stupid here? For once, I don't think so.