I think this will require some moderately intricate JavaScript to detect where a respondent clicks in a <div> element and load (hidden?) text inputs with the coordinates.
↧