Currently when a positional image is found we just add it at the line just after paragraph and with no positional info. Good thing: at least get the image exported. Bad thing: the position looks either bad or real bad.