On the official site, images are parsed when the body contains them. It seems like the open source version doesn't. Or am I misconfiguring something? Thanks, Sam