In this paper we present experimental results rating the experience of users conversing in a casual video/audio dialogue, in a simulated home environment. Here video-realistic images are problematic and prone to numerous "medium effects", such as unaligned eye-gaze, which can be misattributed as personal flaws. We tested three levels of manipulated video to see if they improved user's sense of: (a) presence, (b) portrayal and (c) preference. By blurring the background we found a manipulation that is both preferred and more efficiently coded than the original.