We propose and validate a model of the perception area of content on public displays in order to predict from where users can read. From this model, we derive Screenfinity, a technique to rotate, translate, and zoom content in order to
enable reading while passing by very large displays. Screenfinity is comfortable to read when close, supports different
content for different users, does not waste screen real estate and allows expert passers-by to read content while walking. A laboratory study shows that expert users are able to perceive content when it moves. A field study evaluates the effect of Screenfinity on novice users in an ecologically valid setting. We find first time users can read content without slowing down or stopping; Passers-by stopping did so to explore the technology. Users explore the interaction, the limits of the system, manipulate the technology, and look behind the screen.