In his Paralipomena (1604) Johannes Kepler reported an experimentum that he had seen in the Dresden Kunstkammer. In one of the rooms there, which had been turned in its entirety into a camera obscura, he had witnessed the images formed by a lens. I discuss the role of this experiment in the development and foundation of his new theory of optical imagery, which made a distinction between two concepts of image, pictura and imago. My focus is on how Kepler used his report of the experiment inside the camera obscura to criticize the account of image formation given in Giovanbattista Della Porta's Magia naturalis (1589). I argue that this experiment allowed Kepler to sort out the confusion between images 'in the air'—referring to the geometrical locus of images in the perspectivist tradition of optics—and the experimentally produced 'projected images', which were empirically familiar but conceptually alien to perspectivist optics.