Pictures that look like photographs are currently a standard means of presenting architectural projects. The photorealistic verisimilitude of these pictures appears to require no interpretive effort, yet it is also claimed to mislead architectural perceptions.

This thesis argues that to understand the architectural import of photorealism, it should not be approached as a stylistic property of finalised renderings but as a central mode of architects' digitalised work: for a user of architects' generic 3D software, operations largely take place as if seen through an imaginary camera.

What makes it difficult to recognise and question this photorealism of architectural tools is its fluency, that is, the ease with which it appears to make projects accessible. This fluency is shown to result from deeply entrenched conventions of pictorial realism and architectural representation that intersect in the virtual photography of computer-aided design.

Through thematic historical explorations, the thesis charts ways in which these conventions, under the cover of photorealistic fluency, both inform and confuse architectural perceptions and production.