Can computers generate art and architecture? Recent technologies like Google's DeepDream increasingly say so. But can we extend this trend into architecture, a discipline at once artistic and practical, abstract and lived-in?

We pursue various methods and workflows to allow computers to generate spaces with little to no human intervention. Along the way, we evaluate each approach's successes and downfalls, using these intermediate results to fuel an informal investigation into heuristics for good practice in computer-generated architecture, as well as to refine existing algorithms to produce pleasing results.

In the end, we produce a bird's-eye view of a fully computer-generated architectural landscape using no human involvement whatsoever.