The Schufftan Process

The Schufftan mirror trick process is the second major special effects technique used in the film. It was invented and implemented by Eugen Schufftan in collaboration with Ernst Kunstmann, and was a form of compositing miniatures into the full-scale shot using mirrors.

Schufftan was not originally involved in filmmaking, and had a background as a painter and architect, but he found himself fascinated by the medium of moving pictures. One of things he found disappointing was the lack of a sense of depth in most of the films of his time, and so he devised the mirror trick to achieve this sense of deep perspective by compositing models and real scenery.

The Schufftan process enabled a range of complex shots never before possible in film. In the scene where the mob is chasing the evil Maria, we see a shot of the crowd running on a roadway between two seemingly enormous buildings. In another shot, we see a group of workers crossing the roadway while overhead the stop-motion animated model cars appear to drive overhead. To achieve these effect, a mirror was mounted at 45 degrees in front of the camera lens. This mirror reflected the image of the miniature model, positioned directly behind the camera. Parts of the mirrored surface were scratched away to correspond to areas where the real-life footage would be composited to give the camera an unrestricted view of the real scenery.


This process was also used to create a sense of vast scales without having to create even larger sets than the production alrady demanded. The lower floors of the buildings in the so-called "worker’s city", for example, were constructed full-scale in one of the film’s massize sets. The upper floors were mirror image models.The same technique can be seen in the track scene in the "Stadium of the Sons", in which the track and lower portion of the wall were shot full-scale with people running in the foreground. The wall itself was over 10 metres tall (or nearly 33 feet). The upper portion of the wall and the dome in the background were a mirror image of the model, scaled to 1/20th of the simulated size.